Content uploaded by Mir Sayed Shah Danish
Author content
All content in this area was uploaded by Mir Sayed Shah Danish on Apr 07, 2023
Content may be subject to copyright.
Citation: Danish, M.S.S. AI and
Expert Insights for Sustainable
Energy Future. Energies 2023,16, 3309.
https://doi.org/10.3390/en16083309
Academic Editor: Alan Brent
Received: 21 March 2023
Revised: 4 April 2023
Accepted: 6 April 2023
Published: 7 April 2023
Copyright: © 2023 by the author.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
energies
Article
AI and Expert Insights for Sustainable Energy Future
Mir Sayed Shah Danish
Energy Systems (Chubu Electric Power) Funded Research Division, IMaSS (Institute of Materials and Systems for
Sustainability), Nagoya University, Furocho, Chikusa Ward, Nagoya 464-8601, Aichi, Japan; mdanish@nagoya-u.jp
Abstract:
This study presents an innovative framework for leveraging the potential of AI in energy
systems through a multidimensional approach. Despite the increasing importance of sustainable
energy systems in addressing global climate change, comprehensive frameworks for effectively
integrating artificial intelligence (AI) and machine learning (ML) techniques into these systems
are lacking. The challenge is to develop an innovative, multidimensional approach that evaluates
the feasibility of integrating AI and ML into the energy landscape, to identify the most promising
AI and ML techniques for energy systems, and to provide actionable insights for performance
enhancements while remaining accessible to a varied audience across disciplines. This study also
covers the domains where AI can augment contemporary and future energy systems. It also offers a
novel framework without echoing established literature by employing a flexible and multicriteria
methodology to rank energy systems based on their AI integration prospects. The research also
delineates AI integration processes and technique categorizations for energy systems. The findings
provide insight into attainable performance enhancements through AI integration and underscore
the most promising AI and ML techniques for energy systems via a pioneering framework. This
interdisciplinary research connects AI applications in energy and addresses a varied audience through
an accessible methodology.
Keywords:
AI-compatible energy models; transforming energy models; parameter-based models;
data-driven-based models; energy system modeling; modern energy policies; energy future landscape
1. Introduction
The application of artificial intelligence (AI) technologies is witnessing a significant
upward trend. Expected to reach a market value of 190.61 billion USD by 2050, AI involves
programming computers to make intelligent decisions by identifying patterns and enabling
them to learn from data [
1
]. This rapid development of AI is indelibly associated with
multidimensional constraints, including the lack of expertise and professionals to deploy
it as an emerging part of future technology, as well as institutional, behavioral, cultural,
psychological, ethical, and social concerns. Over time, from its early attempt to become
expert in given systems to today’s revolution, AI has been defined by a singular statement;
this statement is straightforward: “Artificial Intelligence is the study of how to make
computers do things at which, at the moment, people are better.”, if the term of “things”
is ignored [
2
]. Interdisciplinary studies demonstrated a broad knowledge of the topics
involved. Energy systems analysis in the computational intelligence and data science
domain using machine learning (ML) methods is a data-driven model susceptible to data
quality variation. Analysis of systems in the energy domain requires in-depth domain
knowledge. So, onboarding energy models, especially power systems, to a digestible
data set for ML requires interdisciplinary domain knowledge. This study uses a simplified
representation to deal with the most used ML methods in energy systems. Hence, extracting
high-quality datasets from the target model and incorporating domain knowledge, such as
physical and mathematical models, are crucial for accurate system modeling and analysis.
Standardizing energy systems for AI integration involves transitioning from a conven-
tional parameter-based model to a data-driven model. In a parameter-based model, the
Energies 2023,16, 3309. https://doi.org/10.3390/en16083309 https://www.mdpi.com/journal/energies
Energies 2023,16, 3309 2 of 27
system is modeled and controlled using a predefined set of parameters, such as equipment
rating, operating conditions, etc. While this approach can be practical for primary control
and monitoring, it can be constrained in its ability to adapt to changing conditions and
optimizing performance. Conversely, a data-driven model uses real-time data from sensors
and other sources to train and update the model, allowing for more precise monitoring,
control, predictions, optimization, and so on. A data-driven approach can be utilized to
enhance performance and augment the overall efficiency of the energy system. Machine
learning and deep learning (DL), especially neural networks, can be employed to analyze
and interpret the datasets and construct model architectures that can supervise current
operating conditions and predict future system behavior. Standardizing energy systems for
AI integration also requires designing a roadmap that includes initiation, planning, design,
implementation, optimization, etc.; this roadmap can provide real-time interactions and
implement communication protocols that facilitate data sharing and integration with other
systems through different protocols and standards. This approach aims to transform energy
systems models from parameter-based to data-driven, enabling the integration of different
techniques and solutions on a single platform. Ultimately, standardizing energy systems
for AI integration can boost the performance and efficiency of these systems, ensuring
highly reliable operations and control. This, in turn, promotes long-term sustainability
while adhering to techno-economic constraints.
Data-driven models have been extensively studied and posited as viable solutions
for several applications that span multiple interdisciplinary fields, particularly in smart
energy systems. The literature indicates that data-driven models possess the potential to
serve as a powerful tool for enabling parameter-based analysis of a system by leveraging
the rich information encoded in datasets to gain insight into the underlying dynamics
of a system and make predictions about its behavior. These models have demonstrated
their utility in a wide range of applications and are expected to continue playing a more
crucial role in advancing energy systems automation and control. However, it is essential
to note that, in order for these models to be widely adopted and utilized effectively, it is
necessary to demonstrate their scalability and real-world implementation capabilities [
3
].
Dobbe et al. [
4
] conducted case studies based on numerical formulation and data-driven
modeling regarding frequency regulation and distribution system control to provide general
guidelines on integrating learning capabilities for control purposes to make safety risks
a central design tenet. Zienkiewicz et al. [
5
] established a computationally simple, less
data-intensive, fast, and efficient data-driven model through a novel hybrid data-driven
model. Geneva et al. [
6
] propose a novel data-driven framework that improves model
predictions and provides probabilistic bounds for fluid quantities such as velocity and
pressure. Arridge et al. [
7
] aim to provide an account of some of the main contributions in
data-driven inverse problems. Singh et al. [8] introduced a novel ML-based fusion model,
known as PI-LSTM (Physics-Infused Long-Short-Term Memory Networks), that integrates
first-principles physics-based models and Long-Short-Term Memory (LSTM) networks.
Kollmann et al. [
9
] developed a DL model based on a convolutional neural network (CNN)
that predicts optimal metamaterial designs. Runge et al. [
10
] offer a review of studies
published since 2000 that have applied artificial neural networks to forecast building energy
use and demand, focusing on reviewing the applications, data, forecasting models, and
performance metrics used in model evaluations. Baptista et al. [
11
] studied the applicability
of the Kalman filter to filter the estimates of the remaining useful life. Terzi et al. [
12
]
present work that integrates 1314 vertical profiles of PAR acquired by 31 BGC-Argo floats
operating in the Mediterranean Sea between 2012 and 2016 into a one-dimensional model
to simulate the vertical and temporal variability of algal chlorophyll concentrations. A
systematic method to select the optimal network architecture is proposed and tested in [
13
].
Bättig, P., and Schiffmann, J. [
14
] demonstrated that data-driven models were employed
to predict the stiffness and damping coefficients of various nitrile butadiene rubber (NBR)
O-ring sizes. These predictions were based on O-ring geometry, Shore hardness, squeeze,
and excitation frequency. The authors also suggested that the curvature ratio (d/D) should
Energies 2023,16, 3309 3 of 27
be considered when developing reduced-order models. Data-driven approaches are widely
applied for controlling, monitoring, and optimizing energy systems. Among the many
studies on the topic, some are highlighted based on their specific applications, including
economic load dispatch [
15
], voltage stability and regulation [
5
,
16
], restoration of system
fault identification [
17
,
18
], planning and forecasting [
19
,
20
], network observability [
21
,
22
],
frequency regulation and stability [
23
–
25
], system storage and power management [
26
,
27
],
demand side management [
28
,
29
], electricity theft control [
30
,
31
], and unit commitment and
power management [
20
]. Feature engineering methods for data-driven model analysis are
extensively applied for various domains with original and modified applications, including
the analysis of crowd-sourcing materials science [32].
This study makes a tangible contribution to sustainable energy systems by presenting
a novel multidimensional framework for integrating AI and machine learning techniques,
thereby filling a critical gap in the existing literature. The study evaluates the feasibility of
AI and ML integration into the energy landscape and identifies and ranks energy systems
based on their AI integration prospects. By delineating AI integration processes and cate-
gorizing techniques for energy systems, the findings offer valuable insights into potential
performance enhancements and highlight the most promising AI and ML applications
for sustainable energy. Furthermore, the interdisciplinary nature of this research and its
accessible methodology ensure that it caters to a wide range of stakeholders, fostering
collaboration and knowledge exchange across different domains.
In addition, this study can bridge energy engineering, data science, and computer
science, presenting knowledge between domains from fundamental modeling concepts
to advance energy systems numerical modeling, system linearization, simulation, and
optimization in the broader picture. In addition, various ideas of the possibility of applying
ML in energy systems are highlighted; these may provide good hints for further research.
Although AI involves cross-disciplinary studies, encompassing coding, algorithms, statis-
tics, probability, algebra, big data, and critical thinking, this study also presents a roadmap
for acquiring essential knowledge on on-demand intelligence and its adoption. This un-
derstanding will empower readers to apply these tools and techniques to complex models
and case studies, utilizing platforms that support the application process [
33
]. This study
applies the methods and algorithms of the selected methods to create, expand, select, sort,
redact, linearize, validate, and ensure quality-optimized datasets using MATLAB
®
and
Python language as simulation tools. Python language is preferred for simulation due to its
open-source nature, extensive libraries, ease of learning, versatility, large community, and
seamless integration with other tools, unlike other proprietary specializations. Eventually,
this study’s primary objective and potential reside in devising a cutting-edge framework
that seamlessly incorporates artificial intelligence and machine learning methodologies
into sustainable energy systems, as emphasized in the Introduction.
