Demand For Residential Electricity Consumption Of Beijing Based On Artificial Neural Network

Since reform and opening up, China’s economy has been developing in high-speed. Power as a clean, convenient and safe source of energy, has been closely linked with China’s economic development and the improvement of living standards. With a variety of home appliances and electronic products into people’s lives, people’s demand for power is growing, residential electricity consumption has gradually become a important part of the power structure of the whole society, its consumption level is a measure of population modernization and social development level of scale. Beijing as China’s capital and political and cultural center, their development is to a large extent also represents the high level of domestic electric power development. Therefore, the real good job in Beijing residential electricity demand forecasting can help government departments to develop efficient power policies, power planning, reduce the growing pressure on the environment, but also contribute to reasonable arrangements for electricity power enterprise production and management programs to improve people’s living standards.It is not easy to do the job such as the forecast of residential electricity consumption. On the one hand, many factors affect the residential electricity consumption and the data are not easily to obtain, on the other hand,scholars have their own view to form residents demand model.So, considering all aspects, from residential electricity consumption influencing factors to selection of residential electricity consumption model has done a detailed analysis. As a model forecast often neglect to consider certain factors change, the paper considers the advantages and disadvantages of the grey model and artificial neural network model, using the concept of combination forecast model to construct grey neural network forecasting model, and on this basis propose the conventional serial grey neural network forecasting model, added the main variable effects on residential electricity consumption. Then combination prediction model optimized and conventional forecasting model to be tested and compared using 1990-2012 data from Beijing. Finding using the optimized serial grey neural network model to predict the future of residential electricity consumption have higher prediction accuracy. So finally we use a series of improved RBF Neural network in 2017-2022 Beijing residential electricity consumption. At last, the nine policy recommendations have been proposed for Beijing Municipal Government based on the analysis historical and the prediction results.