Research On Deep Learning Hybrid Algorithm Used In Mine Economic Evaluation

It is very important to estimate the cost of main raw materials in mine economic evaluation.Therefore,there is significance in the research of obtaining accurate price predictions that has applied value in raising the evaluation level scientifically with reducing investment risk of enterprises.Inspired by the idea of hybrid model,this paper combines two models that can complement each other so as to obtain more accurate prediction results.First,to increase the accuracy of price predictions,this paper proposes a hybrid model of gated recurrent unit neural network model based on empirical mode decomposition(EMDGRU-Averaging,EGA).The EGA model first uses ensemble empirical mode decomposition to decompose the original time series data into data components at different frequencies,which reflect the fluctuation law of the original series on different time scales.Then,use a gated recurrent unit neural network to predict those data components separately,in order to handle the long-term and short-term dependencies in the time series data.Finally,to reduce the complexity,the EGA model averages the training results to get the predicted price.Second,to reduce the impact caused by missing data,this paper proposes a data completion prediction(Tucker-EMD-GRU-Averaging,TEGA)model based on Tucker decomposition.As time series are correlated,the TEGA model first fills in the missing data with the average price of the previous three days to complete time series data,then uses the Tucker decomposition algorithm to fix the interpolated data by mining their hidden relationship.Finally,it uses the EGA model to forecast the interpolated data to obtain the predicted price.In this paper,in order to verify the generalization and robustness of the two models proposed in this paper,a large number of experiments have been conducted on the data set of a certain raw material required for mining provided by a company.This paper uses three evaluation indexes to compare and evaluate the chosen representative comparison models from different angles,including mean absolute error,root mean square error and pearson correlation coefficient.According to the results,the EGA model can effectively improve the accuracy of price forecasts,the TEGA model can still make economic evaluations obtain relatively ideal price predictions in the case of missing data.