ChapterPDF Available

Minimizing Daily Electricity Cost Using Bird Chase Scheme with Electricity Management Controller in a Smart Home


Abstract and Figures

Integration of Demand Side Management (DSM) strategies within Smart Grid (SG) helps the utilities to mange and control the power consumer load to meet the power demand. Schemes adapted by DSM are used for reducing the load on utilities at peak time, which is achieved by managing the user appliances according to the changes in load on utility and individual smart home. This work is focused on hourly scheduling of the appliances being used in a smart home targeting the daily electricity cost minimization. A new heuristic scheme is introduced for hourly appliances scheduling on user side in this paper. The proposed scheme works at the electricity management controller level, installed in a smart home, within a SG infrastructure. The proposed scheme results are compared with other heuristic schemes as well. From extensive simulations it is depicted that proposed scheme performs best and outperforms other schemes in term of electricity cost minimization.
Content may be subject to copyright.
Minimizing Daily Electricity Cost Using
Bird Chase Scheme with Electricity
Management Controller in a Smart Home
Raza Abid Abbasi1, Nadeem Javaid1(B
2, Amanulla2,
Sajjad Khan1, Hafiz Muhammad Faisal1, and Sajawal Ur Rehman Khan1
1COMSATS Institute of Information Technology, Islamabad 44000, Pakistan
2Quaid-i-Azam University Islamabad, Islamabad 44000, Pakistan
Abstract. Integration of Demand Side Management (DSM) strategies
within Smart Grid (SG) helps the utilities to mange and control the
power consumer load to meet the power demand. Schemes adapted by
DSM are used for reducing the load on utilities at peak time, which
is achieved by managing the user appliances according to the changes
in load on utility and individual smart home. This work is focused on
hourly scheduling of the appliances being used in a smart home targeting
the daily electricity cost minimization. A new heuristic scheme is intro-
duced for hourly appliances scheduling on user side in this paper. The
proposed scheme works at the electricity management controller level,
installed in a smart home, within a SG infrastructure. The proposed
scheme results are compared with other heuristic schemes as well. From
extensive simulations it is depicted that proposed scheme performs best
and outperforms other schemes in term of electricity cost minimization.
Keywords: STLF ·Smart grid ·Xgboost ·Machine learning
1 Introduction
Energy consuming devices are growing exponentially, whereas energy produc-
ing sources are decreasing, i.e., oil, coal and gas. Rapidly growing at a pace of
0.3% annually, residential consumers has large share of 27% among others in US
electricity consumers [1]. Development in every field of life is cause of an evo-
lution in energy management system as well. Different type of energy resources
are explored to compete the energy demand of the community [2]. The use of
Smart Grid (SG) and its features like Demand Side Management (DSM), load
forecasting and cost forecasting is coming out to be the best substitute for the
energy management so far. SG is built as a communications network for the
energy industry. Main element of the SG is AMI which supports communication
between consumer and utility. Smart meters are used to figure out the energy
usage pattern of the consumer.
Springer Nature Switzerland AG 2020
L. Barolli et al. (Eds.): AINA 2019, AISC 926, pp. 82–94, 2020.
Minimizing Cost Using BCS 83
SG is the intelligent power grid which combines the information and commu-
nication technology in traditional grid to enable communication between pro-
ducer and consumer. SG is capable of efficient use of energy being produced.
Smart Meters (SMs) data of big volume, big velocity and big value can be used
for forecasting energy consumption and pricing. It helps the supplier to forecast
energy consumption demand and energy prices appropriately.
Power consumption prediction has importance to many applications, i.e.,
voltage and frequency adjustments, Demand Response (DR) and essential
administration. long term load forecasting (e.g., 1 year to 10 years) is essential
for outlining distribution and transmission network. Short Term Load Forecast-
ing (STLF) (e.g., from hours to days) in future is useful for DR, security affairs
and online setup of the power management system. STLF want to assure the
authenticity of the system accessories and handle power failure losses. It include
peak energy usage, daily energy load and Very STLF which forecasts using units
of less than an individual day.
In SG energy provider can handle the DR of attached devices to reduce load
on SG. DR are the modifications brought in the energy demand of the con-
sumer for the rules defined by the utility. Objective of DSM is to apply different
schemes for encouraging the consumer to transfer the energy consumption time
from on peak hours to off peak hours which increases efficiency of SG. The goal
of DSM is to minimize load on the grid by educating the users to change their
energy consumption pattern. DSM helps in decreasing peak demand, alleviat-
ing price fluctuation, and delay in system upgrade. DSM may involve the cus-
tomer to contribute in the system, where user can minimize the price of energy
through energy preservation, peak clipping and load transfer. In-fact, load shift-
ing can be considered as the mostly used technique where energy load is handled
using a DR.
2 Related Work
In SG utility is capable of managing the DR of attached loads i.e., Smart Homes
(SHs) to curtail the total pressure at the source [3].
