No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
A novel method based on extreme learning machine to predict heating and cooling load through design and structural attributes
Abstract
In the present day environment, smart buildings require optimization of energy consumption through monitoring, consumption prediction and making policy decisions accordingly. Attributes related to building design and structure play a vital role in heating load(HL) and cooling load(CL) of the building which directly affects the energy performance of the buildings. For prediction of HL and CL, emerging machine learning approaches can help in improving accuracy and efficiency in real time. This paper provides improvements in energy load assessment of the buildings. It is the first is the in-depth study and analysis of design and structural attributes and their correlation with HL and CL, the novel methods based on ELM and its variants online sequential ELM(OSELM) to predict HL and CL. This study also proposes OS-ELM based online/real-time prediction when data is coming in stream The total 24 models have been developed including 12 models based on ELM and 12 models based on OSELM with different feature sets and activation functions. Models have been compared on the basis of accuracy, computational performance and efficiency with few existing models. The experimental results show that the proposed models learn better and outperform other popular machine learning approaches such as the artificial neural net-work(ANNs), support vector machine(SVM), radial basis function network(RBFN), random forest(RF) and existing work in the energy and building domain.
... Their physical characteristics -including geometry, height, function, material, condition, and style -are the key parameters that not only support sustainable urban development but also reflect economic progress and cultural evolution over time . Such rich building-level data has been instrumental in a range of applications, such as urban climate simulations for improved environmental planning (Creutzig et al., 2019), building energy modeling for resource optimization (Kumar et al., 2018;Roth et al., 2020), estimation of urban material stocks for the circular economy (Raghu et al., 2023), and disaster impact assessments to inform effective response and recovery efforts (Westrope et al., 2014). Moreover, these data support more nuanced analyses of population distributions , socio-economic conditions (Feldmeyer et al., 2020), as well as deeper understanding of the impact on human behaviors (Wang et al., 2016) and public perception . ...
Building properties, such as height, usage, and material composition, play a crucial role in spatial data infrastructures, supporting applications such as energy simulation, risk assessment, and environmental modeling. Despite their importance, comprehensive and high-quality building attribute data remain scarce in many urban areas. Recent advances have enabled the extraction and tagging of objective building attributes using remote sensing and street-level imagery. However, establishing a method and pipeline that integrates diverse open datasets, acquires holistic building imagery at scale, and infers comprehensive building attributes remains a significant challenge. Among the first, this study bridges the gaps by introducing OpenFACADES, an open framework that leverages multimodal crowdsourced data to enrich building profiles with both objective attributes and semantic descriptors through multimodal large language models. Our methodology proceeds in three major steps. First, we integrate street-level image metadata from Mapillary with OpenStreetMap geometries via isovist analysis, effectively identifying images that provide suitable vantage points for observing target buildings. Second, we automate the detection of building facades in panoramic imagery and tailor a reprojection approach to convert objects into holistic perspective views that approximate real-world observation. Third, we introduce an innovative approach that harnesses and systematically investigates the capabilities of open-source large vision-language models (VLMs) for multi-attribute prediction and open-vocabulary captioning in building-level analytics, leveraging a globally sourced dataset of 30,180 labeled images from seven cities. Evaluation shows that fine-tuned VLM excel in multi-attribute inference, outperforming single-attribute computer vision models and zero-shot ChatGPT-4o.
Instagram is one of the most popular and widely used social network platforms. It is used as a digital tool to connect with other users and also to share information and influence them for marketing and advertising purposes. The influence of popular users is broadly determined by post’s engagement rate in terms of likes, comments, and shares, and the number of followers as well. An objective and comprehensive measure of popularity is necessary to understand the factors that will help make an influencer marketing campaign more successful and beneficial for business activities. This research work attempts to take various features of an influencer account and Instagram posts dataset and develop a novel model that accurately quantifies and determines the influence of a user on Instagram. The research is based on datasets of top regional Instagram influencers and their posts based on categories signified through hashtags and captions. Our research attempts to develop a model using principal component analysis to quantify influence and using it to rank influencers. In our experiment, the proposed model after experimentation, gave the Instagram username “iqbaal.e” influence score as 874,712.9526, username “1nctdream” as 753,830.5847 and username “weareone. exo” as 668,054.4360. The proposed model ranks were compared with other ranks for Instagram users based on other measures such as follower rank etc. User names “huyitian”, “bintangemon” and “bimopd” are top social media influencers based on the proposed model for better business advertising and digital marketing outcomes with collected data and experiment context. This proposed approach gives an exploration for the stakeholders to quantify the impact of influencer in social media and demonstrate an innovative approach.
