基于神经网络的闸阀模糊可靠度计算

Fuzzy reliability calculation of gate valve based on neural network

为提高闸阀可靠度计算准确性,将模糊准则引入闸阀可靠性设计,将闸阀应力与阀体通径、内压之间隐式关系显式化,并基于神经网络构建地震工况下闸阀数学模型;利用蒙特卡洛法,计算闸阀在常规方法下的可靠度;利用正态型隶属函数描述闸阀强度的模糊性,计算闸阀在考虑模糊强度时的可靠度;对比强度的模糊性对可靠度计算结果的影响,分析不同隶属函数下的闸阀模糊可靠度。与传统随机可靠度计算方法相比,该方法对不确定性描述更合理,理论上对可靠度的计算更加准确。

In order to improve the accuracy of the reliability of gate valves,by introducing the fuzzy criteria into the reliability design of gate valve,based on the neural network modeling,the implicit relationship between the gate valve stress and the diameter and internal pressure of valve body was explicit,and a mathematical model of gate valve under the seismic conditions was established.The reliability of gate valve under the conventional method was calculated by Monte Carlo method.The normal membership function was used to describe the fuzziness of the gate valve strength,and the reliability of gate valve when considering the fuzzy strength was calculated.The influence of the strength fuzziness on the calculation results of reliability was compared,and the fuzzy reliability of gate valves under different membership functions was analyzed.Compared with the traditional random reliability calculation method,this method has a more reasonable description of uncertainty,and it can calculate the reliability more accurately in theory.