结构可靠度分析的智能计算法

模糊数学、神经网络、遗传算法并称为智能计算方法.对功能函数没有明确表达式的问题进行可靠度分析,本文提出了基于响应面重构的智能计算法,即构造一个模糊神经网络响应面并采用遗传算法进行可靠度计算.算例分析表明,该方法与传统的二次多项式响应面方法相比,计算精度较好,可大大减少有限元分析次数.该方法用于大型复杂海洋工程结构的可靠度分析,可相应提高工作效率和解题质量,具有实际应用价值.

Intelligence Method for Structural Reliability Analysis

It is well known that the first-order second-moment method can be used when the performance function is known. When the performance function cannot be expressed exactly such as complex large-scale ocean engineering structure, in the Monte Carlo method with finite element analysis, random finite element method and the response surface method, the response surface method can be the best option because of its clear thought and simple programming. The response surface method can be implemented by using a series of the finite element calculations to fit a response surface instead of the unknown real limit state surface, and the existent finite element method program can be applied to the reliability analysis directly. Fuzzy theory, neural network and genetic algorithm are altogether called intelligent computing methods. Intelligent computing method based on response surface restructure is put forward in this paper, which can be used when the performance function is unknown in the structural reliability analysis. In this method, a response surface of fuzzy neural network is made, and then the structural reliability is calculated by genetic algorithm. Through the real examples and contrast analysis, it can be known that this method is much better than the response surface method of quadratic polynomials, by which the quantity of calculation of finite element analysis is reduced a lot, and the accuracy of calculation is increased. If this method is used for the complex large-scale ocean engineering structure reliability analysis, the efficiency and quality of calculation can be greatly improved. So, it is suitable for the engineering application.