Research On Prediction Model Of Coal And Gas Outburst Based On Optimized Quantum Gate Neural Network

The occurrence of coal and gas outburst not only causes serious economic losses to coal enterprises,but also threatens the life safety of underground workers.Carrying out the risk prediction of coal and gas outburst can realize the "early detection and early prevention" of gas outburst accidents and minimize the losses caused by gas outburst.The prediction of coal and gas outburst is a typical nonlinear prediction problem,and the prediction accuracy is often unsatisfactory.For this reason,this dissertation proposes a prediction model of gas outburst risk level based on sd PSO and QGNN.First of all,based on the "comprehensive hypothesis" of gas outburst,this dissertation analyzes the conditions,process and influencing factors of gas outburst,and establishes a gas outburst prediction index system,which fully reflects the influence of geology,gas and physical properties of coal on gas outburst.Considering the similarities and differences of the role and influence degree of each prediction index in the process of gas outburst occurrence and evolution,the grey correlation analysis is used to calculate the correlation degree between gas outburst intensity and each prediction index,so as to determine the main control factors of gas outburst and reduce the redundancy of information.In view of the nonlinear and uncertain characteristics of gas outburst prediction,a quantum gate neural network is proposed to improve the ability to solve the uncertainty problem.This dissertation analyzes the network structure and learning algorithm of quantum gate neural network,and optimizes the hidden layer angle offset matrix(38)and output layer angle offset matrix(37)of quantum gate neural network by introducing the sub-dimension evolution to improve the particle swarm update mode,which improves the global search ability and the stability of particle swarm optimization algorithm,improves the local extremum points of quantum gate neural network,and is easy to fall into the shortage of local convergence.In order to verify the optimization performance of particle swarm optimization algorithm based on sub-dimension evolution,three test functions,Sphere,Rastrigin and Griewank,which are nonlinear and variable dependent and uncorrelated,are selected to optimize through PSO and sd PSO respectively.The success rates of three-dimensional test functions,PSO and sd PSO are 100%,25%,12%;100%,83%,62%,and the success rates of 10-dimensional test functions,PSO and sd PSO are 33%,0,0;100%,40% and 53%,respectively.The experimental results show that the success rates of PSO for multimodal nonlinear functions are significantly lower than that of sd PSO,which shows that sd PSO has improved the global search ability and the stability of the particle swarm optimization algorithm.Based on this,the optimization process of sd PSO algorithm to optimize QGNN is proposed,and the risk grade prediction model of coal and gas outburst of sd PSO-QGNN is established.Finally,the risk grade prediction model of coal and gas outburst based on sd PSO-QGNN is validated with the actual data of gas outburst,and the test results are compared with those of BP,IPSO-SVM and memetic-ELM,the prediction accuracy is 97.5%,82.5%,90% and 87.5% respectively,and the convergence time is 1.2974 s,1.9962 s,1.1123 s and 4.6742 s respectively.The results show that although the convergence time of sd PSO-QGNN is not the least,its prediction accuracy is the highest.In a controllable time range,the prediction of coal and gas outburst pays more attention to the prediction accuracy of the model.In conclusion,the prediction performance and generalization ability of the model are good,and it can accurately predict the risk level of gas outburst.