摘要
为提高传统可靠性分析算法求解工程问题的计算效率,将可靠性优化中的双循环法与通用生成函数相结合,建立优化的数学模型进行可靠性分析,并通过k-means聚类和同类项合并缩减计算成本,提高效率.计算结果表明:当优化模型的极限功能函数为线性方程或非线性方程时,通用生成函数的误差分别为6.1%和10%,与一次二阶矩法和二次二阶矩法相比,其在精确度上的优势随着极限功能函数的非线性程度增加而更突出,计算效率明显高于蒙特卡罗法.在随机变量增加且不服从正态分布的工程实例中,基于通用生成函数的可靠性优化方法也比传统的可靠性优化方法精度高、效率快、普适性强.
The double cycle method in reliability optimization is combined with the universal generating function to establish the optimized mathematical model to solve the engineering problems instead of the traditional reliability analysis algorithm.In reliability analysis,k-means clustering and similar item combination are used to reduce the calculation cost and improve the efficiency.The calculation results show that when the limit functions of the optimization model are linear equation and nonlinear equation,the error of the universal generating function is 6.1%and 10%,respectively.Compared with the first-order second-order moment method and the second-order moment method,the accuracy advantage of the universal generating function is more prominent with the increase of the nonlinear degree of the limit function,and the calculation efficiency is significantly higher than that of Monte Carlo method.In the case of increasing random variables and not obeying normal distribution,the reliability optimization method based on universal generating function is more accurate,efficient and universal than the traditional reliability optimization method.
作者
蒋国盛
周金宇
朱达伟
庄百亮
JIANG Guosheng;ZHOU Jinyu;ZHU Dawei;ZHUANG Bailiang(College of Mechanical Engineering,Jiangsu Institute of Technology,Changzhou 213001,China;College of Mechanical and Electrical Engineering,Jinling University of Technology,Nanjing 211169,China;Jiangsu Institution,China Academy of Mechinery Science and Technology,Changzhou 213001,China)
出处
《扬州大学学报（自然科学版）》
CAS
北大核心
2020年第1期16-20,共5页
Journal of Yangzhou University：Natural Science Edition
基金
国家工业和信息化部重大专项资助项目(2018ZX04026001-008)
江苏省教育厅高校自然科学研究资助项目(16KJA460002).
关键词
通用生成函数
可靠性优化
K-MEANS聚类
universal generating function
reliability optimization
k-means clustering
