Author(s)

Jiale Xu, Zhen Wang, Dongxiao Li

Corresponding Author

Jiale Xu

ABSTRACT

This paper mainly studies the comprehensive prediction of coal price, based on the time series of influencing factors. On the basis of ARIMA prediction, the coal price prediction model based on BP neural network is established. Through quantitative analysis, the order of main factors affecting coal price and Qinhuangdao Port steam coal is given. It is required to establish a prediction model based on the annex of Qinhuangdao Port steam coal price and the main influencing factors.Firstly, the stationarity test is carried out on the data of each influencing factor, and then the stable series is constructed by the difference of the unstable data. Then, the time series model based on the influencing factors is established, and the prediction of the data of each influencing factor is obtained. Then we establish BP. The neural network takes the influencing factors as the input layer, tests the trained network with fresh data, and analyzes the residual error of the prediction data. Finally, the prediction data of each influencing factor is replaced into the BP neural network model which has been trained and has a high degree of fitting. The prediction model is established by comprehensively considering the changes of influencing factors caused by various situations in the future, and the multiple regression model is established by introducing dummy variables and determining the multiplier of cross terms. This paper studies the trend of coal price under the economic crisis, supply side reform, snow disaster, debt crisis, imported low-cost coal and so on, and puts forward targeted suggestions to the government for various emergencies.

KEYWORDS

Time series analysis, BP neural network model, multiple regression analysis, cubic spline interpolation