Short-Term Electricity Forecasting Based On A Hybrid Optimization Model

As one of the most important energy sources,electricity plays a vital role in the power system and is the main driving force for the development of the country and society.Especially,short-term electricity demand forecast is more important in power system planning.However,unlike other energy sources,electricity cannot be stored on a large scale.Overestimation and underestimation of short-term electricity consumption will waste the energy or cause unnecessary loss.Therefore,accurate and effective forecasting of electricity demand can help power system operators and market participants to propose bidding strategies and ensure consumers' electricity supply based on corresponding forecast information,thereby reducing the cost of electricity consumption and reducing energy consumption.Based on the seasonal effect and nonlinearity of short-term demand data,a hybrid optimization forecasting model(SEA-EEMD-GPEE)is proposed for analysis and prediction of hourly electricity consumption data accurately.Due to the original power consumption sequence is often affected by seasonal effects,this paper first performs seasonal index adjustment(SEA)on the sequence to eliminate the seasonal items of the sequence,and then uses the set empirical mode decomposition(EEMD)method to decompose the series which was removed seasonal effects.The sequence is decomposed into several sub-sequences at different frequencies.These IMFs are integrated into the three sub-sequences of the original sequence,which are the high-frequency sub-sequences,the intermediate-frequency sub-sequences,and the low-frequency sub-sequences.GRNN is used to predict the high-frequency sequences,PSO-Elman neural network is used to predict the mid-frequency,among them,the parameters in the Elman neural network are optimized using PSO,and extreme learning machine models(ELM)is established for low-frequency sequences.Finally the hybrid optimization forecast model SEA-EEMD-GPEE is obtained.This paper analyzes and predicts the hourly electricity demand data of California,and establishes four comparison models(GRNN,SEA-GRNN,SEA-EM-D-GEE,SEA-EEMD-GEE).Empirical research shows that the prediction effect of SEA-EEMD-GPEE prediction model is superior to other four comparison models and has higher prediction accuracy.