This paper proposes a computationally efficient method for traffic sign recognition (TSR). This proposed method consists of two modules: 1) extraction of histogram of oriented gradient variant (HOGv) feature and 2) a single classifier trained by extreme learning machine (ELM) algorithm. The presented HOGv feature keeps a good balance between redundancy and local details such that it can represent distinctive shapes better. The classifier is a single-hidden-layer feedforward network. Based on ELM algorithm, the connection between input and hidden layers realizes the random feature mapping while only the weights between hidden and output layers are trained. As a result, layer-by-layer tuning is not required. Meanwhile, the norm of output weights is included in the cost function. Therefore, the ELM-based classifier can achieve an optimal and generalized solution for multiclass TSR. Furthermore, it can balance the recognition accuracy and computational cost. Three datasets, including the German TSR benchmark dataset, the Belgium traffic sign classification dataset and the revised mapping and assessing the state of traffic infrastructure (revised MASTIF) dataset, are used to evaluate this proposed method. Experimental results have shown that this proposed method obtains not only high recognition accuracy but also extremely high computational efficiency in both training and recognition processes in these three datasets.
Building energy use associated with non-domestic buildings accounts for approximately 19% of the total UK CO2 emissions. Energy consumption in a non-domestic building is a complex issue due to a wide variety of uses and energy services and therefore the energy demand of individual buildings need to be understood. A pilot study was undertaken to analyse the relationship between the electrical energy demand profiles and user activities for a university building. To gain insight into how the building is used, operated and managed on a daily basis, an online questionnaire was distributed to staff and students as well as interviews conducted with key management personnel. Analysis was performed on the half-hourly electrical demand data for the case-study building to identify key trends and patterns in energy use. The shape and magnitude of energy demand profiles shows a significant trend which does not seem to be strongly connected to occupancy patterns. It was found that the building was mostly controlled by a building management system (BMS) where building users have minimal access to the controls. However, it was interesting to find that the detailed information on the occupancy patterns could help the management team to redesign control strategies for optimum energy performance of the building.
Extreme learning machine (ELM) has gained increasing interest from various research fields recently. In this review, we aim to report the current state of the theoretical research and practical advances on this subject. We first give an overview of ELM from the theoretical perspective, including the interpolation theory, universal approximation capability, and generalization ability. Then we focus on the various improvements made to ELM which further improve its stability, sparsity and accuracy under general or specific conditions. Apart from classification and regression, ELM has recently been extended for clustering, feature selection, representational learning and many other learning tasks. These newly emerging algorithms greatly expand the applications of ELM. From implementation aspect, hardware implementation and parallel computation techniques have substantially sped up the training of ELM, making it feasible for big data processing and real-time reasoning. Due to its remarkable efficiency, simplicity, and impressive generalization performance, ELM have been applied in a variety of domains, such as biomedical engineering, computer vision, system identification, control and robotics, etc. In this review, we try to provide a comprehensive view of these advances in ELM together with its future perspectives.
The present research work concerns development of regression models to predict the monthly heating demand for single-family residential sector in temperate climates, with the aim to be used by architects or design engineers as support tools in the very first stage of their projects in finding efficiently energetic solutions. Another interest to use such simplified models is to make it possible a very quick parametric study in order to optimize the building structure versus environmental or economic criteria. All the energy prediction models were based on an extended database obtained by dynamic simulations for 16 major cities of France. The inputs for the regression models are the building shape factor, the building envelope U-value, the window to floor area ratio, the building time constant and the climate which is defined as function of the sol-air temperature and heating set-point. If the neural network (NN) methods could give precise representations in predicting energy use, with the advantage that they are capable of adjusting themselves to unexpected pattern changes in the incoming data, the multiple regression analysis was also found to be an efficient method, nevertheless with the requirement that an extended database should be used for the regression. The validation is probably the most important level when trying to find prediction models, so 270 different scenarios are analysed in this research work for different inputs of the models. It has been established that the energy equations obtained can do predictions quite well, a maximum deviation between the predicted and the simulated is noticed to be 5.1% for Nice climate, with an average error of 2%. In this paper, we also show that is possible to predict the building heating demand even for more complex scenarios, when the construction is adjacent to non-heated spaces, basements or roof attics.
