Coal is an important basic energy,especially thermal coal occupies a very high strategic position in China,but the prediction of coal price is very difficult.The recurrent neural network(RNN)was introduced to predict the price of thermal coal.On this basis,in view of the characteristics that the price of thermal coal fluctuates greatly with time,the long short term memory model(LSTM)was established by optimizing the RNN model,and the support vector regression machine model(SVR)was introduced.The LSTM-SVR combined model was formed by series to reduce the risk of single model prediction and improve the accuracy of prediction results.At the same time,the moving average method was used to improve the correlation between the characteristic data and the price of thermal coal.The results show that the development trend of Shanxi thermal coal price in the first half of 2023 predicted by the LSTM-SVR combined model has a high linear fitting with the actual coal price,and the prediction accuracy rate reaches 95.69％.The model predicts that the price of thermal coal in Shanxi will gradually decrease in 2024,from a maximum of about 1200 yuan/t to 700 yuan/t.The research results are of great significance for coal enterprises to adjust their business strategies,optimize their internal capital structure and maintain the long-term stable development of the whole industry.
关键词
动力煤/价格预测/循环神经网络/长短期记忆模型/LSTM-SVR组合模型
Key words
Thermal coal/Price prediction/Recurrent neural network/Long short term memory model/LSTM-SVR combined mode
维普
万方数据