The Electrical Load Forecasting Based On Multi-elements Time Series Analysis Method Of LWT

In order to distribute electric energy and plan the operation of power grid,precise electrical load forecasting of residential estate is essential.However the forecast methods now are poor at processing big data with great random factors.We propose the Multi-elements Time Series Analysis Method of Lifting Wavelet Transform(LWT-MTSAM)to proceed electrical load forecasting,verifying it’s accuracy through simulation.Initially,in terms of the volatility and the massive nature of the power load,using LWT which is simple in calculation and applicable in wide range situation,extracting the main features of power load of residential area,removing the high-frequency noise in load sequence,reducing the uncertainty of the original data;Using Time Series Analysis Method(TSAM),according to the regular factors contained in the data,using it’s strong capability to big data processing,finding and extending the law inside.Then through the significant test to rectify the result in time,in order to accurately predict the power load.Secondly,The prediction ability to power load of traditional TSAM is still not strong enough due to the random factors and weak regularity inside,so this paper put forward the improved method based on this traditional method——MTSAM.The power load series is divided into multiple sub sequences affected by random factors such as the temperature,the influence of holidays and a regular sequence which is not affected by any random factors.Making the most suitable temperature range and holiday correction coefficient,respectively predicting each sub sequence according to it’s characteristics,finally adding prediction results together,achieving the full power load forecasting sequence.Through results feedback of neural network learning algorithm,continuously feedback forecast results,adjusting the most suitable temperature range and holiday correction coefficient,to ensure the accuracy of forecasting later.Terminally,making use of the big data of temperature and historical power load of some estate in Jia Ding district Shanghai,confirming the feasibility of these two power load forecasting methods mentioned above through simulation.Besides,briefly introducing the frame of forecasting system based on this method.