For the past 50 years, a wide variety of building energy simulation programs have been developed, enhanced and are in use throughout the building energy community. This paper is an overview of a report, which provides up-to-date comparison of the features and capabilities of twenty major building energy simulation programs. The comparison is based on information provided by the program developers in the following categories: general modeling features; zone loads; building envelope and daylighting and solar; infiltration, ventilation and multizone airflow; renewable energy systems; electrical systems and equipment; HVAC systems; HVAC equipment; environmental emissions; economic evaluation; climate data availability, results reporting; validation; and user interface, links to other programs, and availability.
According to conventional neural network theories, single-hidden-layer feedforward networks (SLFNs) with additive or radial basis function (RBF) hidden nodes are universal approximators when all the parameters of the networks are allowed adjustable. However, as observed in most neural network implementations, tuning all the parameters of the networks may cause learning complicated and inefficient, and it may be difficult to train networks with nondifferential activation functions such as threshold networks. Unlike conventional neural network theories, this paper proves in an incremental constructive method that in order to let SLFNs work as universal approximators, one may simply randomly choose hidden nodes and then only need to adjust the output weights linking the hidden layer and the output layer. In such SLFNs implementations, the activation functions for additive nodes can be any bounded nonconstant piecewise continuous functions g : R --> R and the activation functions for RBF nodes can be any integrable piecewise continuous functions g : R --> R and integral of R g(x)dx not equal to 0. The proposed incremental method is efficient not only for SFLNs with continuous (including nondifferentiable) activation functions but also for SLFNs with piecewise continuous (such as threshold) activation functions. Compared to other popular methods such a new network is fully automatic and users need not intervene the learning process by manually tuning control parameters.
We present a nonlinear version of the recursive least squares (RLS) algorithm. Our algorithm performs linear regression in a high-dimensional feature space induced by a Mercer kernel and can therefore be used to recursively construct minimum mean-squared-error solutions to nonlinear least-squares problems that are frequently encountered in signal processing applications. In order to regularize solutions and keep the complexity of the algorithm bounded, we use a sequential sparsification process that admits into the kernel representation a new input sample only if its feature space image cannot be sufficiently well approximated by combining the images of previously admitted samples. This sparsification procedure allows the algorithm to operate online, often in real time. We analyze the behavior of the algorithm, compare its scaling properties to those of support vector machines, and demonstrate its utility in solving two signal processing problems-time-series prediction and channel equalization.
It is clear that the learning speed of feedforward neural networks is in general far slower than required and it has been a major bottleneck in their applications for past decades. Two key reasons behind may be: (1) the slow gradient-based learning algorithms are extensively used to train neural networks, and (2) all the parameters of the networks are tuned iteratively by using such learning algorithms. Unlike these conventional implementations, this paper proposes a new learning algorithm called extreme learning machine (ELM) for single-hidden layer feedforward neural networks (SLFNs) which randomly chooses hidden nodes and analytically determines the output weights of SLFNs. In theory, this algorithm tends to provide good generalization performance at extremely fast learning speed. The experimental results based on a few artificial and real benchmark function approximation and classification problems including very large complex applications show that the new algorithm can produce good generalization performance in most cases and can learn thousands of times faster than conventional popular learning algorithms for feedforward neural networks.1
Occupancy detection is very interesting research problem which may help in understanding ambient dynamics of the environment, resource utilisation, energy conservation and consumption, electricity usages and patterns, security and privacy related aspects. In addition to this, achieving good accuracy for occupancy detection problem in the home and commercial buildings can help in cost reduction substantially. In this paper, we explain one experiment in which data for occupancy and ambient attributes have been collected. This paper develops machine learning-based intelligent occupancy detection model and compare the results with several machine learning techniques in a detailed manner.