Huangs et al. [4] suggested plan for smart buildings based on multi agent
minority game to curtail the peak load from source. It focused on reducing
power input from grid and increasing solar power usage to manage load demand.
However, it can not work if a SH does not have solar panel. In [5], Ruelens
proposed the use of batch Reinforcement Learning (RL) to DR which proved
to be more reliable for large-scale implementation as compared to traditional
model-based approaches. They proposed a model-free Monte Carlo scheme that
implements a metric depending on the state-action value function or Q-function
and implement it in finding the next day electricity required for house-hold heat
pumps by mining the historic weather information. It was modeled on a limited
set of appliances. In residential sector, energy needed for electric water heaters
was handled in [6]. It suggested conversion of electric energy into heat energy
during the absence of peak hours which will help in reducing load during peak
84 R. A. Abbasi et al.
hours. It measures temperature state of water heaters for load transferring, and
provides adjustments provision to the utility. DR management was done for a
limited set of appliances; however during conversion of energies, i.e., electric to
heat energy; storage cost of heat energy was not taken into consideration. To
provide a capable DR administration data analytic can be used wisely. In [7],
Sakurama and Miura proposed a decentralized DR management algorithm which
determines control signal by using communication network of SMs. They revealed
that the DR scheme can be more efficient when the behaviour of consumer using
energy is available.
Jindal et al. [8] proposes a comprehensive scheme based on combination of
Decision Tree (DT) and Support Vector Machine (SVM) classifier for analyzing
the data gathered from SH for theft detection in electric grid networks. It is capa-
ble of identifying theft at any level in transmission and distribution. Proposed
scheme can be considered a two-level approach for data processing and data
analysis because output produced by DR scheme is used as input to the SVM
classifier. In [9], authors used data analytics techniques on a rich, multi-year,
and high-frequency annotated data for evaluating STLF on SMs in an average
size city. They used different state-of-the-art forecasting algorithms, i.e., autore-
gressive model, kernel methods, exponential smoothing and state-space mod-
els. They found that STLF accuracy improves at larger meter aggregation, i.e.,
Feeder, substation, and system-wide level. Cumulative Sum (CUSUM) is used
in combination with SheWart algorithm to identify SMs with unusual electricity
usage pattern in [10]. Preprocessing is performed on data to find uncommon
meter readings by analyzing long missing or zero data and outliers. In first step,
temperature effect is normalized and then CUSUM and SheWart is used. Afore-
mentioned studies focused on data analytics; however, did not focus on managing
the DR of SHs for peak load curtailment. For this area, Jindal et al. [11] outline
SHs with extra electricity consumption for DR of those SHs. They SVM for clas-
sification purpose but did not consider real time modification in energy demand
of user and was applicable only to SHs exceeding specific consumption limit.
Cluster-Based Aggregate Forecasting (CBAF) algorithm was used by Misiti
et al. [12] and Alzate and Sinn [13] for load forecasting in a grid. Where Mis-
iti et al. evaluated just for industrial users where Alzate and Sinn worked for
house-hold consumers and different enterprises. They utilized wavelet included
plan and kernel spectral clustering for discovering likeness between time series.
However, they did not research the impact of various user sizes to the change
given by CBAF. In [14], Tri Kurniawan Wijaya et al., used different Machine
Learning (ML) algorithms for individual house-hold STLF and total anticipat-
ing (e.g., 1 h and 1 day in future) in private power utilization. They used scale
independent and robust accuracy metrics, i.e., normalized mean absolute error
and normalized root mean square error for accuracy measurement at house hold
level. Anish Jindal et al. [15], proposed scheme for DSM which aims at load
curtailment from grid at peak load time. They introduced different factors for
analysis, i.e., appliance priority index, appliance adjustment factor and appli-
ance curtailment priority along with an incentive scheme to increase the user
Minimizing Cost Using BCS 85
participation. Different algorithms were introduced depending on these factors,
with respect to customer perspective and utility perspective.
A STLF scheme for electricity has been proposed by Jihoon Moon et al. [16],
for higher education institutes such as universities, etc. They used 2-stage pre-
dictive analytics for improving results. In proposed scheme, electric load pattern
is found using moving average method over day of the week, where daily electric
load is predicted using random forest method. Dogan Keles et al. [17], worked
on forecasting STLF at the European power exchange next day market using
a model developed on the basis of Artificial Neural Network (ANN). Different
cluster based algorithms are used for proper selection and organization of input
data. Then best fitting ANN configurations are found and used for in-sample
and out-of-sample analysis.
SG is of great importance with respect to the pricing point of view as well.
In [18], M. Erol-Kantarci et al. suggested that due to the flexibility of DSM,
it allows consumers to take part in the process of power management system.
They can minimize their electricity cost by load shifting and conservation. It will
encourage consumer to transfer energy consumption on peak or off peak hours
depending on the variable electricity pricing. Customer wants to know the exact
price of electricity, therefore, they need price forecasting to be exact point [18].
