Short-term Forecasting of Individual Residential Load Based on Deep Learning and K-means Clustering

In order to currently motivate a wide range of various interactions between power network operators and electricity customers,residential load forecasting plays an increasingly important role in demand side response(DSR).Due to high volatility and uncertainty of residential load,it is significantly challenging to forecast it precisely.Thus,this paper presents a short-term individual residential load forecasting method based on a combination of deep learning and k-means clustering,which is capable of effectively extracting the similarity of residential load and performing residential load forecasting accurately at the individual level.It first makes full use of k-means clustering to extract similarity among residential load and then employs deep learning to extract complicated patterns of residential load.The presented method is tested and validated on a real-life Irish residential load dataset,and the experimental results suggest that it can achieve a much higher prediction accuracy,in comparison with a published benchmark method.

Electrical demand aggregation effects on the performance of deep learning-based short-term load forecasting of a residential building

Modern power grids face the challenge of increasing renewable energy penetration that is stochastic in nature and calls for accurate demand predictions to provide the optimized power supply.Hence,increasing the self-consumption of renewable energy through demand response in households,local communities,and micro-grids is essential and calls for high demand prediction performance at lower levels of demand aggregations to achieve optimal performance.Although many of the recent studies have investigated both macro and micro scale short-term load forecasting(STLF),a comprehensive investigation on the effects of electrical demand aggregation size on STLF is minimal,especially with large sample sizes,where it is essential for optimal sizing of residential micro-grids,demand response markets,and virtual power plants.Hence,this study comprehensively investigates STLF of five aggregation levels(3,10,30,100,and 479)based on a dataset of 479 residential dwellings in Osaka,Japan,with a sample size of(159,47,15,4,and 1)per level,respectively,and investigates the underlying challenges in lower aggregation forecasting.Five deep learning(DL)methods are utilized for STLF and fine-tuned with extensive methodological sensitivity analysis and a variation of early stopping,where a detailed comparative analysis is developed.The test results reveal that a MAPE of(2.47-3.31%)close to country levels can be achieved on the highest aggregation,and below 10%can be sustained at 30 aggregated dwellings.Furthermore,the deep neural network(DNN)achieved the highest performance,followed by the Bi-directional Gated recurrent unit with fully connected layers(Bi-GRU-FCL),which had close to 15%faster training time and 40%fewer learnable parameters.

Control Strategies of Large-scale Residential Air Conditioning Loads Participating in Demand Response Programs

Residential air conditioning(RAC)loads have great potential to be included in demand response(DR)programs.This paper studies large-scale RAC loads participating in DR programs,such as modeling,parameters identification,DR characteristics and control strategies.First,an aggregate model of large-scale RAC loads are established based on the buildings’performance with heat storage and insulation,avoiding the calculation of a single RAC model.Then,parameters of the aggregate model are identified based on the RACs’power and outdoor temperatures.Based on the aggregate model,DR characteristics of RAC loads are analyzed,including the dynamic relationship between power,outdoor and indoor temperature,and the potential of DR combined with the users’comfort.Next,the DR control strategies adapted for large-scale RAC loads are established by adjusting the temperature set-points.The DR strategies consider users’comfort and calculate the control signals of each RAC load according to the DR power,including adjustment temperature and adjustment time,which are sent to each RAC load for execution.In the DR process,the control center does not need to obtain the users’indoor temperature,which is conducive to protecting the users’privacy.DR strategies of RAC loads when the control degree within/beyond the DR potential are both proposed,and a load recovery control strategy is also introduced.Finally,the effectiveness and accuracy of the proposed model and DR control strategies are verified by simulation results.

Technologies and Practical Implementations of Air-conditioner Based Demand Response

Nowadays,the most notable uncertainty for an electricity utility lies in the electrical demand of end-users.Demand response(DR)has acquired considerable attention due to uncertain generation outputs from intermittent renewable energy sources and advancements of smart grid technologies.The percentage of the air-conditioner(AC)load over the total load demand in a building is usually very high.Therefore,controlling the power demand of ACs is one of significant measures for implementing DR.In this paper,the increasing development of ACs,and their impacts on power demand are firstly introduced,with an overview of possible DR programs.Then,a comprehensive review and discussion on control techniques and DR programs for ACs to manage electricity utilization in residential and commercial energy sectors are carried out.Next,comparative analysis among various programs and projects utilized in different countries for optimizing electricity consumption by ACs is presented.Finally,the conclusions along with future recommendations and challenges for optimal employment of ACs are presented in the perspective of power systems.