Ensemble weather predictions are introduced in the operation of district heating systems to create a heat load forecast with dynamic uncertainties. These provide a new and valuable tool for time-dependent risk assessment related to e.g. security of supply and the energy markets. As such, it is useful in both the production planning and the online operation of a modern district heating system, in particular in light of the low-temperature operation, integration of renewable energy and close interaction with the electricity markets. In this paper, a simple autoregressive forecast model with weather prediction input is used to showcase the new concept. On the study period, its performance is comparable to more complex forecast models. The total uncertainty of the heat load forecast is divided into a constant model uncertainty plus a time-dependent weather-based uncertainty. The latter varies by as much as a factor of 18 depending on the ensemble spread. As a consequence, the total forecast uncertainty varies significantly. The forecast model is applied to the operation of three heat exchanger stations. Applying an optimized temperature control can significantly lower supply temperatures compared to current operation. Improving the temperature control with dynamic time-dependent weather-based uncertainties can lower the supply temperature further and reduce heat losses to the ground. The potential benefit of using dynamic uncertainties is larger for systems with relatively smaller pumping capacities.
At present, most of the buildings are using process of heating, ventilation and air conditioning (HVAC)-systems. HVAC systems are also responsible for consumption of huge amount of energy. Home automation techniques are being used to reduce the waste of resources especially energy that is available to us in the form of temperature, electricity, water, sunlight, etc. Forecasting and predicting the future demand of the energy can help us to maintain and to reduce the cost of energy in the buildings. In this paper, we use the experiments the small medium large system (SML system) which is the house built at the university of CEU cardinal Herrera (CEU-UCH) for competition named Solar Decathlon 2013. With the data available from this experiments, we try to predict and forecast the future temperature condition intelligently for energy conservation with the model based on Extreme Learning Machine (ELM). This will help in determining the energy needs of the buildings and further will help in efficient utilization and conservation of energy.
Extreme learning machines (ELMs) basically give answers to two fundamental learning problems: (1) Can fundamentals of learning (i.e., feature learning, clustering, regression and classification) be made without tuning hidden neurons (including biological neurons) even when the output shapes and function modeling of these neurons are unknown? (2) Does there exist unified framework for feedforward neural networks and feature space methods? ELMs that have built some tangible links between machine learning techniques and biological learning mechanisms have recently attracted increasing attention of researchers in widespread research areas. This paper provides an insight into ELMs in three aspects, viz: random neurons, random features and kernels. This paper also shows that in theory ELMs (with the same kernels) tend to outperform support vector machine and its variants in both regression and classification applications with much easier implementation.
Ship detection on spaceborne images has attracted great interest in the applications of maritime security and traffic control. Optical images stand out from other remote sensing images in object detection due to their higher resolution and more visualized contents. However, most of the popular techniques for ship detection from optical spaceborne images have two shortcomings: 1) Compared with infrared and synthetic aperture radar images, their results are affected by weather conditions, like clouds and ocean waves, and 2) the higher resolution results in larger data volume, which makes processing more difficult. Most of the previous works mainly focus on solving the first problem by improving segmentation or classification with complicated algorithms. These methods face difficulty in efficiently balancing performance and complexity. In this paper, we propose a ship detection approach to solving the aforementioned two issues using wavelet coefficients extracted from JPEG2000 compressed domain combined with deep neural network (DNN) and extreme learning machine (ELM). Compressed domain is adopted for fast ship candidate extraction, DNN is exploited for high-level feature representation and classification, and ELM is used for efficient feature pooling and decision making. Extensive experiments demonstrate that, in comparison with the existing relevant state-of-the-art approaches, the proposed method requires less detection time and achieves higher detection accuracy.
This paper described the development of regression models which are able to predict office building annual heating, cooling and auxiliary energy requirements for different HVAC systems as a function of office building heating and cooling demands. In order to represent the office building stock, a large number of building parameters were explored such as built forms, fabrics, glazing levels and orientation. Selected parameters were combined into a large set of office building models (3840 in total). As different HVAC systems have different energy requirements when responding to same building demands, each of the 3840 models were further coupled with five HVAC systems: VAV, CAV, fan-coil system with dedicated air (FC), and two chilled ceiling systems with dedicated air, radiator heating and either embedded pipes (EMB) or exposed aluminium panels (ALU). In total 23,040 possible scenarios were created and simulated using EnergyPlus software. The annual heating and cooling demands and their HVAC system's heating, cooling and auxiliary energy requirements were normalised per floor area and fitted to two groups of statistical models. Outputs from the regression analysis were evaluated by inspecting models best fit parameter values and goodness of fit. Based on the described analysis, the specific regression models were recommended.
