摘要
对功能函数不能明确表达的问题进行可靠度分析,常采用响应面法。其中二次多项式响应面法应用较为广泛,采用与此方法相同的思路,提出了RBF神经网络响应面法,并通过算例与常用的BP神经网络响应面法进行了对比分析,该方法在学习速度、迭代次数等方面均优于BP神经网络响应面法。该方法用于大型复杂结构的可靠性分析,可相应提高工作效率和解题质量,具有一定实际应用价值。
The response surface method for structural reliability analysis can be used when performance function is unknown, in which the response surface method of quadratic polynomials is often used. In the same way, a response surface method of RBF artificial neural network(ANN) is put forward. By the contrast analysis, this method is better than the response surface of BP ANN in learning speed and iterative degree. If this method is used in the complex large-scale structure reliability analysis, the efficiency and quality of calculation can be greatly improved. So, it is suitable for engineering application.
出处
《海洋工程》
CSCD
北大核心
2004年第3期81-85,共5页
The Ocean Engineering
基金
国家自然科学基金重点资助项目(59895410)
辽宁省自然科学基金资助项目(002067)
