Structural Reliability Analysis Based On Artificial Neural Network

The existing structural reliability calculation methods are often used for reliability analysis of structural members.When it is necessary to perform reliability analysis on the entire structure or reliability analysis of large-scale structures,it is often difficult to express the limit state function of the structure.In most structure dependability are analysis,the limit state function solved asked is often non-linear,involving a large number of random variables not confirmed,it is analysis that this kind of function is complicated course consuming time even more In addition among some actual project that reliability analysis sometimes at all unable to is it produce intact limit state function to set up,the necessary data can only be got through the experiment method.To such a situation,in this paper is present a new method based on the artificial neural network and the traditional reliability analysis method.The main research contents are as follows:(1)By using a single-layer rigid frame structure with artificial neural network based on First-order reliability method(FORM)and third-order moment method based on artificial neural network Reliability analysis of the structure,in order to verify the accuracy of various methods,at the same time using Monte Carlo sampling algorithm for reliability analysis of the structure.(2)After verifying the effectiveness of the method proposed in this paper through several small calculation examples,the reliability analysis method based on the artificial neural network FORM method is used for a six-layer frame structure,and the third-order moment based on the artificial neural network Method and Monte Carlo sampling algorithm for structural static seismic reliability analysis.(3)Then,an 18-story frame cylinder structure is analyzed by ANSYS modeling to obtain the response of the structure under different random variables.Then use the method based on artificial neural network to analyze the structural dynamic seismic reliability.Through the above research,it is found that the calculation efficiency of seismic reliability analysis based on artificial neural network algorithm is higher than that of Monte Carlo sampling algorithm,and the calculation accuracy is higher than the thirdorder moment method.It can also meet engineering needs.