Occupant presence and behaviour in buildings has been shown to have large impact on heating, cooling and ventilation demand, energy consumption of lighting and appliances, and building controls. Energy-unaware behaviour can add one-third to a building's designed energy performance. Consequently, user activity and behaviour is considered as a key element and has long been used for control of various devices such as artificial light, heating, ventilation, and air conditioning. However, how are user activity and behaviour taken into account? What are the most valuable activities or behaviours and what is their impact on energy saving potential? In order to answer these questions, we provide a novel survey of prominent international intelligent buildings research efforts with the theme of energy saving and user activity recognition. We devise new metrics to compare the existing studies. Through the survey, we determine the most valuable activities and behaviours and their impact on energy saving potential for each of the three main subsystems, i.e., HVAC, light, and plug loads. The most promising and appropriate activity recognition technologies and approaches are discussed thus allowing us to conclude with principles and perspectives for energy intelligent buildings based on user activity.
We develop a statistical machine learning framework to study the effect of eight input variables (relative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, glazing area distribution) on two output variables, namely heating load (HL) and cooling load (CL), of residential buildings. We systematically investigate the association strength of each input variable with each of the output variables using a variety of classical and non-parametric statistical analysis tools, in order to identify the most strongly related input variables. Then, we compare a classical linear regression approach against a powerful state of the art nonlinear non-parametric method, random forests, to estimate HL and CL. Extensive simulations on 768 diverse residential buildings show that we can predict HL and CL with low mean absolute error deviations from the ground truth which is established using Ecotect (0.51 and 1.42, respectively). The results of this study support the feasibility of using machine learning tools to estimate building parameters as a convenient and accurate approach, as long as the requested query bears resemblance to the data actually used to train the mathematical model in the first place.
Building cooling load prediction is one of the key factors in the success of energy-saving measures. Many computational models available in the industry have been developed from either forward or inverse modeling approaches. However, these models usually require extensive computer resources and lengthy computation. This paper discusses the use of the multi-layer perceptron (MLP) model, one of the artificial neural network (ANN) models widely adopted in engineering applications, to estimate the cooling load of a building. The training samples used include weather data obtained from the Hong Kong Observatory and building-related data acquired from an existing prestigious commercial building in Hong Kong that houses a mega complex and operates 24 h a day. The paper also discusses the practical difficulties encountered in acquiring building-related data. In contrast to other studies that use ANN models to predict building cooling load, this paper includes the building occupancy rate as one of the input parameters used to determine building cooling load. The results demonstrate that the building occupancy rate plays a critical role in building cooling load prediction and significantly improves predictive accuracy.
Over the past few decades, application of artificial neural networks (ANN) to time-series forecasting (TSF) has been growing rapidly due to several unique features of ANN models. However, to date, a consistent ANN performance over different studies has not been achieved. Many factors contribute to the inconsistency in the performance of neural network models. One such factor is that ANN modeling involves determining a large number of design parameters, and the current design practice is essentially heuristic and ad hoc, this does not exploit the full potential of neural networks. Systematic ANN modeling processes and strategies for TSF are, therefore, greatly needed. Motivated by this need, this paper attempts to develop an automatic ANN modeling scheme. It is based on the generalized regression neural network (GRNN), a special type of neural network. By taking advantage of several GRNN properties (i.e., a single design parameter and fast learning) and by incorporating several design strategies (e.g., fusing multiple GRNNs), we have been able to make the proposed modeling scheme to be effective for modeling large-scale business time series. The initial model was entered into the NN3 time-series competition. It was awarded the best prediction on the reduced dataset among approximately 60 different models submitted by scholars worldwide.
