Study On The Prediction And Early Warning Method Of Mine Water Inflow Based On LSTM-GRU

Coal occupies an important position in China’s energy structure and is vital to the development of the national economy.However,mine water gushing is an important issue in mine safety production.When the variation of water surge is abnormal,it will not only affect the normal production of coal mines,but also may cause property damage and human casualties.Therefore,the prediction,control and early warning of mine water gushing accident is essential to ensure the energy safety and sustainable development of China’s economy and society.In this thesis,in order to solve the problem of mine water surge prediction and early warning,the LSTM model and GRU model are combined to establish a mine water surge prediction model based on LSTM-GRU network,which can take into account the advantages of LSTM model and GRU model,while avoiding their disadvantages.Based on the historical mine water surge data in the study area and the relevant laws and regulations on mine safety production,we combined the principles to determine the threshold value of mine water surge abnormality,and integrated the use of the water surge gradient warning model and the drainage capacity of the mine to carry out graded warning of mine water surge.Finally,a prediction and early warning method of mine water influx based on LSTM-GRU network model was formed.The following research work was mainly carried out:(1)studied the prediction method of mine water surge,and established a mine water surge prediction model based on LSTM-GRU network.Introducing deep learning theory,the combination of long and short-term memory network(LSTM)and gated recurrent unit(GRU),the study object of mine water influx was selected,and a combined LSTM-GRU neural network model was established for prediction.Tingnan coal mine as the study area,Tingnan coal mine historical mine water influx as sample data,before modeling data pre-processing work,mainly with interpolation method to complement the data and normalization,to ensure the quality of data.In modeling,a7:3 ratio was used to divide the data set into a training set and a test set,and the gradient descent algorithm with better model training effect was selected to determine the network model parameters and regularization parameters.To demonstrate the prediction accuracy of the LSTM-GRU model,a traditional time-series ARIMA model was selected as the control group,and a single LSTM model was built to predict the same data through ablation experiments.Four evaluation indexes were used to verify that the combined prediction model proposed in this thesis has high prediction accuracy and reliability,and the prediction effect is better than the traditional time series prediction model and the prediction accuracy of the combined model is better than that of the single model.(2)research based on the history of mine surge data mine surge water classification warning method.Based on the historical mine surge data and relevant laws and regulations on mine safety production in the study area,the threshold value of mine surge abnormality was determined by using the principle.The graded early warning of mine water influx was carried out by using the integrated water influx gradient early warning model and the drainage capacity of the mine.Taking the actual water surge event in Tingnan coal mine as an example,the warning threshold was delineated according to the actual measured data of the water surge in Tingnan coal mine,and the validity test was conducted on the predicted mine water surge sequence based on the prediction of historical data of mine water surge by LSTM-GRU model.The test results show that the method of dividing the level of mine water gushing warning and the method of setting the warning threshold is reasonable and practical,and provides some theoretical method support for the prevention and control of mine water gushing accidents.On this basis,this thesis uses the collected mine surge data,using LSTM-GRU network model,established for coal mining mine surge prediction and warning method.The actual mine water gushing accident test,proved the applicability and effectiveness of the proposed method in the coal mine site,can provide reference and reference for the prevention and control of water gushing accident in coal mine safety production.There are 35 figures,8 tables and 83 references in this paper.