Parameter Back Analysis Of Rock Mass Based On Neural Network And Genetic Algorithm

In the solution of Geotechnical, it is intractable to ascertain parameter of rock and soil exactly. Back analysis is a good method to get parameter of rock. Some research of method of parameter back analysis is made.Because direct method, one of many kinds of back analysis methods, has broad applicability, direct method is adopted in this text.Constitutive model is very important for parameter back analysis. A fitting constitutive model can make it easily to ascertain parameter. Elastic constitutive and elastic-plastic constitutive are introduced in this text.Calculation efficiency will be improved greatly if we substitute Artificial Neural Networks for Finite Element Method. BP Neural Networks is adopted to constitute mapping of rock parameter and rock displacement and calculate rock displacement in back analysis of rock parameter. Orthogonal experimental design is adopted to constitute study examples. Accuracy of forecast of net is ensured and time of test is reduced. Genetic Algorithm is used to search most appropriate neural network structure. The result is good. Uniform design method is used to constitute test examples. Back test is used to test the result. The result proves that forecast capability of Neural Networks is good enough.After the mapping of rock parameter and rock displacement is constituted by neural network, Genetic Algorithm is used to search optimum rock parameter. The result is good.Back analysis program is compiled based on Neural Network and Genetic Algorithm. Only constitute General and correct study examples with correct forward analysis program, all kinds of mapping of relation of input and output can be got by Neural Network, so the program in this text is all-purpose in some area.The back analysis program is used to calculate rock parameters of an engineering example. The result is good and correct. The error of displacement based on the parameter gained by the back analysis program is little.