With the development of modern computer technology, a large amount of building energy simulation tools is available in the market. When choosing which simulation tool to use in a project, the user must consider the tool's accuracy and reliability, considering the building information they have at hand, which will serve as input for the tool. This paper presents an approach towards assessing building performance simulation results to actual measurements, using artificial neural networks (ANN) for predicting building energy performance. Training and testing of the ANN were carried out with energy consumption data acquired for 1 week in the case building called the Solar House. The predicted results show a good fitness with the mathematical model with a mean absolute error of 0.9%. Moreover, four building simulation tools were selected in this study in order to compare their results with the ANN predicted energy consumption: Energy_10, Green Building Studio web tool, eQuest and EnergyPlus. The results showed that the more detailed simulation tools have the best simulation performance in terms of heating and cooling electricity consumption within 3% of mean absolute error.
An Earth-to-Air Heat Exchanger (ETAHE) is a low energy cooling and heating building component. It uses the ground's thermal storage to dampen ambient air temperature oscillations by delivering the air through a horizontally buried duct. To reduce airflow resistance, some hybrid ventilated buildings have recently adopted large cross-sectional area ducts. This paper describes the development of an Artificial Neural Network based Heat Convection (ANN-HC) algorithm to predict local average Nusselt Numbers along the duct surfaces. Furthermore, the ANN-HC algorithm is integrated with a transient three-dimensional heat transfer model based on finite element analysis of heat conduction in the ground domain surrounding the ETAHE to establish a new thermal modeling method for ETAHEs. A case study is presented to demonstrate the working principle of the new method. It is shown that the method can very well simulate the interactions between an ETAHE and its environment.
In this paper, support vector machine (SVM) is used to predict hourly building cooling load. The hourly building cooling load prediction model based on SVM has been established, and applied to an office building in Guangzhou, China. The simulation results demonstrate that the SVM method can achieve better accuracy and generalization than the traditional back-propagation (BP) neural network model, and it is effective for building cooling load prediction.
This study presents four modeling techniques for the prediction of hourly cooling load in the building. In addition to the traditional back propagation neural network (BPNN), the radial basis function neural network (RBFNN), general regression neural network (GRNN) and support vector machine (SVM) are considered. All the prediction models have been applied to an office building in Guangzhou, China. Evaluation of the prediction accuracy of the four models is based on the root mean square error (RMSE) and mean relative error (MRE). The simulation results demonstrate that the four discussed models can be effective for building cooling load prediction. The SVM and GRNN methods can achieve better accuracy and generalization than the BPNN and RBFNN methods.
The methodology to predict building energy consumption is increasingly important for building energy baseline model development and measurement and verification protocol (MVP). This paper presents support vector machines (SVM), a new neural network algorithm, to forecast building energy consumption in the tropical region. The objective of this paper is to examine the feasibility and applicability of SVM in building load forecasting area. Four commercial buildings in Singapore are selected randomly as case studies. Weather data including monthly mean outdoor dry-bulb temperature (T0), relative humidity (RH) and global solar radiation (GSR) are taken as three input features. Mean monthly landlord utility bills are collected for developing and testing models. In addition, the performance of SVM with respect to two parameters, C and ɛ, was explored using stepwise searching method based on radial-basis function (RBF) kernel. Finally, all prediction results are found to have coefficients of variance (CV) less than 3% and percentage error (%error) within 4%.
This paper reports the development of a building energy demand predictive model based on the decision tree method. This method is able to classify and predict categorical variables: its competitive advantage over other widely used modeling techniques, such as regression method and ANN method, lies in the ability to generate accurate predictive models with interpretable flowchart-like tree structures that enable users to quickly extract useful information. To demonstrate its applicability, the method is applied to estimate residential building energy performance indexes by modeling building energy use intensity (EUI) levels. The results demonstrate that the use of decision tree method can classify and predict building energy demand levels accurately (93% for training data and 92% for test data), identify and rank significant factors of building EUI automatically. The method can provide the combination of significant factors as well as the threshold values that will lead to high building energy performance. Moreover, the average EUI value of data records in each classified data subsets can be used for reference when performing prediction. One crucial benefit is improving building energy performance and reducing energy consumption. Another advantage of this methodology is that it can be utilized by users without requiring much computation knowledge.
