The Apply Of Genetic Algorithm In Stress Inversion

Initial geo-stress field is the basic information in geologic environment and crust stability evaluation, It not only decides the stability and regional stability of the rock,but influencing the engineering design and construction.so,for a long time,the character of initial geo-stress field around the engineering area should a focal point in engineering and academic circles research.The most straightforward way to provide the initial geo-stress field is geo-strss measurement.But,questions such as the limited measuring points,discreteness in measurement dates are existed,so,basing on actual survey geo-stress data,it is an effective way to inverse the initial geo-stress field by inversion analysis.on the basis of a large number of references to read at home and abroad,first summarizing the current status of the initial stress of the rock back analysis;then introduced the principle and method of the initial stress measurement, artificial Neural Networks and Genetic Algorithm in Inversion Analysis; at last, an instance of an engineering, the use of genetic algorithm optimization of neural network structure based on the finite element method, inversion of the initial stress.Moreever, in the calculation process,orthogonal test method was used in structuring samples,it is not only guaranteeing the net's precison but also reducing the number of sample; the traditional BP neural network inversion method is easy to fall into local minimum point and the shortcomings of slow convergence speed,in this paper,genetic algorithms combined with neural networks, has greatly enhanced the convergence speed of neural networks.At last,according to the comparision between calculation results and measured results,it proves that the inversion analysis method in this paper is feasible and reasonable;it is valuable in the similar projects.