Electricity Price Forecasting Based on Improved ARIMA Model

Electricity price forecasting is important because it can prompt market participants to trade electricity more effectively and optimize resource allocation. The purpose of this paper is to improve the accuracy of electricity price forecasting by improving the traditional ARIMA model, and to solve the limitations of existing models in dealing with nonlinear characteristics, volatility aggregation and other issues. The research method combines wavelet transform and ARIMA model. Firstly, multi-resolution analysis of electricity price data is carried out by using wavelet transform to extract the signal characteristics of different frequency components, and then the ARIMA model is established to predict the decomposed subsequence. The experimental results show that the improved model shows a significant performance improvement in electricity price forecasting, and the prediction effect of the first-order wavelet transform is better than that of the second-order signal decomposition. Compared with the traditional ARIMA model, the improved model can capture the nonlinear characteristics and volatility clustering phenomenon of electricity price data more effectively, and also provides a strong technical guarantee for the efficient operation and decision support of the electricity market.