Inversion Of Rock Mass Parameters Of Mining Subsidence Based On BP Neural Network

In the research of coal mining subsidence,numerical simulation technology has become one of the most important research methods in the current mining subsidence prediction,which provides new methods and approaches for the study of mining subsidence.In the numerical simulation analysis,the reasonable selection of rock mechanics parameters is a key element of numerical simulation.The rock mass parameters obtained in the indoor test or on-site local test have certain limitations,which makes the numerical simulation results have a certain gap with the actual situation,and it is difficult to accurately reflect,the objective law of coal mining subsidence.In this paper,taking a working face in the Sandaogou mining area as an example,the inversion of the parameters of the mining subsidence rock mass in the mining area is studied.The BP neural network was used to establish the model between the subsidence value of the mining area and the rock mass parameters,and the rock mass parameters were inversely analyzed.For the selection of inversion parameters,using numerical simulation and grey correlation analysis,it is concluded that the parameter with higher sensitivity to the surface subsidence value in the mining area is the elastic modulus,and it is used as the inversion object.The orthogonal design principle combined with the numerical simulation method is used to obtain the training data needed to establish the BP neural network model.The built-in parameters of the network are reasonably set,and the influence of different learning algorithms on the training results of the network model is compared and analyzed.The genetic algorithm is used to optimize the network model,which improves the accuracy of the model prediction parameters.The experimental results show that:the optimized network model has a significant improvement on the accuracy of the prediction parameters.The main distribution range of relative error is reduced from 11%to 17%to between 4%and 8%.Based on the constructed network model and the measured sinking data of the mining area,the inversion of the elastic modulus of the rock mass parameters in the mining area is completed,and the inversion results are verified by numerical simulation.The experimental results show that the inversion of the rock mass parameter elastic modulus results in less error between the numerical simulation results and the measured values,and the relative errors are within 10%.It can meet the actual needs of the project and accurately reflect the objective laws of mining subsidence.