The paper discusses the transformation of energy system modeling from parameter-
based to data-driven models in Section 2, while Section 3examines the competitive energy
policy landscape in the era of AI. Section 4delves into the role of AI and ML in the
transformation of the energy sector for a sustainable future, including process flow, mathe-
matical representation, and simulation results. The article concludes with a summary of
the findings in the Conclusion section.
2. Modeling: From Parameter to Data
AI applications for energy transition and climate change mitigation in the context of
promoting renewable energy technologies and power systems operation with maximum
efficiency and optimum results at large-scale performance are limited to research that will
shortly lead to extensive exploitation. Key areas that will shape the future of sustainable
energy and require the emergence of intelligent technologies include optimal planning,
automation, reliable operation, financial optimization, predictive maintenance, real-time
monitoring, demand forecasting, smart grid management, renewable energy integration,
energy storage management, efficiency improvement, and emission reduction. These
Energies 2023,16, 3309 4 of 27
interconnected domains will be critical in driving advancements in sustainable energy
systems. Automation of energy systems planning and operation, aided by AI technologies,
provides modern infrastructures primarily operated by robotic technologies, ensuring high
accuracy and intelligent control in terms of actuation, interaction, and responsiveness to
dynamic conditions. As contemporary energy landscapes transition towards Industry 5.0
standards, they integrate the concept of “man and machine”, known as collaborative robots
(cobots) [
34
]. Cobots, designed with advanced kinematic and dynamic capabilities [
35
],
aim to incorporate systems and society through intelligent technologies, fostering agility
and resilience. Although there are concerns about job displacement and labor market
changes, cobots can support societies with aging populations rather than entirely replacing
the workforce.
Usually, energy systems offer sophisticated models in terms of interdependency and
interrelation of components and system behavior. Therefore, modeling complex characteris-
tics leads to non-linear patterns that can be time-consuming to handle, such a complicated
task. Evaluation of the dynamic characteristics of an energy system requires an in-depth
analysis before exploring its mathematical models for swapping to ML compiling ar-
chitecture models. Underlying relationships among transformation processes require a
feedback loop to confirm the model performance accuracy with its original version and
to evaluate the impact of sensitivity and variability. Therefore, a feedback loop among
the model processes and interactions at different data exchange levels is required because
inappropriate dataset input leads to poor characteristics and overall performances of the
regenerated model.
Data-driven methods are a promising solution for enabling system parameter-based
analysis, leading to improved operation and reliability [
36
]. However, a lack of standard-
ization is evident in the transformation from system-parameter-based to data-driven-based
models. There is a need for benchmarking among various methodologies, tools, and
techniques being employed. In addition, the scalability and real-world implementation
capabilities of these data-driven models must be demonstrated to achieve widespread
adoption and integration into various industrial domains. Establishing a robust framework
and benchmarking protocols to evaluate these models’ performance, comparability, and
generalizability is important. More research is needed to explore integrating other sources
of building energy flexibility, such as thermal energy storage, batteries, electric vehicles,
and onsite generation. The next step is to keep energy policies updated and aligned with
fast-growing technologies. This requires a principal change in system modeling approaches
from system parameter models to data-driven models, as shown in Figure 1. Measurement
system analysis is used for data collection at two levels: (1) actual system parameters
and performances in real-time, and (2) past or existing data for demand analysis with
high accuracy.
The thematic framework proposed in Figure 1offers a comprehensive roadmap for
transforming an energy system into its decomposable components through a systematic
procedure. This approach minimizes the probability of errors by converting a system
parameter-based model to a data-driven model suitable for ML analysis. A vital advantage
of this framework is its ability to ensure a process that validates both the system model and
the performance equivalency.
A primary transition in system modeling approaches from parameter-based to data-
driven models is required to keep energy systems aligned with fast-growing technologies.
In other words, modeling and analysis methods in old-fashioned energy systems require
swapping to utilize ML inputs (datasets). Some immediate benefits of the power system
data-driven model for various applications compared to the parameter-based system are
summarized in Figure 2, which is idealized from [37].
Energies 2023,16, 3309 5 of 27
Energies 2023, 16, x FOR PEER REVIEW 5 of 26
Figure 1. The thematic framework illustrates the systematic process of transferring the system pa-
rameter-based model to a data-driven based model.
The thematic framework proposed in Figure 1 offers a comprehensive roadmap for
transforming an energy system into its decomposable components through a systematic
procedure. This approach minimizes the probability of errors by converting a system pa-
rameter-based model to a data-driven model suitable for ML analysis. A vital advantage
of this framework is its ability to ensure a process that validates both the system model
and the performance equivalency.
A primary transition in system modeling approaches from parameter-based to data-
driven models is required to keep energy systems aligned with fast-growing technologies.
In other words, modeling and analysis methods in old-fashioned energy systems require
swapping to utilize ML inputs (datasets). Some immediate benefits of the power system
data-driven model for various applications compared to the parameter-based system are
summarized in Figure 2, which is idealized from [37].
Figure 1.
The thematic framework illustrates the systematic process of transferring the system
parameter-based model to a data-driven based model.
The merits and demerits of data-driven models in the energy systems shown in
Figure 3highlight the potential improvements that data-driven models can bring to var-
ious aspects of the energy sector. Accurate forecasting is essential for efficient energy
management; data-driven models excel at identifying patterns and trends in historical
data. Optimal allocation of resources is a priority in the energy sector, and these models
can analyze historical and real-time data to improve efficiency. Enhanced maintenance is
another key advantage, as predictive analytics can identify potential equipment failures,
thereby increasing reliability. Furthermore, the deployment of renewable energy resources
into the energy sector is an urgent necessity to address and mitigate the impacts of climate
change. Data-driven models can facilitate this by accurately predicting production from
renewable sources technically and financially. The drawbacks of data-driven models stem
from the prevalent limitations and challenges encountered while applying such models
within the energy sector. Data quality and availability are critical factors, and limitations in
these areas can hinder the effectiveness of models. The complexity of data-driven models
can present challenges with respect to the specialized knowledge and resources required,
which may be barriers for some energy utilities companies. Overfitting is another common
Energies 2023,16, 3309 6 of 27
technical issue in data-driven models, which can lead to inaccuracies when applied to new,
unseen data. Adaptability is also crucial in the rapidly evolving energy sector; data-driven
models may struggle to keep up with new technologies, regulations, or competitive market
conditions.
Energies 2023, 16, x FOR PEER REVIEW 6 of 26
Figure 2. A visual representation of the advantages and disadvantages of data-driven models in the
energy sector shows the balance between the potential improvements and the challenges these mod-
els face in addressing the sector’s unique requirements and demands.
The merits and demerits of data-driven models in the energy systems shown in Fig-
ure 3 highlight the potential improvements that data-driven models can bring to various
aspects of the energy sector. Accurate forecasting is essential for efficient energy manage-
ment; data-driven models excel at identifying paerns and trends in historical data. Opti-
mal allocation of resources is a priority in the energy sector, and these models can analyze
historical and real-time data to improve efficiency. Enhanced maintenance is another key
advantage, as predictive analytics can identify potential equipment failures, thereby in-
creasing reliability. Furthermore, the deployment of renewable energy resources into the
energy sector is an urgent necessity to address and mitigate the impacts of climate change.
Data-driven models can facilitate this by accurately predicting production from renewable
sources technically and financially. The drawbacks of data-driven models stem from the
prevalent limitations and challenges encountered while applying such models within the
energy sector. Data quality and availability are critical factors, and limitations in these
areas can hinder the effectiveness of models. The complexity of data-driven models can
present challenges with respect to the specialized knowledge and resources required,
which may be barriers for some energy utilities companies. Overfiing is another common
technical issue in data-driven models, which can lead to inaccuracies when applied to
new, unseen data. Adaptability is also crucial in the rapidly evolving energy sector; data-
Figure 2.
A visual representation of the advantages and disadvantages of data-driven models in
the energy sector shows the balance between the potential improvements and the challenges these
models face in addressing the sector’s unique requirements and demands.
Engineering planning and design are crucial to reliable supply with high customer
satisfaction. Accurate system supply forecasting for maintaining stability and meeting
current and future demands necessitates transcending traditional forecasting approaches
that solely depend on historical data and system parameters. When integrating renewable
resources and smartness of supply-demand through instantaneous metering loads, systems
operate using a dynamic nature that traditional methods cannot replicate. Utilizing machine
learning for load forecasting allows power systems to adopt a real-time measurement data
approach rather than relying on system parameter-based data. This can be applied under
steady-state conditions at a single operating point without observing the system’s dynamic
behavior [38].
Energies 2023,16, 3309 7 of 27
Energies 2023, 16, x FOR PEER REVIEW 7 of 26
driven models may struggle to keep up with new technologies, regulations, or competitive
market conditions.
Figure 3. An overview of the research process hierarchy recommended for research and develop-
ment of sustainable energy policies.
Engineering planning and design are crucial to reliable supply with high customer
satisfaction. Accurate system supply forecasting for maintaining stability and meeting
current and future demands necessitates transcending traditional forecasting approaches
that solely depend on historical data and system parameters. When integrating renewable
resources and smartness of supply-demand through instantaneous metering loads, sys-
tems operate using a dynamic nature that traditional methods cannot replicate. Utilizing
machine learning for load forecasting allows power systems to adopt a real-time meas-
urement data approach rather than relying on system parameter-based data. This can be
applied under steady-state conditions at a single operating point without observing the
systemʹs dynamic behavior [38].
