Research on the Probability Prediction Method of Electricity Price Response Load based on CNNLSTM and Quantile Regression Fusion

This article explores the importance of power load forecasting in the power industry, especially the uncertainty when considering dynamic electricity pricing factors. Traditional methods cannot handle the impact of electricity prices on load, so deep learning techniques, especially convolutional neural networks (CNN) and short-term memory networks (LSTM), have been introduced. These two networks, combined with quantile regression, are used for probability density load forecasting, providing a complete probability distribution of load demand and supporting the smart grid. This article provides a detailed introduction to the principles of CNN and LSTM networks, and explains the concept of quantile regression. The study validated the effectiveness of this method using an Australian dataset. Through this method, real-time electricity price factors can be better considered, the accuracy of power load forecasting can be improved, and strong support can be provided for the planning and operation of the power system.