In this exact point forecasting, they can decide to switch on or switch off a device
depending on threshold. Data analytics need to work here on big historical price
data for price classification. We can classify the price by categorizing the prices
according to different thresholds on the basis of exact point prediction scheme
values. While relating forecasting, exact price is difficult due to different variable
factors. Therefore, price taxonomy for electricity using big price data is in focus
for researchers.
For managing price classification challenge, researches used conventional clas-
sifiers, i.e., SVM, Naive Bayes, neural network and DT mostly [19]. In [20], it is
stated that SVM has better accuracy, however higher complexity. Data redun-
dancy and high dimensionality of the features are the serious threat to ML
3 Problem Statement
Replacement of traditional hierarchical grids with SG brought new challenges for
energy conservation and efficient usage. Since the evolution of DSM, it came up
with a number of challenges like load shifting, communication fairness, security
and privacy. DSM aims at handling the irregular consumption of energy on
customer side because it increases load on the utility. To fulfill the demand
during peak hours, extra electric power is generated which result in an increase
in the cost. The existing studies have already done a lot of work to reduce the
cost, Peak to Average Ratio (PAR) and discomfort by efficiently scheduling the
household appliances.
In [21], authors proposed an Integer Linear Programming (ILP) based opti-
mization mechanism for home DSM in SG. However, It works well only for large
86 R. A. Abbasi et al.
number of appliances. In [22], authors worked on DSM and introduced schedul-
ing by formulating Mixed Integer Non Linear Programming (MINLP) scheme for
mixed appliances. It helps to reduce the total electricity cost. Home Area Net-
work (HAN) controller was proposed in [23], for reducing peak loads. They used
a load reduction algorithm which is modelled using a unified modeling language
state machine model. However, cost of HAN is high. Authors in [24,25]proposed
classification model by classifying the user devices in different classes and then
scheduled them according to prioritization at real time. It has two classes where
low i.e., delay tolerant and high i.e., delay intolerant, priority classes exist.
Flexible rescheduling of appliances is required for dynamic scheduling accord-
ing to the user need. However, there is a trade-off in PAR vs cost and cost vs
comfort. Cost and PAR can be reduced by load shifting and integrating renew-
able energy resources. In this work we will use our proposed scheme, i.e., Bird
Chase Scheme (BCS); for daily electricity cost minimization.
4 Proposed System Model
Home Energy Management Systems (HEMS) are getting popular. HEMS is
based on DSM which enables efficient use of energy. In a SG infrastructure,
houses are equipped with SMs. SM enables two way communication between
utility and consumer. Devices inside a home are connected to each other and
Energy Management Controller (EMC) through HAN. Hence, smart appliances
in a house share information with EMC using HAN. EMC receives the informa-
tion sent from smart appliances and then perform required processing on that
information for taking important decisions. EMC receives the total load and
other requirement and shares that with the utility using SM. Utility receives the
information sent through SM and perform necessary actions accordingly. Utility
sends the pricing signal along with the user demanded energy to the consumer
using SM. SM shares the information received from utility with the EMC. Here,
EMC has both the pricing information received from utility and information
received from smart appliances, i.e., appliance power rating information, Length
of Operation Time (LoT) information, operational time interval information.
EMC uses the information received from utility and appliances and perform opti-
mal scheduling of appliances. It uses the Jaya, Crow Search Algorithm (CSA)
and BCS algorithms for scheduling. The objective of the scheduling is to min-
imize PAR, maximize consumer satisfaction and cost reduction. The discussed
system model is displayed in Fig. 1.
In our proposed scheme, we consider a single home, using 11 smart appli-
ances of different categories. We categorize appliances in three main categories.
Fixed appliances, shiftable appliances and interruptible appliances. Fixed appli-
ances are those that needs to be switch on according to the specified schedule.
Shiftable appliances are those, that can be scheduled but can not be interrupted
once their status is set on. Interruptible appliances can be scheduled as well as
interrupted at any time. Moreover, we have a condition that washing dryer will
always start working after washing machine has finished its operation. Table 1,
Minimizing Cost Using BCS 87
shows the categorized appliances with their corresponding LoT and power rat-
ing. The proposed schemes objective is to: minize PAR, maximize user confort
by reducing waiting time and reducing total cost. Equation1is used for total
energy consumption for a single day.
Load =
(Prate St),S
t=[1/0] (1)
Total cost for a single day is calculated using the mathematical Eq. 2.
PAR =(Max(Load)/Av g (Load)) (2)
Equation 3is used for PAR calculation.
Cost =
(Pprice Prate St),S
t=[1/0] (3)
Here, in Eq. 1,Prate is the device power rating, and Stis defining the status
i.e., on or off of a device. If status is 1 it means that device is on at that time
and if staus is 0 it means device is off at that time. In Eq. 2,PAR is peak
to average ration. PAR is computed by selecting highest load from single day
energy consumption then dividing that by the average of that day. In Eq. 3,
Pprice represents the power prices for single day.