Integrating intelligent technologies in the energy sector will significantly impact pol-
icy and system modeling due to changing priorities, investments, and competitive energy
markets. Developing countries, particularly those with predominantly rural communities
adversely affected by climate change and financial instability, are more vulnerable to these
shifts, as they may lack the resources necessary for significant initial investments. Feature
engineering methods, which involve generating new features or aribute modifications,
employ techniques such as creation, filtering, selection, expansion, transformation, or for-
mulation or utilize time-based system performance values to develop novel features [39].
Feature engineering deals with the existing system dataset to improve quality through
optimal feature changes such as adding, modifying, expanding, transforming, filtering,
selecting, encoding, formulation, etc. [33]. In contrast to feature creation, the feature selec-
tion method (forward and backward selections) of the expansion of the existing data set
is used to reduce the size of the feature set; this can cause redundancy and sparsity or
multicollinearity in the data [40]. In addition, ML bestows accuracy with high processing
Figure 3.
An overview of the research process hierarchy recommended for research and development
of sustainable energy policies.
Integrating intelligent technologies in the energy sector will significantly impact policy
and system modeling due to changing priorities, investments, and competitive energy
markets. Developing countries, particularly those with predominantly rural communities
adversely affected by climate change and financial instability, are more vulnerable to these
shifts, as they may lack the resources necessary for significant initial investments. Feature
engineering methods, which involve generating new features or attribute modifications,
employ techniques such as creation, filtering, selection, expansion, transformation, or
formulation or utilize time-based system performance values to develop novel features [
39
].
Feature engineering deals with the existing system dataset to improve quality through
optimal feature changes such as adding, modifying, expanding, transforming, filtering,
selecting, encoding, formulation, etc. [
33
]. In contrast to feature creation, the feature
selection method (forward and backward selections) of the expansion of the existing data
set is used to reduce the size of the feature set; this can cause redundancy and sparsity or
multicollinearity in the data [
40
]. In addition, ML bestows accuracy with high processing
speed, turns complicated knowledge-based status models of system performance and
interconnected behavior into a simple output, creates models from real-time data, and
allows the machine to learn and map optimum output.
Generally, energy systems consist of various setups aligned together in a dynamic oper-
ation nature for a reliable and techno-economic supply. Dynamic and transient behavior of
systems offers sophisticated observation of multidimensional and non-linear relationships
that require a competency analysis package to overcome these challenges. In addition to
the challenge of assessing system insight, envisioning the system’s future for planning and
expansion presents significant difficulties that are not readily achievable using conventional
tools and methods. AI and its specific subsets of ML and DL tools and techniques allow
the energy sector to corroborate with optimizations using numerous packages of in-build
functions that support multidimensional-nonlinear-interrelated systems (subsystems) with
high accuracy and less computational time.
Energies 2023,16, 3309 8 of 27
3. Competitive Energy Policy at the Era of AI
The study paradigm is founded based on triple methodology patterns (except the
“Critical” that is often used for social science studies), establishing a clear understanding of
emerging theories and anticipated outcomes of the research practices, as shown in Figure 3
(adapted from [
41
–
43
]). From the building block perspective, the research paradigm in the
field of energy consistently addresses the importance and reality of practices and analysis
of applied methodologies aligned with the research paradigm in the context of ontology,
epistemology, and research methodology requirements. From the behavioral point of
view, the research paradigm follows mixed-rational positivism (quantitative, experimental),
constructivism (qualitative), and pragmatists (predict and optimize) paradigms. The
predictive approach demonstrates its potential to inform energy policy in the age of AI,
supported by the recommended prescriptive framework’s practical applications.
A research paradigm forms the foundation of the research process, guiding the choice
of philosophy, methodology, and methods. Research philosophy encompasses ontology,
which studies the nature of reality, and epistemology, which examines the nature of knowl-
edge. Research methodology, informed by paradigm and philosophy, includes prognostic
and prescriptive frameworks that predict outcomes and recommend optimal actions. Re-
search methods, such as positivism, constructivism, pragmatism, postpositivism, and
interpretivism, reflect specific data collection and analysis approaches, each with a unique
emphasis on objectivity, subjectivity, or practicality.
The author has extensively researched energy system sustainability, focusing on ad-
dressing real world challenges beyond theoretical suggestions. The proposed topic led to
an in-depth exploration of the literature, revealing the fundamental foundation of long-
term sustainable energy planning. The study highlights the importance of examining
the perspectives on current and future energy trends and the points of view from which
we observe these trends. It is crucial to understand that before proposing solutions, the
study requires the root causes of the challenges, balances demand and the expected growth
rate, and aligns with the philosophy and methodology of effective research with real-time
implications. Consequently, these concepts are briefly discussed in Figure 3to emphasize
the importance of a systematic outlook. Although a detailed examination is beyond the
scope of this study, these ideas will be applied in future research through various scenarios.
Energy policy is a set of laws, regulations, and actions that a government or utility
company takes to manage the generation/import/export, transmission/distribution, and
consumption of energy; these are piloted by an energy policy paradigm in a broad over-
arching framework. The energy policy structure includes institutional arrangements and
administrative frameworks that govern the development, implementation, and evaluation
process of energy policy. A systematic and integrated approach that employs various
analytical tools and techniques, stakeholder participation, evidence-based decision-making,
etc., is defined under the policy methodology. Goals and objectives, values and principles,
and methods and tools, including regulations, incentives, research and development pro-
grams, etc., are the components of a policy paradigm. Tracking and assessing the reliability
and sustainability of the energy supply chain, circular economic resilience, reduction of
greenhouse gas emissions, enhancement of energy efficiency and conservation, investments
in energy infrastructure, and access to affordable energy are just a few of the key aspects
that utilize energy policy indicators and indices. Various indices are introduced to measure
progress towards achieving Sustainable Development Goal 7 (Affordable and Clean En-
ergy), including the Renewable Energy Share Index, Renewable Energy Capacity Index,
Renewable Energy Generation Index, Energy Efficiency Index, Energy Intensity of the
Economy, Renewable Energy Access Index, Renewable Energy Prices Index, Renewable En-
ergy Investment Index, and Renewable Energy Employment Index [
44
]. These indices can
provide quantitative and comparative tools for stakeholders to facilitate communication
and develop strategies using various potential applications, such as measuring progress
toward achieving sustainable energy goals, benchmarking progress between countries,
identifying areas for improvement in energy policies and programs, monitoring and report-
Energies 2023,16, 3309 9 of 27
ing on implementation, promoting renewable energy and energy efficiency, evaluating the
effectiveness of policies, and supporting decision-making processes.
Integrating AI and ML into energy systems can be approached from many perspec-
tives and applied to various use cases. In this context, we focus on the techno-economic
energy balance landscape as an area of interest (Figure 4). This framework encompasses a
wide range of applications, including optimizing generation, economic dispatch, and man-
agement of energy resources and infrastructure. Incorporating AI and ML techniques into
this landscape can lead to more efficient and sustainable energy systems, enabling better
demand forecasting, capacity planning, and resource allocation. Furthermore, integrating
these advanced technologies can help identify new opportunities for cost reduction and
revenue generation while promoting cleaner energy sources and minimizing environmental
impacts. By harnessing AI and ML for techno-economic energy balance, decision-makers,
and policy developers can enhance the resilience and reliability of energy systems, en-
suring a stable power supply even amid fluctuating demand and resource availability.
Furthermore, AI and ML mitigate risks associated with extreme weather events, equipment
failures, and other unforeseen disruptions.
Choosing a method depends on various factors, necessitating an examination of the
problem’s nature and complexity, the availability of data, and the model’s accuracy and
interoperability. Nonetheless, the proposed ML methods are optimally suited for the
intended applications or operations.
Energies 2023,16, 3309 10 of 27
Energies 2023, 16, x FOR PEER REVIEW 10 of 26
Figure 4. An overview of the techno-economic energy balancing landscape highlighting the integra-
tion of AI and ML into various energy system applications by distinguishing the triangle of AI and
ML methods.
Figure 4.
An overview of the techno-economic energy balancing landscape highlighting the integra-
tion of AI and ML into various energy system applications by distinguishing the triangle of AI and
ML methods.
Energies 2023,16, 3309 11 of 27
4. AI and ML: Transforming the Energy Sector
Although AI is gradually permeating all aspects of life, 93% of the existing technical
infrastructure does not meet the optimal integration requirements, not to mention the
upfront costs associated with its deployment. To overcome this issue, platforms must
prepare for a critical change that demands inter-domain collaboration. In addition to data
scientists and AI experts, other engineering domains will deal with AI directly or indirectly,
offering expensive programming, new tools, and techniques. Therefore, this section focuses
on representing selected applied data science tools and techniques in energy engineering,
followed by examples that enable energy field researchers, students, and practitioners
to acquire essential knowledge of applied AI in energy planning and design. This study
aims to categorize the essential ML methods and tools primarily used in energy systems
studies. It then presents these methods to properly approach the appropriate white-box,
gray-box, and black-box techniques [
45
], utilizing suitable programming languages such as
Python, MATLAB
®
, and so on. This creates an interface for simulation packages designed
to achieve the best solutions.
Many studies have employed diverse estimations and weighting adjustments to
reduce biases stemming from nonresponse and non-coverage systems [
46
]. In this context,
priority is given to pairwise comparison using magnitude scaling, ideal point methods [
47
],
and Analytic Hierarchy Process (AHP) methods [
48
]. The weights are determined by
considering the relative importance of each element in the scenarios, with expert judgment
and the Delphi approach for support [
49
,
50
]. However, evaluating qualitative scenarios
with quantitative measures is challenging due to the subjectivity of qualitative data and
proxy limitations [
51
]. This may result in inaccuracies and biases, thereby complicating
interpretation [
52
]. Using appropriate methods and careful understanding is crucial to
ensuring validity and robustness. Therefore, a multidimensional approach utilizing a
systematic methodology (Figure 5), including the AHP method [
53
], is employed to estimate
qualitative weighting scores for the analyzed scenarios [
54
]. AHP is a structured technique
for organizing and analyzing complex decisions involving qualitative and quantitative
factors, providing a consistent and structured approach for weighing non-quantitative
factors and prioritizing scenarios [
55
]. These scenarios can be tailored based on specific
deployment needs and priorities.
