Research on Medium and Long-term Electricity Price Hybrid Forecasting Model Based on Information Entropy and GM-MLP-ARIMA

The electric power industry, as the cornerstone of national economy and people's livelihood, has made the fair and reasonable electricity pricing mechanism a focal point of research. The formulation of electricity prices involves the interests of numerous market participants, hence the importance of accurate electricity price forecasting. Electricity price forecasting requires an in-depth analysis of supply and demand conditions as well as the influence of market participants. Additionally, factors such as power generation costs, coal prices, and green certificate prices also affect the formation of electricity prices. With the deepening of China's power industry market reform, the competitive pressure faced by market participants in the power sector is continuously increasing. This has led to higher demands for the accuracy and stability of electricity price forecasting. Electricity price forecasting is crucial for market participants in formulating bidding strategies, adjusting production plans, conducting risk management, planning, and financial analysis. However, due to uncertain factors such as power generation costs, market supply and demand, climate conditions, and policy changes, as well as the non-stability of the market environment, electricity price forecasting is highly challenging, especially for medium and long-term forecasting. Medium and long-term electricity price forecasting is the foundation of power industry planning, investment, production, dispatching, and trading, and is vital for ensuring the safe and economic operation of the power system. Generally, the longer the prediction time span, the worse the accuracy and stability, and the larger the error. Traditional electricity price forecasting methods mainly rely on historical data to establish prediction models and use single mathematical methods for forecasting. These methods often fail to fully consider all influencing factors, and the models are difficult to adapt to long-term market changes, resulting in significant differences between medium and long-term prediction results and actual values.To address these issues, this paper proposes a multi-model combination electricity price forecasting method based on information entropy. This method not only considers historical concurrent data but also integrates various factors affecting electricity prices, such as installed capacity, power generation, electricity load, and coal prices. It employs improved GM(1,1) grey model, multilayer perceptron (MLP) neural network model, and ARIMA time series model, among other forecasting models. By combining the advantages of these models and evaluating their stability and accuracy based on their prediction errors, a rolling correction forecast of medium and long-term electricity prices is conducted to provide reasonable electricity price prediction values. The effectiveness of the proposed medium and long-term electricity price forecasting method is verified through simulation tests using actual market data. The test results indicate that this multi-model combination method has good adaptability and can significantly improve the accuracy of medium and long-term predictions. This not only provides more accurate price signals for market participants in the electricity market, helping them make trading decisions, protect their market interests, and promote the rational allocation of resources, but also has significant implications for ensuring the economic and reliable operation of the power system.