Table 1. Control parameters
Classes Appliances LoT (h) PR (kWh)
Fixed appliances Light 12 0.1
Fan 16 0.1
Oven 9 3
Blender 41.2
Shiftable appliances Washing machine 50.5
Cloth dryer 4 4
Dish washer 41.5
Interruptable appliances AC 12 1.1
Refrigerator 12 1.2
Iron 61.1
Vacuum cleaner 50.5
5 Metaheuristic Optimization Algorithms
Conventional techniques like, LP, Convex Programming (CP), ILP, Mixed ILP
(MILP) and MI Non LP (MINLP) doesn’t perform well over large number of
88 R. A. Abbasi et al.
Fig. 1. System model
computations. Due to their deterministic nature, they are not sufficient for real
time optimization. Where, heuristic techniques are problem dependent. They
are designed to solve a particular problem. meta-heuristic algorithms can be
designed to solve almost any problem. Therefore, we selected heuristic program-
ming algorithms to solve the home appliances scheduling problem. Our main
objective is total energy consumption cost minimization. The selected meta-
heuristic algorithms are discussed now.
5.1 CSA
CSA is derived from the intelligent behavior of crows. Alireza Askarzadeh intro-
duced CSA in 2016. It is a novel meta-heuristic algorithm used for optimization.
CSA has been extensively used to solve numerous complex engineering problems.
Crows are intelligent enough to store their food for future use. While saving food
they ensure it that they are not followed by any other crow, who can steal their
food. If they find someone following them or keeping an eye on them, they change
their direction and find some other place for that. They continue this process
untill they find best possible location for hiding food. By considering locations
as possible solutions, crows as searchers, environment as the search space we
can map with the optimization process. The food source quality is considered as
the fitness function, and the best food source in the whole environment can be
mapped with the global best solution. CSA steps for scheduling appliances in a
SH for cost optimization as are listed below: 1. Initialize the problem and the
parameters 2. Initialize the memory and crow positions 3. Evaluate the positions
using objective function 4. Creating new positions for crows 5. Checking feasi-
bility of the new positions 6. Evaluate new positions using objective function 7.
Update the memory of the crows 8. comparing termination criteria.
Minimizing Cost Using BCS 89
5.2 Jaya
Jaya is a Sanskrit word, which means victory. Jaya algorithm aims to become
winner by approaching best solution. This algorithm always try to get closer to
the best solution and get away from the worst solution. In first step we initial-
ize population size which is number of possible schedulings, number of design
variables which is number of appliances in our case and termination criterion. In
second step it identifies the best and worst solutions in the population. In third
step it updates population solutions based on previously selected best and worst
solutions. In fourth step we compare the updated solution with the best solution
corresponding to the design variable. If new solution is better we replace the
previous one, otherwise keep the same. If termination criterion is met, we report
the optimum solution otherwise repeat steps one to five.
5.3 BCS
BCS is based on the birds intelligence behavior to secure their food from other
birds in the flock. We can consider locations as possible solutions, birds as
searchers, environment as the search space. We can map them with the opti-
mization process. The food source quality is considered as the fitness function,
and the best food source in the whole environment can be mapped with the
global best solution. The steps followed in BICS are listed below.
Algorithm 1. Proposed: Bird Chase Scheme
Require: Minimum cost Scheduling
Initialize the bird awareness probability, memory and population
while Termination criteria is not met do
for i=1 to size(p) do
Apply Rosenbrock fitness function on p(i)
end for
generate random number of birds
for j=1 to size(p) do
update bird position using awareness probability
end for
Apply mutate on updated population
for i=1 to size(p) do
Apply Rosenbrock fitness function on updated p(i)
end for
Apply mutate on updated population
Find local fit solution
Apply scheduling conditions
end while
Find global fit solution
scheduling Appliances
90 R. A. Abbasi et al.
6 Simulation Results and Discussion
This section highlights the proposed system performance, which is validated
through extensive simulations in MATLAB software. Then, the results achieved
are discussed hereafter. The results are produced for Real Time Pricing (RTP)
tariff using our proposed algorithm BCS. We compare the results of BCS with
other implemented techniques such as Jaya and CSA. We study a single home for
the purpose of simulation with eleven appliances, i.e., D = 11 those needs to be
scheduled. For the purpose of simulation, these appliances are categorized into
three different groups: Fixed Appliances (FA), Shift-able Appliances (SA), Inter-
ruptible Appliances (IA). We selected these appliances because of their frequent
use in summer. Group FA includes light, fan, oven and blender. These appliances
will can neither be interrupted nor shifted. Group SA includes washing machine,
cloth dryer and dish washer. These appliances can be shifted, however can not
be interrupted once they are turned on. Group IA includes AC, refrigerator, iron
and vaccume cleaner. These appliances can be shifted as well as interrupted.
Real Time Pricing (RTP), pricing scheme is used for load, cost and PAR
calculation. Figure 2is displaying the RTP pricing signal for 24 h. Where, x-axis
represents the 24 h and y-axis represents price in cents per kilowatt-hour.