Energies 2023, 16, x FOR PEER REVIEW 11 of 26
4. AI and ML: Transforming the Energy Sector
Although AI is gradually permeating all aspects of life, 93% of the existing technical
infrastructure does not meet the optimal integration requirements, not to mention the up-
front costs associated with its deployment. To overcome this issue, platforms must pre-
pare for a critical change that demands inter-domain collaboration. In addition to data
scientists and AI experts, other engineering domains will deal with AI directly or indi-
rectly, offering expensive programming, new tools, and techniques. Therefore, this section
focuses on representing selected applied data science tools and techniques in energy en-
gineering, followed by examples that enable energy field researchers, students, and prac-
titioners to acquire essential knowledge of applied AI in energy planning and design. This
study aims to categorize the essential ML methods and tools primarily used in energy
systems studies. It then presents these methods to properly approach the appropriate
white-box, gray-box, and black-box techniques [45], utilizing suitable programming lan-
guages such as Python, MATLAB
®
, and so on. This creates an interface for simulation
packages designed to achieve the best solutions.
Many studies have employed diverse estimations and weighting adjustments to re-
duce biases stemming from nonresponse and non-coverage systems [46]. In this context,
priority is given to pairwise comparison using magnitude scaling, ideal point methods
[47], and Analytic Hierarchy Process (AHP) methods [48]. The weights are determined by
considering the relative importance of each element in the scenarios, with expert judg-
ment and the Delphi approach for support [49,50]. However, evaluating qualitative sce-
narios with quantitative measures is challenging due to the subjectivity of qualitative data
and proxy limitations [51]. This may result in inaccuracies and biases, thereby complicat-
ing interpretation [52]. Using appropriate methods and careful understanding is crucial
to ensuring validity and robustness. Therefore, a multidimensional approach utilizing a
systematic methodology (Figure 5), including the AHP method [53], is employed to esti-
mate qualitative weighting scores for the analyzed scenarios [54]. AHP is a structured
technique for organizing and analyzing complex decisions involving qualitative and
quantitative factors, providing a consistent and structured approach for weighing non-
quantitative factors and prioritizing scenarios [55]. These scenarios can be tailored based
on specific deployment needs and priorities.
Figure 5.
Analysis process flowchart of policy building block scenarios and AI techniques, offering a
consistent and structured approach to weighing nonquantitative factors, prioritizing, and comparing
the scenario.
Energies 2023,16, 3309 12 of 27
4.1. The Process Flow
This study focuses on multicriteria objective analysis using the Analytic Hierarchy
Process (AHP) (Python codes are given in Appendix A). AHP is a structured decision-
making technique that compares alternatives based on multiple criteria and sub-criteria.
The methodology involves constructing a hierarchy of criteria, assigning weights to criteria
and sub-criteria, and evaluating alternatives based on each sub-criterion. The final output
for each alternative is a normalized score that reflects its overall preference, following the
steps in Figure 6.
Energies 2023, 16, x FOR PEER REVIEW 12 of 26
Figure 5. Analysis process flowchart of policy building block scenarios and AI techniques, offering
a consistent and structured approach to weighing nonquantitative factors, prioritizing, and compar-
ing the scenario.
4.1. The Process Flow
This study focuses on multicriteria objective analysis using the Analytic Hierarchy
Process (AHP) (Python codes are given in Appendix A). AHP is a structured decision-
making technique that compares alternatives based on multiple criteria and sub-criteria.
The methodology involves constructing a hierarchy of criteria, assigning weights to crite-
ria and sub-criteria, and evaluating alternatives based on each sub-criterion. The final out-
put for each alternative is a normalized score that reflects its overall preference, following
the steps in Figure 6.
Figure 6. The two-stage decision-making process consists of the evaluation framework and decision
analysis.
The proposed flowchart and the framework aim to assess the potential applications
of AI and ML techniques (referenced in Table 1) in various energy policy scenarios while
considering sustainability requirements. To achieve this objective, the following criteria
are identified to evaluate the impact of AI and ML methodologies in energy systems:
• Technical feasibility (TF)
• Environmental Impact (EI)
• Economic Viability (EV)
• Social Acceptability (SA)
• Regulatory Compliance (RC)
These criteria are subsequently assessed through a sub-criteria evaluation to gain
deeper insights into the relevant and measurable aspects. The sub-criteria for the five cri-
teria prioritize environmental impact, focusing on carbon emission reduction (as depicted
in Figure 7).
Figure 6.
The two-stage decision-making process consists of the evaluation framework and decision
analysis.
The proposed flowchart and the framework aim to assess the potential applications
of AI and ML techniques (referenced in Table 1) in various energy policy scenarios while
considering sustainability requirements. To achieve this objective, the following criteria are
identified to evaluate the impact of AI and ML methodologies in energy systems:
•Technical feasibility (TF)
•Environmental Impact (EI)
•Economic Viability (EV)
•Social Acceptability (SA)
•Regulatory Compliance (RC)
Energies 2023,16, 3309 13 of 27
Table 1.
The relationship between scenario codes and AI techniques, where 1 indicates that the
technique is used in the given scenario and 0 indicates that it is not.
No. 1
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
References
[56]
[57]
[58,59]
[58,60]
[61]
[61–63]
[64]
[65–67]
[68]
[69]
[61]
[70]
[69]
[71,72]
[71]
[73]
[74,75]
1Code
AI and
Machine
Learning
Methods
Scenario
EEDM
RPS
NM
FIT
CPM
EMD
ECS
AIEC
RETI
SATM
ESST
EVCI
CCS
EDA
EB
ETP
EE
ERD
EMD
ICEC
2
Neural Networks 1
1111110111001010101
3
Decision Trees 1
1111110111001110101
4
Linear Regression 1
0001110110000100000
5
Support Vector Machines 0
1010100000000000000
6
Computer Vision 1
0000000110001110000
7
Natural Language Processing 1
0000000000001101001
8
Time Series Analysis 1
0000000000000000000
9
Clustering 1
0000000000000000000
10
Interdisciplinary Tools 0
0000001000000000000
11
Deep Learning 0
0000000000000010000
12
Evolutionary Algorithms 0
0000000000000000100
13
Particle Swarm Optimization 0
0000000100000000100
14
Convolutional Neural Networks 0
0000000000000000100
15
Generative Models 0
0000000000000000100
16
Sentiment Analysis 0
0000000000001001001
17
Text Classification 0
0000000000001001001
18
Random Forest 0
0001000000000000000
These criteria are subsequently assessed through a sub-criteria evaluation to gain
deeper insights into the relevant and measurable aspects. The sub-criteria for the five
criteria prioritize environmental impact, focusing on carbon emission reduction (as depicted
in Figure 7).
Energies 2023,16, 3309 14 of 27
Energies 2023, 16, x FOR PEER REVIEW 13 of 26
Figure 7. Evaluating criteria using sub-criteria for the five evaluation criteria, focusing on environ-
mental impact.
The analytical hierarchy process (AHP) is used to assign weights to criteria to reflect
their relative importance. This involves creating a pairwise comparison matrix to show
each criterion’s importance when compared to others, normalizing the matrix, and calcu-
lating priority weights. The consistency ratio (CR) is estimated to check the consistency of
the pairwise comparison matrix; if it is less than 0.1, the matrix is considered consistent.
Evaluating alternatives involves several steps, including assessing each item based
on sub-criteria, normalizing the data, calculating the weighted scores, and ranking the
items. First, each item is evaluated on a scale of 1 to 5, based on how well it satisfies each
sub-criterion. Then, the data are normalized by dividing each element by the sum of its
column. The weighted score is then calculated for each item by multiplying its normalized
score for each sub-criterion by the weight of that sub-criterion and proceeding with sum-
ming these products across all sub-criteria. Finally, the items are ranked according to their
total weighted score, with higher scores indicating beer performance.
Analyzing AI and ML techniques in energy systems provides insight into promising
areas of focus. High-ranked items are prioritized for investment and promotion, while
lower-ranked items require further evaluation. The model involves defining the objective,
identifying criteria and sub-criteria, assigning weights using AHP, evaluating
Figure 7.
Evaluating criteria using sub-criteria for the five evaluation criteria, focusing on environ-
mental impact.
The analytical hierarchy process (AHP) is used to assign weights to criteria to reflect
their relative importance. This involves creating a pairwise comparison matrix to show each
criterion’s importance when compared to others, normalizing the matrix, and calculating
priority weights. The consistency ratio (CR) is estimated to check the consistency of the
pairwise comparison matrix; if it is less than 0.1, the matrix is considered consistent.
Evaluating alternatives involves several steps, including assessing each item based on
sub-criteria, normalizing the data, calculating the weighted scores, and ranking the items.
First, each item is evaluated on a scale of 1 to 5, based on how well it satisfies each sub-
criterion. Then, the data are normalized by dividing each element by the sum of its column.
The weighted score is then calculated for each item by multiplying its normalized score
for each sub-criterion by the weight of that sub-criterion and proceeding with summing
these products across all sub-criteria. Finally, the items are ranked according to their total
weighted score, with higher scores indicating better performance.
Analyzing AI and ML techniques in energy systems provides insight into promising
areas of focus. High-ranked items are prioritized for investment and promotion, while
lower-ranked items require further evaluation. The model involves defining the objective,
Energies 2023,16, 3309 15 of 27
identifying criteria and sub-criteria, assigning weights using AHP, evaluating alternatives,
and interpreting results. The weights can be adjusted to reflect priorities and preferences.
The consistency of the pairwise comparison matrix must be checked. The proposed model
offers a structured and comprehensive approach to evaluating AI and ML energy system
deployment techniques.
