Study On Dynamic Prediction Of Molybdenum Concentrate Market Price Based On LSTM

In recent years,with the continuous upgrading of industrial production and the rapid growth of the housing and infrastructure construction industry in China,and the recovery of the global economy,molybdenum concentrate which as a basic raw material for the production of industrial and construction industry has shown a good form of price fluctuation.The accurate prediction of the future medium and long term market price of molybdenum concentrate is beneficial to the development of reasonable production plan and mining cost control of molybdenum mining enterprises in China,as well as the rational planning and utilization of molybdenum mineral resources in China.Therefore,it is necessary to study the long-term price forecast of molybdenum concentrate.In this paper,theories and technologies related to prediction are adopted to conduct in-depth research on medium and long-term price prediction of molybdenum concentrate under the fluctuation of molybdenum market.The main research contents are as follows:(1)Using the relevant content of management economics and literature analysis method to conduct a comprehensive analysis of the market and price influencing factors of molybdenum concentrate,so as to obtain the development status of China's molybdenum market,supply and demand of molybdenum concentrate and molybdenum concentrate relevant factors affecting the price of the mining market.(2)Decomposing complex nonlinear and non-stationary time series by decomposition method,and then establishing linear ARIMA and nonlinear LSTM prediction models based on the linear and nonlinear characteristics of the components obtained by decomposition,so as to propose EMD-ARIMA-LSTM multi-step time series prediction methods.The simulation model of the prediction model was carried out by using China's molybdenum concentrate price,global copper price and global aluminum price.The results show that the EMD-ARIMA-LSTM prediction method is better than the ARIMA,LSTM,EMD-ARIMA and EMD-LSTM prediction models in mid-and long-term predictive performance.(3)On the basis of the analysis of molybdenum concentrate market and price influencing factors,a dynamic forecasting method of molybdenum concentrate market price is proposed based on improved EMD-ARIMA-LSTM.Firstly,the multi-dimensional molybdenum data is constructed from the price dimension,supply and demand dimension and macroeconomic dimension of molybdenum.Then,the multi-dimension molybdenum data is used to construct the mid-and long-term market price dynamic prediction model of molybdenum concentrate based on EMD-ARIMA-LSTM multi-step prediction method and BP neural network.(4)The simulation experiments of the proposed molybdenum concentrate market price dynamic prediction method were carried out by using the example data,and MAPE and RMSE were used as the measurement indexes of prediction accuracy,and the prediction performance of the prediction method was analyzed and applied.