Analysis And Prediction Of Coal Quality Data Based On Modern Neural Network

The analysis and prediction of coal quality data based on neural network and deep learning functions is to analyze and predict the calorific value and sales volume of coal quality data with the method of modern neural network,so as to achieve the auxiliary guiding role for coal mine production and operation activities.Coal quality test is an important link in the process of coal production and sales.The calorific value of coal is an important index to demarcate the price in the process of sale and a decisive condition to distinguish the use of coal.In the production and operation of coal mines,the calorific value of coal and other indexes are determined by physical methods through coal quality test to understand the composition of coal deeply and accurately,so as to guide the mining,production and sale of coal.In recent years,with the proposal of deep learning,neural network has begun to glow a new round of life,and has been widely used in many artificial intelligence applications,showing excellent performance.The new technology in deep learning plays an important role in the analysis and prediction of the complex non-linear relationship among many indexes of coal quality.Therefore,based on the real demand of production practice,combined with the characteristics of modern neural network technology and coal quality data set,this paper designs and realizes a set of coal calorific value,sales volume analysis and shortterm prediction method based on modern neural network method.The main contents can be summarized as follows:(1)Design and implementation of coal calorific value analysis based on multilayer neural networkAiming at the correlation analysis of coal calorific value and coal quality index,an analysis system based on multi-layer neural network is designed,which can accurately fit the correlation between coal calorific value and coal quality index with an accuracy of 99.40%.It can meet the supervision requirements of the coal quality data entered into the system in the production practice,and achieve the purpose of early warning the records that do not conform to the correlation relationship 。(2)Design a short-term forecast of heat output and sales volume based on LSTM neural networkAiming at the short-term prediction of coal calorific value and sales volume,a short-term prediction system based on long-term memory neural network is designed to realize the short-term prediction of coal calorific value and sales volume.Among them,the single variable and multi variable network inputs are used for the prediction of calorific value,and the final accuracy reaches 98.81%;for the prediction of coal sales,a prediction method based on multi model is designed,which improves the prediction accuracy of the original LSTM neural network,and the final prediction accuracy reaches 95.13%,meeting the guidance of production time assignment.