Research On Short-term Electricity Price Forecasting Model Based On VMD-ISSA-GRU

As a key indicator of the electricity market,electricity price forecasting has always attrac ted much attention.Accurate electricity price forecast can provide relevant information for bot h parties in the electricity market to formulate the best bidding strategy and maximize benefits.At the same time,under the background of a new round of electricity market reform in my co untry,an accurate electricity price forecast model is designed.It can also provide reference opi nions for the future development of the power financial market and the effective trading of bot h parties in the power market.In order to improve the prediction accuracy and prediction speed,a hybrid model based o n variational modal decomposition,improved sparrow search algorithm and GRU neural netw ork model is proposed for the non-stationarity and high volatility of electricity price time serie s.Predictive analysis of electricity price time series.The model first uses the variational moda l decomposition algorithm(VMD)to decompose the original electricity price time series into I MF components of different frequencies,and then uses the improved sparrow search algorithm to optimize the number of neurons in each hidden layer of the GRU model and the number of iterations.Finally,the optimized GRU models are used to predict each IMF component,and th e predicted value of each component is superimposed and reconstructed to obtain the final pre dicted value.The improved sparrow algorithm is to combine Tent chaotic map,mixed sine and cosine algorithm and Levy flight strategy to improve the population initialization method,fin der position update formula and follower position update formula of sparrow search algorithm respectively.This paper uses RMSE,MAE,MAPE,R~2 as four evaluation indicators to evalua te the prediction effect of the proposed model.Based on the electricity price data of the Nordic electricity market from 2013 to 2017,thi s thesis uses the benchmark model and VMD-ISSA-GRU to forecast short-term electricity pric es.The research results show that compared with the GRU model,VMD-GRU can see that the prediction errors of RMSE,MAE and MAPE of the VMD method are reduced by 63.57%,49.75%and 43.21%on average,respectively.Compared with VMD-GRU,VMD-ISSA-GRU has an average reduction of 68.95%,71.54%and 72.86%in the prediction errors of RMSE,MAE and MAPE of the improved model using ISSA.Compared with SSA-GRU,ISSA-GRU has a n average reduction of 15.19%,33.59%and 38.07%in the prediction errors of RMSE,MAE a nd MAPE of the improved model using ISSA.It shows that the VMD algorithm and the impro ved sparrow algorithm are of great significance to improving the prediction accuracy of the sh ort-term electricity price prediction model.