可靠性分析新方法研究与应用

【作者】 杨子政；

【导师】 吕震宙；

【作者基本信息】 西北工业大学 ， 人机与环境工程， 2006， 硕士

【摘要】 对于工程常见的非线性程度较高的隐式极限状态方程,经典响应面法由于自身固有的缺点,无法快速得到足够精确的计算结果。为此,本文研究了基于神经网络技术和Kriging技术的可靠性分析方法,并提出相应的改进新方法,最后将这些方法应用于复合材料可靠性分析中。 第一章为文献综述和选题背景。首先简要介绍了结构可靠性分析方法,回顾快速概率积分法、Monte Carlo法和响应面法的基本思想。然后追溯了神经网络技术的发展,说明令其可以用于结构可靠性分析的相对优势。随后对复合材料可靠性研究现状进行总结,指出将神经网络等技术应用于复合材料可靠性分析中正成为值得挖掘的研究方向之一。最后介绍了本文的主要研究内容。 第二章对大部分工程可靠性分析中的隐式极限状态方程,建立了一种基于样本筛选方案的神经网络的可靠性分析方法,并提出了一种基于加权线性响应面的神经网络可靠性分析法。与已有的神经网络可靠性分析方法相比,提高了可靠性分析的精度。 第三章引入采用灵活、自适应的响应函数形式的Kriging技术,建立基于该技术的结构可靠性分析方法,基于前一章中提出的采样策略,提出了改进的基于Kriging的结构可靠性分析方法,并以算例证明其可行性和合理性。 第四章对于利用结构标准有限元程序寻找结构可靠性分析中最可能失效点的方法进行了验证,并将其应用于工程可靠性分析问题中。 第五章将第二、三章提出的可靠性分析方法应用到复合材料层压结构的可靠性分中去,编制了与有限元软件相连接的神经网络和Kriging复合材料层压结构可靠性分析程序,为复合材料结构可靠性分析提供了新的分析方法。更多还原

【Abstract】 In order to predict the failure probability of a complicated structure, the structural responses usually need to be estimated by a numerical procedure, such as finite element method. To reduce the computational effort required for reliability analysis, response surface method could be used. However the inflexible function form affects the fitting precision of the conventional response surface method. In this thesis, a new sampling strategy is presented, based on which a new artificial neural network (ANN)-based reliability analysis method and an advanced structural reliability analysis method incorporating Kriging technique are proposed and approved firstly. Then both of them are successfully used in the reliability analysis of composite structure. Also, searching for the most probable failure point directly by use of ANSYS optimization module is applied in engineering structure validly.In chapter one, the substance of four kinds of reliability analysis methods commonly used in structural safety are reviewed. Development of artificial neural network (ANN) is surveyed. Recent advancements in composite structural reliability analysis are summarized. Finally, the main work in this thesis is introduced.For implicit limit state equations in most engineering reliability analysis, a new method is presented on the basis of artificial neural network (ANN), where the training samples are appropriately selected, and an ANN reliability method based on the weight linear response surface method is also presented in chapter two.In chapter three, Kriging technique characterized by flexible and adaptive function form, in conjunction with selection strategy of sampling points, is employed to develop a new way for the reliability analysis of the implicit performance problems.The calculation of the most probable failure point can be translated into an optimization problem by introducing a modified joint probability density function (MJPDF) and searching trouble is solved directly through a built-in optimization module of ANSYS. In chapter four, after this method is validated, it is used to analyzethe reliability of the engineering structure.In chapter five, all methods presented in chapter two and three are applied in the reliability analysis of composite laminated structure under load of different level, and the method is programmed with the standard finite element software. It is shown that the results are suitable and reasonable.更多还原