6.1 RTP Tariff
As mentioned earlier, proposed scheme will be evaluated using RTP. Here, we
will discuss the RTP tariff. The RTP is based on the per hour use of electricity
and is known as dynamic price rate as well. RTP provides information about
real cost of energy at a specific time to the consumer. Electric prices vary with
time. RTP enables the user to adjust their working hours from on peak hours
to off peak hours, resulting in saving. RTP is enabled using SM that allows the
exchange of information between utility and consumer.
1 2 3 4 5 6 7 8 9 101112131415161718192021222324
Time (hour)
Price (cent/kWh)
Fig. 2. Pricing signal
Figure 3is displaying the power consumption pattern for unscheduled scheme
and scheduled schemes i.e., CSA, Jaya and BCS. X-axis is showing the time in
hour and y-axis is showing load per hour in kWh. It is note-able that all three
scheduling schemes are performing better load distribution as compared to non
scheduled. It is evident from the figure that proposed BCS is scheduling the
Minimizing Cost Using BCS 91
1 2 3 4 5 6 7 8 9 10111213141 5161 7181 9202 1222 324
Time (hour)
Per hour load (kWh)
Fig. 3. Per hour load
1 2 3 4 5 6 7 8 9 10111213141 5161 7181 9202 1222324
Time (hour)
Per hour cost (cent)
Fig. 4. Hourly cost
energy consumption in an proficient way. While using BCS load during the last
four hours is maximum as a result of reducing load during peak hours.
Figure 4shows the hourly cost for Jaya, CSA and our proposed scheme BCS.
It is visible from plot that BCS is performing better than competitors while
shifting the load from on peak hours to off peak hours. Moreover, we found that
even Jaya and CSA are performing better than unscheduled.
The PAR performance of Jaya, CSA and BCS is displayed in Fig. 5. X-axis
is showing bars for Unscheduled, Jaya, CSA and BCS where y-axis is showing
the value of PAR. The graph is depicting that the performance of our proposed
scheme BCS is not optimal in case of Peak to Average Ratio. Here, CSA outper-
formed all other schemes. The PAR is reduced to 53.64%, 52.51% and 39.64%
by CSA, Jaya and BCS respectively. Here, CSA outperformed all other schemes.
It is worth mentioning that all schemes are performing better than unscheduled.
Waiting time of a consumer is displayed in Fig. 6. The consumer waiting time
is actually depicting the user satisfaction level. In current scheme, user comfort
is measured in terms of time a user will wait for a particular appliance to turn
on. The wait is discomfort in real. If waiting time is high, it means discomfort is
high. So, the waiting time and user comfort are indirectly proportional to each
other. The calculated waiting time is 5.267235 h, 5.264078 h and 5.953346 h for
Jaya, CSA and BCS respectively. It is evident that CSA outperformed the other
techniques in the case of user comfort.
92 R. A. Abbasi et al.
Unscheduled Jaya CSA BCS
Fig. 5. Peak to average ratio
Waiting time (hour)
Fig. 6. Waiting time
Unscheduled Jaya CSA BCS
Total cost (cent)
Fig. 7. Total cost
In Fig. 7, total electricity cost for 24h is displayed. The plot is displaying the
cost calculated by the use of different schemes. The unscheduled cost is 1410.98
cents, where, Jaya, CSA and BCS costs 1137.7832 cents, 1157.3978 cents and
1108.6057 cents respectively. The cost is reduced by all schemes as compared
to the unscheduled. However, The calculated values are depicting that proposed
scheme BCS outperformed all other techniques. Electricity cost is reduced by
19.94%, 18.95% and 24.58% from Jaya, CSA and JSCA techniques respectively
as compared to the unscheduled. From graphs we can see that their is a trade-off
between PAR, user comfort, i.e., waiting time and cost. The proposed scheme
minimizes the cost but is compromising PAR and user comfort.
Minimizing Cost Using BCS 93
7 Conclusion
In this proposed solution we focused on DSM for minimizing the total electricity
cost. A new load shifting strategy for appliance scheduling is proposed, that
works at EMC level in a SG. It works on scheduling different appliances used
in houses, considering their usage patterns. We scheduled appliances for a single
day. We used CSA, Jaya and our proposed BCS for scheduling appliances. By
comparing the results of different schemes used for simulations, it is evident, that
BCS performed best as compared to CSA and Jaya in terms of total electricity
cost. Electricity cost is reduced by 24.58% using BCS. However, there is a trade-
off between different parameters, like PAR, consumer waiting time and cost.
1. U.S. Energy Information Administration, Annual Energy Outlook 2018, pp. 121,
122, Residential and commercial energy consumption grows gradually from 2017 to
2050. Accessed Oct 2018
2. Luo, F., Ranzi, G., Wan, C., Xu, Z., Dong, Z.Y.: A multi-stage home energy
management system with residential photovoltaic penetration. IEEE Trans. Ind.
