Prediction Of Coal Mine Gas Concentration And Safety Warning Management Based On LSTM

In the process of coal mine safety production,gas accidents have always been an important factor affecting the safety of underground personnel.In order to effectively prevent gas disasters,various gas prediction and early warning technologies are gradually applied to coal mine safety management equipment.Accurate and effective gas prediction and early warning is of great significance for ensuring the safety of workers.The change of gas monitoring data in fully mechanized coal mining faces is important for coal mine safety management.Therefore,this article takes the collected gas monitoring data in fully mechanized coal mining faces as an example to carry out gas concentration prediction and early warning research.The gas concentration in fully mechanized coal mining faces is affected by multiple factors,using only univariate prediction will affect the accuracy of gas concentration prediction.Therefore,it is necessary to select factors that have a high correlation with the gas concentration in fully mechanized coal mining faces to build a multifactor gas concentration prediction model.The gas monitoring data collected by sensors in fully mechanized coal mining faces contain many abnormal and missing values due to the influence of external factors such as environment.Therefore,first,the near average method and Lagrange interpolation method are used to process the abnormal and missing values,and Pearson correlation coefficient is used to eliminate the factors that have low correlation with the gas concentration in fully mechanized coal mining faces.Finally,five factors are selected as model input features to build a gas concentration prediction model.Based on the gas monitoring data,an LSTM gas concentration prediction model is established,which is optimized from three aspects: the number of neurons in the LSTM hidden layer,the number of hidden layers,and the batch size.Univariate and multivariate LSTM models,LSTM models with different prediction steps,and the prediction effects of different models are verified.In order to improve the performance of LSTM model in gas concentration prediction,CNN convolutional neural network and Attention attention mechanism were introduced into LSTM network,and gas concentration prediction models based on Attention LSTM and Attention CNN-LSTM were established.The prediction effects of different models were compared using three evaluation indicators,RMSE,MAE,and MAPE.The experimental results show that the prediction effect of the Attention CNN-LSTM model is better than that of the Attention LSTM model and LSTM model.The RMSE,MAE,and MAPE of the model are only 0.0157,0.0109,and 0.0217.On the basis of gas prediction,the warning analysis of gas concentration,first of the warning level of gas concentration,and determine the basic indicators of the gas concentration warning and gas concentration early warning threshold,finally through the examples of the gas concentration prediction warning analysis,the results show that the actual value of the warning situation is consistent with the forecast value of the warning.This article combined LSTM neural network to accurately predict the gas concentration in fully mechanized mining faces,providing a new technical means for processing gas monitoring data.Based on the prediction,early warning analysis was conducted on the gas concentration in fully mechanized mining faces.The effectiveness of the early warning method was verified through experiments,which can provide a certain reference basis for gas control and coal mine safety management.Figure [29] Table[20] Reference[82]...