4.2. Mathematical Representation of the Process
Pairwise Comparison Matrix: A pairwise comparison matrix determines the relative
importance of elements (criteria or sub-criteria) in the decision-making process, assigning a
predefined scale (e.g., 1 to 9). In the below matrix, the elements
aij
represent the preference
of element i over element j [76]:
a11 a12 . . . a1n
a21 a22 . . . a2n
. . . . . . . . . . . .
an1 an2 . . . ann
(1)
A
is the pairwise comparison matrix with dimensions
(n×n)
, where
n
is the number
of elements being compared.
Criteria weights: Calculating the criteria weights can be performed using various
methods, including the Analytic Hierarchy Process (AHP), the Analytic Network Process
(ANP), or direct weighting, to quantify the relative importance of each criterion in the
decision-making process. If
wi
is the weight of criterion
i
, the weights of the criteria are
represented as the following vector [76]:
W=(w1, w2, . . . , wn)(2)
where i =1, 2, ..., n.
Sub-criteria Weights: Similar to criteria weights, sub-criteria weights calculated weight-
ing to create pairwise comparison matrices for each criterion and calculate the correspond-
ing weights for each sub-criterion. For each criterion
i
,
wij
is the weight of sub-criterion
j
.
Then, the sub-criteria weights for the criterion i vector are given as follows [76]:
Wi=wi1, wi2, . . . , wimi(3)
where j =1, 2, ..., mj.
Alternatives and weighted performance matrix: The performance matrix (
P
) is a
prerequisite to track the performance of each alternative regarding each criterion. Multiply
each column of the performance matrix (
P
) by the corresponding weight from the weight
vector of the criteria (W). The weighted performance matrix Pwis given as:
Pw=P·W(4)
P
is the performance matrix with dimensions
(n×m)
, where
n
is the number of
alternatives and m is the number of criteria. Let Wbe the weight vector of the criteria.
Weighted Scores: Multiplying the weighted performance matrix (
Pw
) by the criteria
weights vector (
W
) will result in a weighted score for each alternative, which considers
both the performance of the alternatives and the importance of each criterion.
S=Pw·W(5)
where Sis the score vector for each alternative.
Normalized scores: Calculating the sum of all weighted scores (
∑n
i=1Si
) is a prereq-
uisite for normalizing the scores. Divide each weighted score by the sum of all weighted
Energies 2023,16, 3309 16 of 27
scores to obtain the normalized scores (
Sn
). Normalized scores contribute to comparing
alternatives on a scale from 0 to 1, as given below [76]:
Sum =
n
∑
i=1
Si(6)
Sni=Si
∑n
i=1SiFor all i =1, 2, . . . , n. (7)
where,
n
is the number of alternatives, and
S
and
Sn
are the original and normalized
score vectors.
4.3. Simulation and Results
In this study, we compare five alternatives for a given problem, using the Analytic
Hierarchy Process (AHP) to evaluate their performance and rank them based on their
normalized scores. Alternative 4 (A4) emerges as the top choice, significantly outperforming
the other options. While Alternatives 1, 2, and 5 exhibit relatively close scores, they do
not surpass the performance of A4. Alternative 3 is identified as the least preferred option.
The AHP provides a valuable foundation for our analysis, but it is crucial to recognize that
decision-making often involves multiple factors and complex considerations. Therefore,
our findings should be used as a starting point for further discussions, incorporating
additional factors before reaching a final decision.
Alternative 4 (A4) has the highest normalized score (0.3071), indicating that it is the
most preferred option. The other alternatives have relatively close scores, with A2 second
(0.2146), followed by A5, A1, and A3. The normalized scores highlight each alternative’s
performance: A1 (0.1994) and A2 (0.1989) are good options; A3 (0.1823) is the least preferred;
A4 (0.2192) is the best choice; A5 (0.2002) performs slightly better than A1 and A2, but not as
well as A4. The study’s findings reveal that A4 is the best choice among the five alternatives,
with a remarkable normalized score of 0.2192. While A1, A2, and A5 provide viable options,
they do not surpass A4’s performance. A3, which has the lowest score (0.1823), falls behind
the others. Based on AHP, A4 is the recommended alternative. However, decision-making
encompasses multiple factors, and assigned weights might not capture the full complexity.
The results shown in Figures 8–11 can serve as a basis for further analysis and discussions,
integrating additional factors before reaching a final decision.
Energies 2023,16, 3309 17 of 27
Energies 2023, 16, x FOR PEER REVIEW 17 of 26
Figure 8. Criteria (equal importance assigned to all five criteria) and sub-criteria weights (varying
importance assigned to sub-criteria within each criterion).
Figure 8.
Criteria (equal importance assigned to all five criteria) and sub-criteria weights (varying
importance assigned to sub-criteria within each criterion).
Energies 2023,16, 3309 18 of 27
Energies 2023, 16, x FOR PEER REVIEW 18 of 26
Figure 9. Overall sub-criteria weights (consolidated view of sub-criteria importance, cal-
culated by multiplying criteria weights and sub-criteria weights).
Figure 10. (a) Weighted performance matrix (performance scores of the alternatives across all sub-
criteria), and (b) Normalized scores (relative performance of alternatives, calculated by dividing the
weighted performance matrix scores by their sum).
Figure 9.
Overall sub-criteria weights (consolidated view of sub-criteria importance, calculated by
multiplying criteria weights and sub-criteria weights).
Energies 2023, 16, x FOR PEER REVIEW 18 of 26
Figure 9. Overall sub-criteria weights (consolidated view of sub-criteria importance, cal-
culated by multiplying criteria weights and sub-criteria weights).
Figure 10. (a) Weighted performance matrix (performance scores of the alternatives across all sub-
criteria), and (b) Normalized scores (relative performance of alternatives, calculated by dividing the
weighted performance matrix scores by their sum).
Figure 10.
(
a
) Weighted performance matrix (performance scores of the alternatives across all sub-
criteria), and (
b
) Normalized scores (relative performance of alternatives, calculated by dividing the
weighted performance matrix scores by their sum).
Energies 2023,16, 3309 19 of 27
Energies 2023, 16, x FOR PEER REVIEW 19 of 26
Figure 11. Exploring the influence of the proposed sub-criteria on ai-powered energy policy through
visualization of sub-criteria weights across five criteria.
Figure 11.
Exploring the influence of the proposed sub-criteria on ai-powered energy policy through
visualization of sub-criteria weights across five criteria.
Energies 2023,16, 3309 20 of 27
The results demonstrate the prioritization of different criteria and sub-criteria in
decision-making. The equal weighting of the criteria (0.2 for each) indicates that all criteria
have the same importance. The sub-criteria weights show varying levels of importance
within each criterion. The overall sub-criteria weights, calculated by multiplying the criteria
and sub-criteria weights, provide a consolidated view of the importance of the sub-criteria.
The weighted performance matrix represents the performance scores of the alternatives
across all sub-criteria; it shows that the fourth alternative (11.76) performs the best, while
the third alternative (9.78) performs the worst. Normalized scores, calculated by dividing
the weighted performance matrix scores by their sum, further support these findings. As
mentioned earlier, the fourth alternative has the highest normalized score (0.21919851),
indicating its superiority, while the third alternative has the lowest score (0.18229264),
indicating its inferior performance. The results provide a strong argument for selecting the
fourth alternative as the most suitable option based on the criteria and sub-criteria. This
study can be considered to be a roadmap for various scenarios based on objectives and
priorities from different perspectives.
Integrating AI into energy systems has excellent potential to improve efficiency, opti-
mize resource allocation, and facilitate the transition to a sustainable future [
77
]. However,
it also carries inherent risks, including increased vulnerability to cyberattacks, unintended
consequences from algorithmic biases, and possible job displacement; these risks must be
carefully assessed and mitigated to ensure the responsible and beneficial implementation
of AI technologies [78].
5. Conclusions
The study applied a multicriteria and multiobjective approach to evaluate various
scenarios. This proposed method enables decision-makers in the energy sector to structure
complex problems, prioritize factors, and systematically compare options. Moreover, the
proposed models and a systematic roadmap can be applied to various decision-making
scenarios, including business strategy, resource allocation, project management, and circular
economy. This study showcases its versatility and adaptability in interdisciplinary settings.
Therefore, this is a novel contribution to interdisciplinary endeavors from methodological,
comprehensiveness, and implementation perspectives. This research does not replicate the
existing literature or validate previous findings but presents an innovative thesis.
Additionally, this study exhaustively examines the pros and cons of data-driven
models in energy systems, comparing them with model-based system parameters. This
analysis provides a balanced perspective on the potential benefits and challenges associated
with the use of data-driven models in the energy sector, empowering stakeholders and
utilities to make informed decisions about adopting these methodologies. Simultaneously,
integrating AI and ML within the techno-economic energy balancing landscape presents
a promising opportunity to enhance the efficiency, sustainability, and resilience of energy
systems. This has the potential to catalyze transformative change across the industry.
Funding: This research received no external funding.
Data Availability Statement:
https://github.com/mirsayedshah/ai_and_expert_insights_for_susta
inable_energy_future.git.
Acknowledgments:
The author extends his profound gratitude to those who contributed significantly
to the completion of this article. Your steadfast support, wise comments and invaluable guidance,
particularly during crucial moments, have played an essential role in refining and shaping this work.
The author sincerely appreciates your generous contributions and commitment to embark on this
journey with us. The author thanks you for your time, effort, and unwavering dedication to the
success of this article.
Conflicts of Interest: The author declares no conflict of interest.
Energies 2023,16, 3309 21 of 27
Appendix A
This appendix contains the Python language code used to analyze potential AI and ML
techniques applications in various energy policy scenarios while considering sustainability
requirements. The code performs the following steps:
•Define criteria and sub-criteria related to energy policy scenarios.
•Construct a pairwise comparison matrix for criteria.
•Normalize the pairwise comparison matrix to obtain criteria weights.