Inform. 15(1), 116–126 (2018)
3. Gungor, V.C., Sahin, D., Kocak, T., Ergut, S., Buccella, C., Cecati, C., Hancke,
G.P.: Smart grid and smart homes: key players and pilot projects. IEEE Ind.
Electron. Mag. 6(4), 18–34 (2012)
4. Huang, H., Cai, Y., Xu, H., Yu, H.: A multi-agent minority-game- based demand-
response management of smart buildings towards peak load reduction. IEEE Trans.
Comput.-Aided Des. Integr. Circuits Syst. 36(4), 573–585 (2017)
5. Ruelens, F., Claessens, B.J., Vandael, S., De Schutter, B., Babuˇska, R., Belmans,
R.: Residential demand response of thermostatically controlled loads using batch
reinforcement learning. IEEE Trans. Smart Grid 8(5), 2149–2159 (2017)
6. Pourmousavi, S.A., Patrick, S.N., Nehrir, M.H.: Real-time demand response
through aggregate electric water heaters for load shifting and balancing wind gen-
eration. IEEE Trans. Smart Grid 5(2), 769–778 (2014)
7. Sakurama, K., Miura, M.: Communication-based decentralized demand response
for smart microgrids. IEEE Trans. Ind. Electron. 64(6), 5192–5202 (2017)
8. Jindal, A., Dua, A., Kaur, K., Singh, M., Kumar, N., Mishra, S.: Decision tree
and SVM-based data analytics for theft detection in smart grid. IEEE Trans. Ind.
Informat. 12(3), 1005–1016 (2016)
9. Mirowski, P., Chen, S., Kam Ho, T., Yu, C.N.: Demand forecasting in smart grids.
Bell Labs Tech. J. 18(4), 135–158 (2014)
10. Wu, Z., Zhao, T., He, L., Shen, X.: Smart grid meter analytics for revenue pro-
tection. In: Proceedings of the IEEE International Conference on Power System
Technology, pp. 782–787 (2014)
11. Jindal, A., Kumar, N., Singh, M.: A data analytical approach using support vector
machine for demand response management in smart grid. In: 2016 IEEE Power
and Energy Society General Meeting (PESGM) 17–21 July, pp. 1–5 (2016)
12. Misiti, M., Misiti, Y., Oppenheim, G., Poggi, J.-M.: Optimized clusters for disag-
gregated electricity load forecasting. REVSTAT 8, 105–124 (2010)
94 R. A. Abbasi et al.
13. Alzate, C., Sinn, M.: Improved electricity load forecasting via kernel spectral clus-
tering of smart meters. In IEEE 13th International Conference on Data Mining
(ICDM), pp. 943–948 (2013)
14. Wijaya, T.K., Vasirani, M., Humeau, S., Aberer, K.: Cluster-based aggregate fore-
casting for residential electricity demand using smart meter data. In: IEEE Inter-
national Conference on Big Data (Big Data)
15. Jindal, A., Singh, M., Kumar, N.: Consumption-aware data analytical demand
response scheme for peak load reduction in smart grid. IEEE Trans. Ind. Electron.
65(11), 8993–9004 (2018)
16. Moon, J., Kim, K.-H., Kim, Y., Hwang, E.: A short-term electric load forecasting
scheme using 2-stage predictive analytics. In: 2018 IEEE International Conference
on Big Data and Smart Computing (2018)
17. Keles, D., Scelle, J., Paraschiv, F., Fichtner, W.: Extended forecast methods for
day-ahead electricity spot prices applying artificial neural networks. Appl. Energy
162, 218–230 (2016)
18. Erol-Kantarci, M., Mouftah, H.T.: Energy-efficient information and communica-
tion infrastructures in the smart grid: a survey on interactions and open issues.
IEEE Commun. Surv. Tutor. 17(1), 179–197 (2015)
19. Huang, D., Zareipour, H., Rosehart, W.D., Amjady, N.: Data mining for electricity
price classification and the application to demand-side management. IEEE Trans.
Smart Grid 3(2), 808–817 (2012)
20. Fleury, A., Vacher, M., Noury, N.: SVM-based multimodal classification of activ-
ities of daily living in health smart homes: sensors, algorithms, and first experi-
mental results. IEEE Trans. Inform. Technol. Biomed. 14(2), 274–283 (2010)
21. Zhu, Z., Tang, J., Lambotharan, S., Chin, W.H., Fan, Z.: An integer linear pro-
gramming based optimization for home demand-side management in smart grid.
In: 2012 IEEE PES Innovative Smart Grid Technologies (ISGT) (2012)
22. Maa, K., Hua, S., Yanga, J., Xua, X., Guanb, X.: Appliances scheduling via cooper-
ative multi-swarm PSO underday-ahead prices and photovoltaic generation. Appl.
