Reliability And Sensitivity Analysis Of The Complex Electromechanical System With Uncertainty

Modern industrial equipment highly integrated with sound,light,electricity,liquid,and other technologies,has sophisticated structure and complex functions.In order to ensure the high reliability and long lifetime of the equipment,the reliability analysis of the system plays a huge role.Reliability analysis is an important method for evaluating the equipment to complete the predetermined function within the specified time,and it characterizes the life characteristics of the product,such as system reliability,average lifetime,and failure rate.In addition,due to the lack of long-term,large-scale statistical tests,there will be huge uncertainty during assessment.Effectively quantifying the uncertainty will directly affect the reliability evaluation results.There are plenty of components in a large-scale system,but not the uncertainty of every component has the same weight impact on the system uncertainty.Therefore,finding the components with a higher degree of influence on the system uncertainty will also be useful in the maintenance or design process.The main research content of this article includes the following parts.The reliability and sensitivity analysis under hybrid uncertainty is completed via the Double Loop Monte Carlo Simulation based on Bayesian network and pinching method.The time and parameter domains are traversed in layers to calculate the reliability and discrete sensitivity under hybrid uncertainty.In this paper,the hybrid uncertainty is visualized by probability box,and the uncertainty propagation is optimized by the genetic algorithm.Through numerical and realistic case,the method successfully found the component that have a greatest impact on system uncertainty and accurately quantifies the degree of impact,which verifies its feasibility in engineering.Bayesian network is a classic method of reliability derivation,but the reasoning process is difficult to separate the system structure and the prior probability.But the Survival Signature can effectively solve this problem.However,Survival Signature is rarely used in multi-state systems.Therefore,this article extend the traditional Survival Signature method,implementing the method from the two-state system to the multi-state system,and derives the state probability distribution by homogeneous Markov model modeling.This article proposed a reliability evaluation method of multi-state systems,and the sensitivity quantification is completed by the Double Loop Monte Carlo simulation.Finally,the method is applied to the numerical example of bridge structure system.The results show that the method is feasible for the system with regular state transfer rules.Finally,the two proposed reliability and sensitivity analysis methods under hybrid uncertainty are applied to the feed control system of XKA28,which is divided into W,X,Y,and Z axis feed control subsystems.According to the number of components of the same type of the subsystem,the W and X axis subsystems are modeled by multi-state Survival Signature method and Bayesian network is used to complete the reliability modeling of Y and Z subsystems.The whole system is modeled by the Survival Signature,and then the uncertainty propagation analysis and sensitivity analysis are completed by the above method.