•Define the weights for the criterion of sub-criteria.
•Calculate the overall weights of sub-criteria.
•Define alternatives for exploration.
•
Create a performance matrix for each alternative, evaluating their performance against
the sub-criteria.
•
Multiply the performance matrix by the overall sub-criteria weights to obtain a
weighted performance matrix.
•Normalize the weighted scores for each alternative.
•
Print the results, including criteria weights, sub-criteria weights, overall sub-criteria
weights, weighted performance matrix, and normalized scores.
The code provides insights into the relative importance of different criteria and sub-
criteria for energy policy scenarios and the evaluation of alternative solutions. The results
can be used to inform decision-making processes and guide the development of energy
policies that leverage AI and ML techniques while addressing sustainability requirements.
Energies 2023,16, 3309 22 of 27
Energies 2023, 16, x FOR PEER REVIEW 21 of 26
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
from matplotlib.colors import LinearSegmentedColormap
# Objective: Potential applications of AI and machine learning
techniques (referenced in Table 1) in various energy policy
scenarios while considering sustainability requirements
# Criteria and sub-criteria
criteria = ['TF', 'EI', 'EV', 'SA', 'RC']
sub_criteria = [ 'EEDM', 'RPS', 'NM', 'FIT', 'CPM', 'EMD', 'ECS',
'AIEC', 'RETI', 'SATM', 'ESST', 'EVCI', 'CCS', 'EDA', 'EB',
'ETP', 'EE', 'ERD', 'EMD', 'ICEC']
# Criteria and sub-criteria abbreviations explaination or
alternative option for using themselved instead of
abbreviations.
# criteria = ['Technical Feasibility', 'Environmental Impact',
'Economic Viability', 'Social Acceptability', 'Regulatory
Compliance']
# sub_criteria = ['Energy Efficiency and Demand Management',
'Renewable Portfolio Standard', 'Net Metering', 'Feed-in
Tariff', 'Carbon Pricing Mechanism', 'Electricity Market
Design', 'Energy Control System', 'Advanced Inverter Energy
Control', 'Renewable Energy Technology Integration', 'Smart
Appliances and Thermostats Management', 'Energy Storage System
Technologies', 'Electric Vehicle Charging Infrastructure',
'Carbon Capture and Storage', 'Energy Data Analytics', 'Energy
Balance', 'Energy Trading Platform', 'Energy Efficiency',
'Energy Resource Diversification', 'Energy Market Design',
'Integrated Community Energy Challenges']
# The pairwise comparison matrix
pairwise_matrix = np.array([
[1, 5, 3, 7, 2],
[1/5, 1, 2, 5, 1/3],
[1/3, 1/2, 1, 3, 1/2],
[1/7, 1/5, 1/3, 1, 1/4],
[1/2, 3, 2, 4, 1]
Energies 2023,16, 3309 23 of 27
Energies 2023, 16, x FOR PEER REVIEW 22 of 26
])
# The pairwise comparison matrix normalization
weights_matrix = pairwise_matrix / pairwise_matrix.sum(axis=1,
keepdims=True)
# Weights for each criterion
criteria_weights = weights_matrix.mean(axis=1)
# Weights for criterion of the sub-criteria
sub_criteria_weights = np.array([
[0.2, 0.1, 0.05, 0.15, 0.2, 0.1, 0.05, 0.05, 0.05, 0.05, 0.05,
0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05],
[0.1, 0.15, 0.2, 0.1, 0.15, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05,
0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05],
[0.05, 0.05, 0.1, 0.1, 0.1, 0.1, 0.05, 0.05, 0.05, 0.05, 0.1,
0.05, 0.05, 0.05, 0.05, 0.1, 0.05, 0.05, 0.05, 0.05],
[0.1, 0.1, 0.15, 0.05, 0.1, 0.1, 0.1, 0.05, 0.05, 0.05, 0.05,
0.1, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05],
[0.15, 0.1, 0.1, 0.1, 0.1, 0.1, 0.05, 0.05, 0.05, 0.05, 0.05,
0.05, 0.1, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05, 0.05]
])
# Overall weights of sub-criteria
overall_sub_criteria_weights = np.dot(criteria_weights,
sub_criteria_weights)
# Alternatives exploration
alternatives = ['A1', 'A2', 'A3', 'A4', 'A5']
# Tperformance matrix for each alternative
performance_matrix = np.array([
[8, 6, 7, 9, 7, 8, 8, 7, 9, 9, 8, 7, 7, 6, 8, 7, 8, 8, 7, 9],
[7, 8, 8, 7, 8, 7, 7, 8, 7, 7, 9, 8, 8, 7, 7, 8, 7, 8, 8, 7],
[6, 7, 7, 8, 7, 7, 7, 6, 8, 7, 7, 6, 6, 8, 7, 7, 6, 7, 7, 8],
[9, 9, 8, 8, 8, 9, 9, 8, 7, 8, 7, 9, 8, 8, 8, 9, 9, 9, 8, 8],
[8, 7, 7, 7, 8, 8, 7, 9, 8, 8, 8, 7, 8, 7, 8, 8, 7, 7, 9, 7]
])
# Multiplying the performance matrix by the overall sub-criteria
weights matrix
Energies 2023,16, 3309 24 of 27
Energies 2023, 16, x FOR PEER REVIEW 23 of 26
weighted_performance_matrix = performance_matrix @
overall_sub_criteria_weights
# Normalizing the weighted scores
normalized_scores = weighted_performance_matrix /
weighted_performance_matrix.sum()
# Print the results
print("Criteria weights:")
print(criteria_weights)
print("Sub-criteria weights:")
print(sub_criteria_weights)
print("Overall sub-criteria weights:")
print(overall_sub_criteria_weights)
print("Weighted performance matrix:")
print(weighted_performance_matrix)
print("Normalized scores:")
print(normalized_scores)
References
1. Thamik, H.; Wu, J. The Impact of Artificial Intelligence on Sustainable Development in Electronic Markets. Sustainability 2022,
14, 3568. hps://doi.org/10.3390/su14063568.
2. Venkatasubramanian, V. The Promise of Artificial Intelligence in Chemical Engineering: Is It Here, Finally? AIChE J. 2019, 65,
466–478. hps://doi.org/10.1002/aic.16489.
3. Kathirgamanathan, A.; De Rosa, M.; Mangina, E.; Finn, D.P. Data-Driven Predictive Control for Unlocking Building Energy
Flexibility: A Review. Renew. Sustain. Energy Rev. 2021, 135, 110120. hps://doi.org/10.1016/j.rser.2020.110120.
4. Dobbe, R.; Hidalgo-Gonzalez, P.; Karagiannopoulos, S.; Henriquez-Auba, R.; Hug, G.; Callaway, D.S.; Tomlin, C.J. Learning to
Control in Power Systems: Design and Analysis Guidelines for Concrete Safety Problems. Electr. Power Syst. Res. 2020, 189,
106615. hps://doi.org/10.1016/j.epsr.2020.106615.
5. Zienkiewicz, A.K.; Ladu, F.; Barton, D.A.W.; Porfiri, M.; Bernardo, M.D. Data-Driven Modelling of Social Forces and Collective
Behaviour in Zebrafish. J. Theor. Biol. 2018, 443, 39–51. hps://doi.org/10.1016/j.jtbi.2018.01.011.
6. Geneva, N.; Zabaras, N. Quantifying Model Form Uncertainty in Reynolds-Averaged Turbulence Models with Bayesian Deep
Neural Networks. J. Comput. Phys. 2019, 383, 125–147. hps://doi.org/10.1016/j.jcp.2019.01.021.
7. Arridge, S.; Maass, P.; Öktem, O.; Schönlieb, C.-B. Solving Inverse Problems Using Data-Driven Models. Acta Numer. 2019, 28,
1–174. hps://doi.org/10.1017/S0962492919000059.
8. Singh, S.K.; Yang, R.; Behjat, A.; Rai, R.; Chowdhury, S.; Matei, I. PI-LSTM: Physics-Infused Long Short-Term Memory Network.
In Proceedings of the 2019 18th IEEE International Conference on Machine Learning and Applications (ICMLA), Boca Raton,
FL, USA, 16–19 December 2019; pp. 34–41.
9. Kollmann, H.T.; Abueidda, D.W.; Koric, S.; Guleryuz, E.; Sobh, N.A. Deep Learning for Topology Optimization of 2D Met-
amaterials. Mater. Des. 2020, 196, 109098. hps://doi.org/10.1016/j.matdes.2020.109098.
10. Runge, J.; Zmeureanu, R. Forecasting Energy Use in Buildings Using Artificial Neural Networks: A Review. Energies 2019, 12,
3254. hps://doi.org/10.3390/en12173254.
References
1.
Thamik, H.; Wu, J. The Impact of Artificial Intelligence on Sustainable Development in Electronic Markets. Sustainability
2022
,14,
3568. [CrossRef]
2.
Venkatasubramanian, V. The Promise of Artificial Intelligence in Chemical Engineering: Is It Here, Finally? AIChE J.
2019
,65,
466–478. [CrossRef]
3.
Kathirgamanathan, A.; De Rosa, M.; Mangina, E.; Finn, D.P. Data-Driven Predictive Control for Unlocking Building Energy
Flexibility: A Review. Renew. Sustain. Energy Rev. 2021,135, 110120. [CrossRef]
4.
Dobbe, R.; Hidalgo-Gonzalez, P.; Karagiannopoulos, S.; Henriquez-Auba, R.; Hug, G.; Callaway, D.S.; Tomlin, C.J. Learning to
Control in Power Systems: Design and Analysis Guidelines for Concrete Safety Problems. Electr. Power Syst. Res.
2020
,189,
106615. [CrossRef]
5.