Soft Comput. 62, 504–513 (2018)
23. Bazydlo, G., Wermi´nski, S.: Demand side management through home area network
systems. Electr. Power Eng. Syst. 97, 174–185 (2018)
24. Hong, S.H., Yu, M., Huang, X.: A real-time demand response algorithm for het-
erogeneous devices in buildings and homes. Energy 80, 123–132 (2015)
25. Logenthiran, T., Srinivasan, D., Shun, T.Z.: Demand side management in smart
grid using heuristic optimization. IEEE Trans. Smart Grid 3, 1244–1252 (2012)
Development of models based on the Internet of Things (IoT) for household framework leads to the establishment of smart appliances more and more for improving the living style and support of residents. Due to this reason, useful energy becomes an increase in demand for the past few decades that especially usages in smart homes and buildings as people of developing rapidly and enhancing their lifestyle based on modern technology. Various parameters like building characteristics, surrounding weather variables, and energy usage pattern are the reliable sources of buildings energy performance. In this paper, a predictive model is proposed by integrating the mechanisms of IoT and classifier ensemble techniques for forecasting the indoor temperature of the smart building. The online learning-methodology is used for training the predictive model for a successive performance over an unfamiliar dataset. Moreover, the recorded real-sensor data is applied in the experimental process, which is based on the classifier ensemble techniques for validating the model. Furthermore, the building works with an energy-efficient way by using the latest IoT architecture, which is based on Edge Computing. The simulation results compared with other existing approaches and models in which the proposed energy prediction techniques prove to be better.
Full-text available
This paper presents a cyber-physical management of smart buildings based on smart-gateway network with distributed and real-time energy data collection and analytics. We consider a building with multiple rooms supplied with one main electricity grid and one additional solar energy grid. Based on smart-gateway network, energy signatures of rooms are first extracted with consideration of uncertainty and further classified as different types of agents. Then, a multiagent minority-game (MG)-based demand-response management is introduced to reduce peak demand on the main electricity grid and also to fairly allocate solar energy on the additional grid. Experiment results show that compared to the traditional static and centralized energy-management system (EMS), and the recent multiagent EMS using price-demand competition, the proposed uncertainty-aware MG-EMS can achieve up to 50× and 145× utilization rate improvements, respectively, regarding to the fairness of solar energy resource allocation. More importantly, the peak load from the main electricity grid is reduced by 38.50% in summer and 15.83% in winter based on benchmarked energy data of building. Lastly, an average 23% uncertainty can be reduced with an according 37% balanced energy allocation improved comparing to the MG-EMS without consideration of uncertainty.
Advances in bilateral communication technology foster the improvement and development of Home Energy Management System (HEMS). This paper proposes a new HEMS to optimally schedule Home Energy Resources (HER) in a high rooftop photovoltaic penetrated environment. The proposed HEMS includes three stages: forecasting, day-ahead scheduling, and actual operation. In the forecasting stage, short-term forecasting is performed to generate day-ahead forecasted photovoltaic solar power and home load profiles; in the day-ahead scheduling stage, a Peak-to-Average Ratio constrained coordinated HER scheduling model is proposed to minimize the 1-day home operation cost; in the actual operation stage, a model predictive control based operational strategy is proposed to correct HER operations with the update of real-time information, so as to minimize the deviation of actual and day-ahead scheduled net-power consumption of the house. An adaptive thermal comfort model is applied in the proposed HEMS to provide decision-support on the scheduling of the heating, ventilating, and air conditioning (HVAC) system of the house. The proposed approach is then validated based on Australian real datasets.
The ever-increasing load demand of the residential sector gives rise to concerns such as-decreased quality of service and increased demand-supply gap in the electricity market. To tackle these concerns, the utilities are switching to smart grids (SGs) to manage the demand response (DR) of the connected loads. However, most of the existing DR management schemes have not explored the concept of data analytics for reducing peak load while taking consumer constraints into account. To address this issue, a novel data analytical demand response (DADR) management scheme for residential load is proposed in this paper with an aim to reduce the peak load demand. The proposed scheme is primarily based on the analysis of consumers' consumption data gathered from smart homes (SHs) for which factors such as-appliance adjustment factor, appliance priority index, etc. have been considered. Based on these factors, different algorithms with respect to consumer's and utility's perspective have been proposed to take DR. In addition to it, an incentive scheme has also been presented to increase the consumers' participation in the proposed scheme. The results obtained show that it efficiently reduces the peak load at the grid by a great extent. Moreover, it also increases the savings of the consumers by reducing their overall electricity bills.
The increasing demand for power in the Electrical Power System (EPS) causes a significant increase of power in the daily load curve as well as transmission line overload. The large variability in energy consumption in the EPS combined with unpredictable weather events can lead to a situation in which, to save the stability of the EPS, power limits must be introduced or even industrial customers in a given area have to be disconnected, which causes financial losses. Nowadays, a Transmission System Operator (TSO) is looking for additional solutions to reduce peak power, because existing approaches (mainly building new intervention power units or tariff programs) are not satisfactory due to the high cost of services in combination with an insufficient power reduction effect. The paper presents an approach to reduce peak loads with the use of Home Area Network (HAN) systems installed at residential units. The algorithm of the HAN system, executed by the HAN controller, is modeled using Unified Modeling Language (UML). Then using model transformation techniques, the UML model is translated into Verilog description, and is finally implemented in the Field Programmable Gate Array (FPGA). The advantages of the proposed approach are that with only a small loss of residential user comfort, there is a gain in energy reduction for a relatively small cost, an effective and convenient design of the HAN algorithm, and the flexible maintenance of HAN systems. The latter gain is possible thanks to using modern FPGAs, which allow for dynamic reconfiguration of the HAN controller. It means that a HAN algorithm of a selected user can be exchanged without power interruption of other residential users. A practical example illustrating the proposed approach and a calculation of the potential gains from its implementation are also presented.
