Research On Coal Price Prediction Model Based On Machine Learning

Coal is the most important resource in China and has always occupied a leading position in China’s energy structure.Due to the complexity and uncertainty of coal prices,as well as the impact of policies and climate,it is difficult to predict coal prices.In today’s increasingly fierce market competition,how to predict the coal market scientifically and efficiently is an important issue to stabilize the energy supply.This article aims to analyze the characteristics of short-term coal price data,construct suitable single prediction models and combined prediction models,and ultimately achieve efficient and accurate coal price prediction,providing reference basis for coal enterprises to avoid market risks.The main work of this paper is as follows:1.The advantages and disadvantages of existing coal price prediction algorithms are analyzed.According to the characteristics of coal price data,the missing values and outliers of the data set are processed,and the factors affecting the coal price are selected from different angles for descriptive analysis.2.Starting from three different perspectives of econometrics,machine learning,and deep learning,features are filtered,and finally a sliding window strategy is used to construct supervised learning data.3.A deep learning prediction model based on Attention mechanism is proposed.Firstly,four single models of RNN,LSTM,Bi LSTM and CNN are constructed,and their advantages and disadvantages are analyzed.In order to further improve the single neural network model,Attention mechanism is introduced to improve the accuracy and reliability of the model.Comparing the models,it is found that the Bi LSTM model and CNN model based on Attention mechanism are more suitable for short-term coal price prediction.4.In this paper,different combination forecasting methods based on series and parallel are studied,and two better single models are fused together.The combination method can effectively solve the problem of unstable prediction results of a single model at the price fluctuation point,and can better meet the needs of practical applications.The experimental results show that the CNN-Bi LSTM series prediction model based on the Attention mechanism proposed in this paper is significantly better than the CNN-Bi LSTM series model and the CNN-Bi LSTM parallel model,and the evaluation index MAPE decreases to 0.16%,which proves the effectiveness of the model.The CNN-Bi LSTM series combination model based on Attention mechanism proposed in this paper can significantly improve the accuracy of short-term coal price prediction,and can be used as an effective model for short-term coal price prediction and analysis.