Zienkiewicz, A.K.; Ladu, F.; Barton, D.A.W.; Porfiri, M.; Bernardo, M.D. Data-Driven Modelling of Social Forces and Collective
Behaviour in Zebrafish. J. Theor. Biol. 2018,443, 39–51. [CrossRef] [PubMed]
6.
Geneva, N.; Zabaras, N. Quantifying Model Form Uncertainty in Reynolds-Averaged Turbulence Models with Bayesian Deep
Neural Networks. J. Comput. Phys. 2019,383, 125–147. [CrossRef]
7.
Arridge, S.; Maass, P.; Öktem, O.; Schönlieb, C.-B. Solving Inverse Problems Using Data-Driven Models. Acta Numer.
2019
,28,
1–174. [CrossRef]
8.
Singh, S.K.; Yang, R.; Behjat, A.; Rai, R.; Chowdhury, S.; Matei, I. PI-LSTM: Physics-Infused Long Short-Term Memory Network.
In Proceedings of the 2019 18th IEEE International Conference on Machine Learning and Applications (ICMLA), Boca Raton, FL,
USA, 16–19 December 2019; pp. 34–41.
9.
Kollmann, H.T.; Abueidda, D.W.; Koric, S.; Guleryuz, E.; Sobh, N.A. Deep Learning for Topology Optimization of 2D Metamateri-
als. Mater. Des. 2020,196, 109098. [CrossRef]
10.
Runge, J.; Zmeureanu, R. Forecasting Energy Use in Buildings Using Artificial Neural Networks: A Review. Energies
2019
,12,
3254. [CrossRef]
11.
Baptista, M.; Henriques, E.M.P.; de Medeiros, I.P.; Malere, J.P.; Nascimento, C.L.; Prendinger, H. Remaining Useful Life Estimation
in Aeronautics: Combining Data-Driven and Kalman Filtering. Reliab. Eng. Syst. Saf. 2019,184, 228–239. [CrossRef]
Energies 2023,16, 3309 25 of 27
12.
Terzi´c, E.; Lazzari, P.; Organelli, E.; Solidoro, C.; Salon, S.; D’Ortenzio, F.; Conan, P. Merging Bio-Optical Data from Biogeochemical-
Argo Floats and Models in Marine Biogeochemistry. Biogeosciences 2019,16, 2527–2542. [CrossRef]
13.
Nokhandan, A.K.; Snieder, E.; Khan, U.T.; ElDyasti, A.; Ghaemi, Z.; Bagheri, M. A Comparison of Two Data-Driven Models to
Predict Hypolimnetic Dissolved Oxygen Concentration: A Case Study of the Seymareh Reservoir in Iran. J. Environ. Inform. Lett.
2019,2, 70–81. [CrossRef]
14.
Bättig, P.; Schiffmann, J. Data-Driven Model for the Dynamic Characteristics of O-Rings for Gas Bearing Supported Rotors. J.
Appl. Mech. 2019,86, 081003. [CrossRef]
15.
Bellizio, F.; Karagiannopoulos, S.; Aristidou, P.; Hug, G. Optimized Local Control for Active Distribution Grids Using Machine
Learning Techniques. In Proceedings of the 2018 IEEE Power & Energy Society General Meeting (PESGM), Portland, OR, USA,
5–9 August 2018; pp. 1–5.
16.
Dobbe, R.; Sondermeijer, O.; Fridovich-Keil, D.; Arnold, D.; Callaway, D.; Tomlin, C. Toward Distributed Energy Services:
Decentralizing Optimal Power Flow with Machine Learning. IEEE Trans. Smart Grid 2020,11, 1296–1306. [CrossRef]
17.
Brahma, S.; Kavasseri, R.; Cao, H.; Chaudhuri, N.R.; Alexopoulos, T.; Cui, Y. Real-Time Identification of Dynamic Events in Power
Systems Using PMU Data, and Potential Applications—Models, Promises, and Challenges. IEEE Trans. Power Deliv.
2017
,32,
294–301. [CrossRef]
18.
Ardakanian, O.; Wong, V.W.S.; Dobbe, R.; Low, S.H.; von Meier, A.; Tomlin, C.J.; Yuan, Y. On Identification of Distribution Grids.
IEEE Trans. Control Netw. Syst. 2019,6, 950–960. [CrossRef]
19.
Sun, M.; Cremer, J.; Strbac, G. A Novel Data-Driven Scenario Generation Framework for Transmission Expansion Planning with
High Renewable Energy Penetration. Appl. Energy 2018,228, 546–555. [CrossRef]
20. Zhao, C.; Guan, Y. Data-Driven Stochastic Unit Commitment for Integrating Wind Generation. IEEE Trans. Power Syst. 2016,31,
2587–2596. [CrossRef]
21.
Deka, D.; Backhaus, S.; Chertkov, M. Structure Learning in Power Distribution Networks. IEEE Trans. Control Netw. Syst.
2018
,5,
1061–1074. [CrossRef]
22.
Liao, Y.; Weng, Y.; Liu, G.; Rajagopal, R. Urban MV and LV Distribution Grid Topology Estimation via Group Lasso. IEEE Trans.
Power Syst. 2019,34, 12–27. [CrossRef]
23.
Glavic, M.; Fonteneau, R.; Ernst, D. Reinforcement Learning for Electric Power System Decision and Control: Past Considerations
and Perspectives. IFAC-Pap. 2017,50, 6918–6927. [CrossRef]
24.
Ernst, D.; Glavic, M.; Wehenkel, L. Power Systems Stability Control: Reinforcement Learning Framework. IEEE Trans. Power Syst.
2004,19, 427–435. [CrossRef]
25.
Karagiannopoulos, S.; Dobbe, R.; Aristidou, P.; Callaway, D.; Hug, G. Data-Driven Control Design Schemes in Active Distribution
Grids: Capabilities and Challenges. In Proceedings of the 2019 IEEE Milan PowerTech, Milano, Italy, 23–27 June 2019; pp. 1–6.
26.
Karagiannopoulos, S.; Aristidou, P.; Hug, G. Data-Driven Local Control Design for Active Distribution Grids Using Off-Line
Optimal Power Flow and Machine Learning Techniques. IEEE Trans. Smart Grid 2019,10, 6461–6471. [CrossRef]
27.
Xiong, R.; Cao, J.; Yu, Q. Reinforcement Learning-Based Real-Time Power Management for Hybrid Energy Storage System in the
Plug-In Hybrid Electric Vehicle. Appl. Energy 2018,211, 538–548. [CrossRef]
28.
Lu, R.; Hong, S.H. Incentive-Based Demand Response for Smart Grid with Reinforcement Learning and Deep Neural Network.
Appl. Energy 2019,236, 937–949. [CrossRef]
29.
Lesage-Landry, A.; Taylor, J.A. Setpoint Tracking With Partially Observed Loads. IEEE Trans. Power Syst.
2018
,33,
5615–5627. [CrossRef]
30.
Duan, C.; Jiang, L.; Fang, W.; Liu, J. Data-Driven Affinely Adjustable Distributionally Robust Unit Commitment. IEEE Trans.
Power Syst. 2018,33, 1385–1398. [CrossRef]
31.
Jokar, P.; Arianpoo, N.; Leung, V.C.M. Electricity Theft Detection in AMI Using Customers’ Consumption Patterns. IEEE Trans.
Smart Grid 2016,7, 216–226. [CrossRef]
32.
Sutton, C.; Ghiringhelli, L.M.; Yamamoto, T.; Lysogorskiy, Y.; Blumenthal, L.; Hammerschmidt, T.; Golebiowski, J.R.; Liu, X.;
Ziletti, A.; Scheffler, M. Crowd-Sourcing Materials-Science Challenges with the NOMAD 2018 Kaggle Competition. Npj Comput.
Mater. 2019,5, 111. [CrossRef]
33.
Wilfling, S. Augmenting Data-Driven Models for Energy Systems through Feature Engineering: A Python Framework for Feature
Engineering. arXiv 2023, arXiv:2301.01720.
34. Huang, S.; Wang, B.; Li, X.; Zheng, P.; Mourtzis, D.; Wang, L. Industry 5.0 and Society 5.0—Comparison, Complementation and
Co-Evolution. J. Manuf. Syst. 2022,64, 424–428. [CrossRef]
35.
Rossato, C.; Pluchino, P.; Cellini, N.; Jacucci, G.; Spagnolli, A.; Gamberini, L. Facing with Collaborative Robots: The Subjective
Experience in Senior and Younger Workers. Cyberpsychology Behav. Soc. Netw. 2021,24, 349–356. [CrossRef] [PubMed]
36.
Feng, Y.; Cui, N.; Chen, Y.; Gong, D.; Hu, X. Development of Data-Driven Models for Prediction of Daily Global Horizontal
Irradiance in Northwest China. J. Clean. Prod. 2019,223, 136–146. [CrossRef]
37.
Danish, M.S.S.; Senjyu, T. AI-Powered Policy for Energy Sustainability 2023. Available online: https://www.researchgate.net/p
ublication/369850945_AI-Enabled_Energy_Policy_for_a_Sustainable_Future (accessed on 3 April 2023).
38.
Dewangan, F.; Abdelaziz, A.Y.; Biswal, M. Load Forecasting Models in Smart Grid Using Smart Meter Information: A Review.
Energies 2023,16, 1404. [CrossRef]
Energies 2023,16, 3309 26 of 27
39.
Wilfling, S.; Falay, B.; Alfalouji, Q.; Schweiger, G. A Dymola-Python Framework for Data-Driven Model Creation and Co-
Simulation. In Proceedings of the Asian Modelica Conference 2022, Tokyo, Japan, 24–25 November 2022; The Modelica
Conference: Tokyo, Japan, 2022; pp. 165–170.
40.
Zhang, G.; Tian, C.; Li, C.; Zhang, J.J.; Zuo, W. Accurate Forecasting of Building Energy Consumption via a Novel Ensembled
Deep Learning Method Considering the Cyclic Feature. Energy 2020,201, 117531. [CrossRef]
41. Kankam, P.K. The Use of Paradigms in Information Research. Libr. Inf. Sci. Res. 2019,41, 85–92. [CrossRef]
42.
