基于PSO优化神经网络响应面技术的非能动系统可靠性分析  

Reliability Analysis of Passive System Based on PSO Optimized Neural Network Response Surface Method

作　　者：丁浩 蔡琦[1] 张永发[1] 蒋立志[1] 魏柯 Ding Hao;Cai Qi;Zhang Yongfa;Jiang Lizhi;Wei ke(Department of Nuclear Science and Engineering,Naval University of Engineering,Wuhan,430033,China)

机构地区：[1]海军工程大学核能科学与工程系,武汉430033

出　　处：《核动力工程》2018年第4期101-106,共6页Nuclear Power Engineering

摘　　要：在非能动可靠性分析数学模型的基础上,结合某型核动力装置非能动余热排出系统原理性试验装置和改进的热工水力程序的运行数据,识别了输入参数的不确定性,比较了不同神经网络响应面技术替代热工水力程序的精度和优度,分析了粒子群优化算法(PSO)优化神经网络响应面分类准确率。数值结果表明,该响应面具有较高的拟合优度,且能够较为准确的对非能动系统系统可靠性进行判定。On the basis of reliability analysis mathematical model, combined with the operating data from an experimental facility and improved thermal-hydraulic codes, the uncertainty of input parameters is identified. Compared with the accuracy and the goodness of different Neural Network Response Surface methods, the one optimized with PSO is analyzed by classification accuracy. The results show that PSO response surface has relatively better fitting performance and can evaluate the reliability of the passive system accurately.

关 键 词：非能动系统 响应面 神经网络 PSO 

分 类 号：TL38[核科学技术—核技术及应用]