Research On Air-conditioning Load Analysis Forecasting Method In Smart Residential Area

At present，air conditioning load, as an important part of electric power load, isreceived increasing attention by researchers. Air conditioning load rises sharply within afew hours in summer, which poses a potential threat to the safe operation of the grid. Inorder to protect the security and stability of the electricity power of residents, airconditioning load forecasting has become an integral part of the grid scheduling system.Using existing resources to establish a reasonable air conditioning load forecasting modeland improve the prediction accuracy is a difficult work for researchers.Based on the "smart power empirical project" nudertaked by Beijing GuodiantongNetwork Technology Co. Ltd, the key technology of data acquisition for air-conditioningload in smart residential area is described, and then air conditioning load is analyzed andpredicted.Study experiences show that air conditioning load is affected by many factors insummer. Air-conditioning load is particularly sensitive to meteorological factors such asweather diversification. Based on this, firstly, the relationship between the residents of airconditioning load and meteorological factors (temperature, wind speed) is analyzed, andthen three BP network prediction models are constructed which based on the integralpoint meteorological factors. Prediction accuracy of models are compared and analyzedto verify the validity.However, BP network has a limitation on the long time training. A RBF neuralnetwork forecasting model is built and improved by K-means algorithm. And then, inorder to improve the accuracy of load forecasting, A RBF model combined with thegenetic algorithm is proposed and simulated on MatlabR2012a experimental platform.Simulation results are compared with the actual load, and the errors are analyzed. Thefeasibility and effectiveness of the improved method is proved.The studies provide a reference in building the platform for Beijing Power SupplyCompany on residents air conditioning load forecasting system. it also has importantapplication value.