This work studies the problem of appliances scheduling in a residential unit. An appliance-scheduling model for the home energy management system (HEMS) is established based on day-ahead electricity prices and photovoltaic (PV) generation. The HEMS receives the meter data and calculates the scheduling strategies, then the HEMS sends control signals to achieve the on/off control of the appliances through the ZigBee (a wireless communication technology with low power consumption in short distance). The study starts with a view to minimizing the summation of the electricity payments, the consumer's dissatisfaction (DS), and the carbon dioxide emissions (CDE), and the constraints specify the restrictions on the operating time and the power consumption of the appliances. A cooperative multi-swarm particle swarm optimization (PSO) algorithm is adopted to solve the combinational optimization problem. The appliances can be categorized into shiftable and non-shiftable appliances. For the shiftable appliances, the start time and power of the appliances can be scheduled flexibly in the case of the announced electricity prices. Furthermore, the plug-in hybrid electric vehicle (PHEV) is introduced to charge or discharge for energy management. Specially, the ability of selling electricity (SE) to the power grid is studied for appliances scheduling. Finally, the simulation results demonstrate that the cooperative multi-swarm PSO algorithm shows good convergence performance under different scenarios. Moreover, The electricity payments can be reduced by considering the carbon dioxide emissions in the objective function and selling electricity to the power grid, which also achieves the peak load curtailment.
Demand response (DR) is one of the most promising solutions to efficient control of smart grids with renewable energy resources. Usually, DR programs are im- plemented by means of centralized control by power supply companies or independent system operators (ISOs). In con- trast, recently, the focus has been on decentralized control to enhance the efficient use of distributed energy resources especially on microgrids. This paper proposes a decen- tralized control system for DR. The key of the proposed method is a new decentralized algorithm for determining appropriate control signals (corresponding to prices and/or incentives) by using communication networks provided by smart meters. The effectiveness of the proposed method is illustrated by a numerical example.
Conference Paper
Many existing issues pertaining to power sector such as-demand response management, theft detection, outage management etc. can be solved efficiently with grid modernization. Out of these, demand response is one such issue which affects the overall grid stability. One way of managing demand response is to balance the load in smart grid (SG). In this paper, a novel scheme for handling the demand response in SG is presented. The household load is managed in such a way that the load profile of the SG is flattened. Unlike existing approaches, the proposed scheme is based on the data analytics and works as follows. Initially, the data is gathered from all the devices and the users with excess load consumption are identified using the support vector machine (SVM). To curtail the load of such users, a novel load balancing algorithm has been designed. This algorithm sheds the excess load in homes so as to balance the overall load in SG. The simulation results show that the proposed scheme effectively flattens the load profile of SG while managing the demand response.
Conference Paper
While electricity demand forecasting literature has focused on large, industrial, and national demand, this paper focuses on short-term (1 and 24 hour ahead) electricity demand forecasting for residential customers at the individual and aggregate level. Since electricity consumption behavior may vary between households, we first build a feature universe, and then apply Correlation-based Feature Selection to select features relevant to each household. Additionally, smart meter data can be used to obtain aggregate forecasts with higher accuracy using the so-called Cluster-based Aggregate Forecasting (CBAF) strategy, i.e., by first clustering the households, forecasting the clusters' energy consumption separately, and finally aggregating the forecasts. We found that the improvement provided by CBAF depends not only on the number of clusters, but also more importantly on the size of the customer base.
Nontechnical losses, particularly due to electrical theft, have been a major concern in power system industries for a long time. Large-scale consumption of electricity in a fraudulent manner may imbalance the demand-supply gap to a great extent. Thus, there arises the need to develop a scheme that can detect these thefts precisely in the complex power networks. So, keeping focus on these points, this paper proposes a comprehensive top-down scheme based on decision tree (DT) and support vector machine (SVM). Unlike existing schemes, the proposed scheme is capable enough to precisely detect and locate real-time electricity theft at every level in power transmission and distribution (T&D). The proposed scheme is based on the combination of DT and SVM classifiers for rigorous analysis of gathered electricity consumption data. In other words, the proposed scheme can be viewed as a two-level data processing and analysis approach, since the data processed by DT are fed as an input to the SVM classifier. Furthermore, the obtained results indicate that the proposed scheme reduces false positives to a great extent and is practical enough to be implemented in real-time scenarios.