DeCarlo, M. 6.2 Paradigms, Theories, and How They Shape a Researcher’s Approach. In Scientific Inquiry in Social Work;
Pressbooks: Montréal, QC, Canada, 2018; pp. 144–152.
43.
Proofed Research Paradigms: Explanation and Examples. Proofed. 2022. Available online: https://proofed.com/writing-tips/res
earch-paradigms-explanation-and-examples/ (accessed on 12 January 2023).
44.
UNEP GOAL 7: Affordable and Clean Energy. Available online: http://www.unep.org/explore-topics/sustainable-developmen
t-goals/why-do-sustainable-development-goals-matter/goal-7 (accessed on 24 November 2022).
45.
Wilfling, S.; Ebrahimi, M.; Alfalouji, Q.; Schweiger, G.; Basirat, M. Learning Non-Linear White-Box Predictors: A Use Case in
Energy Systems. In Proceedings of the 2022 21st International Conference on Machine Learning and Applications (ICMLA),
Nassau, Bahamas, 12–14 December 2022; pp. 507–512. [CrossRef]
46. Kalton, G.; Flores-Cervantes, I. Weighting Methods. J. Off. Stat. 2003,19, 81–97.
47.
Kok, M.; Lootsma, F.A. Pairwise-Comparison Methods in Multiple Objective Programming, with Applications in a Long-Term
Energy-Planning Model. Eur. J. Oper. Res. 1985,22, 44–55. [CrossRef]
48.
Redfoot, E.K.; Verner, K.M.; Borrelli, R.A. Applying Analytic Hierarchy Process to Industrial Process Design in a Nuclear
Renewable Hybrid Energy System. Prog. Nucl. Energy 2022,145, 104083. [CrossRef]
49.
Van Schoubroeck, S.; Springael, J.; Van Dael, M.; Malina, R.; Van Passel, S. Sustainability Indicators for Biobased Chemicals: A
Delphi Study Using Multicriteria Decision Analysis. Resour. Conserv. Recycl. 2019,144, 198–208. [CrossRef]
50.
Project Management Institute. A Guide to the Project Management Body of Knowledge (PMBOK Guide), 6th ed.; Project Management
Institute: Newtown Square, PA, USA, 2017; ISBN 978-1-62825-184-5.
51.
Chen, P. On the Diversity-Based Weighting Method for Risk Assessment and Decision-Making about Natural Hazards. Entropy
2019,21, 269. [CrossRef]
52.
Lowhorn, G.L. Qualitative and Quantitative Research: How to Choose the Best Design. In Proceedings of the Academic Business
World International Conference, Nashville, TN, USA, 28 May 2007; pp. 1–5.
53.
Ramanathan, R. Multicriteria Analysis of Energy. In Encyclopedia of Energy; Cleveland, C.J., Ed.; Elsevier: New York, NY, USA,
2004; pp. 77–88. ISBN 978-0-12-176480-7.
54.
Abu Taha, R.; Daim, T. Multicriteria Applications in Renewable Energy Analysis, a Literature Review. In Research and Technology
Management in the Electricity Industry; Daim, T., Oliver, T., Kim, J., Eds.; Green Energy and Technology; Springer: London, UK,
2013; pp. 17–30. ISBN 978-1-4471-5096-1.
55.
Miciuła, I.; Nowakowska-Grunt, J. Using the AHP Method to Select an Energy Supplier for Household in Poland. Procedia Comput.
Sci. 2019,159, 2324–2334. [CrossRef]
56.
Krarti, M. Evaluation of Energy Efficiency Potential for the Building Sector in the Arab Region. Energies
2019
,12, 4279. [CrossRef]
57.
Li, W.; Lu, C.; Zhang, Y.-W. Prospective Exploration of Future Renewable Portfolio Standard Schemes in China via a Multi-Sector
CGE Model. Energy Policy 2019,128, 45–56. [CrossRef]
58.
Gabr, A.Z.; Helal, A.A.; Abbasy, N.H. Economic Evaluation of Rooftop Grid-Connected Photovoltaic Systems for Residential
Building in Egypt. Int. Trans. Electr. Energy Syst. 2020,30, e12379. [CrossRef]
59.
Huang, H.; Nie, S.; Lin, J.; Wang, Y.; Dong, J. Optimization of Peer-To-Peer Power Trading in a Microgrid with Distributed PV and
Battery Energy Storage Systems. Sustainability 2020,12, 923. [CrossRef]
60.
Mormann, F. Who Is Going to Pay for and Benefit from the Expansion of Solar Power? Natl. Interest. 2021. Available
online: https://nationalinterest.org/blog/reboot/who-going-pay-and-benefit-expansion-solar-power-194923 (accessed on 12
January 2023).
61.
Cao, K.; Xu, X.; Wu, Q.; Zhang, Q. Optimal Production and Carbon Emission Reduction Level under Cap-And-Trade and Low
Carbon Subsidy Policies. J. Clean. Prod. 2017,167, 505–513. [CrossRef]
62.
Li, H.; Li, R.; Shang, M.; Liu, Y.; Su, D. Cooperative Decisions of Competitive Supply Chains Considering Carbon Trading
Mechanism. Int. J. Low-Carbon Technol. 2022,17, 102–117. [CrossRef]
63.
Tong, W.; Mu, D.; Zhao, F.; Mendis, G.P.; Sutherland, J.W. The Impact of Cap-and-Trade Mechanism and Consumers’ Environ-
mental Preferences on a Retailer-Led Supply Chain. Resour. Conserv. Recycl. 2019,142, 88–100. [CrossRef]
64.
Inoue, N.; Matsumoto, S. An Examination of Losses in Energy Savings after the Japanese Top Runner Program? Energy Policy
2019,124, 312–319. [CrossRef]
65.
Sabory, N.R.; Senjyu, T.; Momand, A.H.; Waqfi, H.; Saboor, N.; Mobarez, R.; Razeqi, F. LEED Scores of Residential Buildings in
Poor Cities: Kabul City Case. Sustainability 2021,13, 6959. [CrossRef]
66.
Castrillón-Mendoza, R.; Rey-Hernández, J.M.; Rey-Martínez, F.J. Industrial Decarbonization by a New Energy-Baseline Method-
ology. Case Study. Sustainability 2020,12, 1960. [CrossRef]
67.
Arcelay, I.; Goti, A.; Oyarbide-Zubillaga, A.; Akyazi, T.; Alberdi, E.; Garcia-Bringas, P. Definition of the Future Skills Needs of Job
Profiles in the Renewable Energy Sector. Energies 2021,14, 2609. [CrossRef]
Energies 2023,16, 3309 27 of 27
68.
Majeed, A.; Ahmad, M.; Rasheed, M.F.; Khan, M.K.; Popp, J.; Oláh, J. The Dynamic Impact of Financial Globalization, Environ-
mental Innovations and Energy Productivity on Renewable Energy Consumption: Evidence From Advanced Panel Techniques.
Front. Environ. Sci. 2022,10, 447. [CrossRef]
69. Omitaomu, O.A.; Niu, H. Artificial Intelligence Techniques in Smart Grid: A Survey. Smart Cities 2021,4, 548–568. [CrossRef]
70.
Gładysz, P.; Strojny, M.; Bartela, Ł.; Hacaga, M.; Froehlich, T. Merging Climate Action with Energy Security through CCS—A
Multi-Disciplinary Framework for Assessment. Energies 2022,16, 35. [CrossRef]
71.
Guo, F.; Chen, Z.; Xiao, F.; Li, A.; Shi, J. Real-Time Energy Performance Benchmarking of Electric Vehicle Air Conditioning
Systems Using Adaptive Neural Network and Gaussian Process Regression. Appl. Therm. Eng. 2023,222, 119931. [CrossRef]
72.
Hu, M. Response to Another Look at “2019 Energy Benchmarking Data for LEED-Certified Buildings in Washington, DC:
Simulation and Reality”. J. Build. Eng. 2022,46, 103694. [CrossRef]
73.
Middleton, P. Sustainable Living Education: Techniques to Help Advance the Renewable Energy Transformation. Sol. Energy
2018,174, 1016–1018. [CrossRef]
74.
Gu, J.; Umar, M.; Soran, S.; Yue, X.-G. Exacerbating Effect of Energy Prices on Resource Curse: Can Research and Development Be
a Mitigating Factor? Resour. Policy 2020,67, 101689. [CrossRef]
75.
Pandey, N.; de Coninck, H.; Sagar, A.D. Beyond Technology Transfer: Innovation Cooperation to Advance Sustainable Develop-
ment in Developing Countries. WIREs Energy Environ. 2022,11, e422. [CrossRef]
76.
Khalid, H.M.; Flitti, F.; Mahmoud, M.S.; Hamdan, M.M.; Muyeen, S.M.; Dong, Z.Y. Wide Area Monitoring System Operations in
Modern Power Grids: A Median Regression Function-Based State Estimation Approach towards Cyber Attacks. Sustain. Energy
Grids Netw. 2023,34, 101009. [CrossRef]
77.
Said, Z.; Sharma, P.; Thi Bich Nhuong, Q.; Bora, B.J.; Lichtfouse, E.; Khalid, H.M.; Luque, R.; Nguyen, X.P.; Hoang,
A.T. Intelligent Approaches for Sustainable Management and Valorisation of Food Waste. Bioresour. Technol.
2023
,377,
128952. [CrossRef] [PubMed]
78.
Mahmoud, M.S.; Khalid, H.M.; Hamdan, M.M. Cyberphysical Infrastructures in Power Systems: Architectures and Vulnerabilities,
1st ed.; Academic Press: Cambridge, UK, 2021; ISBN 978-0-323-85261-6.
Disclaimer/Publisher’s Note:
The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.