The installation of a raised floor system can change the thermal behaviour of the building by reducing the interaction between the heat gains and the thermally massive concrete slab. In this study, the influence of the raised floor on the summer design day zone cooling load profile is evaluated with EnergyPlus for an office building located in San Francisco. The effects of structure type, window-to-wall ratio and the presence of carpet on the thermal behaviour of the raised floor are also investigated. The results show that the mere presence of the raised floor largely affects the zone cooling load profile and the peak cooling load over the range of −7 to +40%. The most significant parameters are the zone orientation, i.e. the exposure to direct solar radiation, and the presence of floor carpeting. If carpeting is present, commonly used in U.S. office buildings, the overall impact on zone peak cooling load is reduced, ranging from 0 to 5% greater for the raised floor than without it. Without carpet the peak cooling load is 4% greater with raised floor than without it in the north zone, 22% in the east and west zones, and 12% in the south zone.
Principal component analysis of dry-bulb temperature, wet-bulb temperature and global solar radiation was considered, and a new climatic index (principal component Z) determined for two emissions scenarios – low and medium forcing. Multi-year building energy simulations were conducted for generic air-conditioned office buildings in Harbin, Beijing, Shanghai, Kunming and Hong Kong, representing the five major architectural climates in China. Regression models were developed to correlate the simulated monthly heating and cooling loads and building energy use with the corresponding Z. The coefficient of determination (R2) was largely within 0.78–0.99, indicating strong correlation. A decreasing trend of heating load and an increasing trend of cooling load due to climate change in future years were observed. For low forcing, the overall impact on the total building energy use would vary from 4.2% reduction in severe cold Harbin (heating-dominated) in the north to 4.3% increase in subtropical Hong Kong (cooling-dominated) in the south. In Beijing and Shanghai where heating and cooling are both important, the average annual building energy use in 2001–2100 would only be about 0.8% and 0.7% higher than that in 1971–2000, respectively.
In this paper, we develop an online sequential learning algorithm for single hidden layer feedforward networks (SLFNs) with additive or radial basis function (RBF) hidden nodes in a unified framework. The algorithm is referred to as online sequential extreme learning machine (OS-ELM) and can learn data one-by-one or chunk-by-chunk (a block of data) with fixed or varying chunk size. The activation functions for additive nodes in OS-ELM can be any bounded nonconstant piecewise continuous functions and the activation functions for RBF nodes can be any integrable piecewise continuous functions. In OS-ELM, the parameters of hidden nodes (the input weights and biases of additive nodes or the centers and impact factors of RBF nodes) are randomly selected and the output weights are analytically determined based on the sequentially arriving data. The algorithm uses the ideas of ELM of Huang et al. developed for batch learning which has been shown to be extremely fast with generalization performance better than other batch training methods. Apart from selecting the number of hidden nodes, no other control parameters have to be manually chosen. Detailed performance comparison of OS-ELM is done with other popular sequential learning algorithms on benchmark problems drawn from the regression, classification and time series prediction areas. The results show that the OS-ELM is faster than the other sequential algorithms and produces better generalization performance.
Numerous advances have been made in developing intelligent programs, some inspired by biological neural networks. Researchers from many scientific disciplines are designing artificial neural networks (ANNs) to solve a variety of problems in pattern recognition, prediction, optimization, associative memory; and control. Although successful conventional applications can be found in certain well-constrained environments, none is flexible enough to perform well outside its domain. ANNs provide exciting alternatives, and many applications could benefit from using them. This article is for those readers with little or no knowledge of ANNs to help them understand the other articles in this issue of Computer. It discusses the motivation behind the development of ANNs; describes the basic biological neuron and the artificial computation model; outlines network architectures and learning processes; and presents multilayer feed-forward networks, Kohonen's self-organizing maps, Carpenter and Grossberg's Adaptive Resonance Theory models, and the Hopfield network. It concludes with character recognition, a successful ANN application.
Building Morphology, Transparence, and Energy Performance
- W Pessenlehner
- A Mahdavi
W. Pessenlehner, A. Mahdavi, Building Morphology, Transparence, and Energy
Performance, 2003.
Extreme learning machine for clustering
- C K L Lekamalage
- T Liu
- Y Yang
- Z Lin
- G.-B Huang
C.K.L. Lekamalage, T. Liu, Y. Yang, Z. Lin, G.-B. Huang, Extreme learning machine for clustering, in: J. Cao, K. Mao, E. Cambria, Z. Man, K.-A. Toh (Eds.),
Proceedings of ELM-2014, 1, Springer International Publishing, Cham, 2015,
pp. 435-4 4 4.