Reliability Analysis of Cracked Structure Based on Evolutionary Neural Network and Monte-Carlo

摘要

These main influence factors of three dimensional crack extension lives are analyzed to obtain their statistical distribution types. And we have established the structure functional function and reliability model. Making use of evolutionary neural network technology, which integrates the advantage of genetic algorithms and neural network, we simulate the non-linear structure functional function, and then make a reliability analysis by the importance sampling. Taking the surface crack extension reliability simulation of the torsion shaft as numerical illustration, we obtain the crack extension lives for different given reliability, and analyze the influence of these random variables on the crack extension residual lives. The simulation results are in good accord with the real situation.