Reliability-Based Multidisciplinary Design Optimization Of High-Speed Train Gear Transmission System Under Random And Fuzzy Uncertainty

In the field of engineering,Reliability-Based Multidisciplinary Design Optimization （RBMDO）is an effective strategy for handling complex discipline optimization problems. This strategy fully considers the differences between disciplines within the system and the reliability requirements of the system,and maximizes the use of disciplines.The coupling effect between the two can be obtained to obtain the global optimal solution.At the same time,due to the common uncertainty in complex engineering,the influence of these uncertain factors needs to be considered in the Multidisciplinary Design Optimization（MDO）process.The main research results of this paper are as follows:（1）Through the generalized density function method based on information entropy, a unified representation of random and fuzzy uncertain information is established.Firstly, the uncertainty information is classified,and then in order to solve the optimization design problem under various uncertainty conditions,a generalized density function method ased on information entropy is selected to transform the fuzzy uncertain variables,and unified representation of uncertain variables are established.Finally,two cases are used o verify its effectiveness.（2）Reliability analysis and modeling are realized by first-order and second-order eliability approximation methods.For the reliability analysis in the multidisciplinary eliability optimization design under the influence of random and fuzzy uncertainties,a ariety of reliability analysis methods are used for verification,and the First Order eliability Method（FORM）and the Second Order Reliability Method（SORM）calculates he reliability of the discipline,and uses Monte Carlo Adaptive Importance Sampling（MCAIS）to calculate and compare the reliability,and uses simple random Monte Carlo ampling to simulate the reliability to verify the effectiveness of several reliability olutions.Under the influence of random and fuzzy uncertainties,the validity of eliability solutions are verified,which lay the foundation for the RBMDO.（3）The surrogate model is constructed with BP（Back Propagation）neural network, the feasibility of BP neural network as a surrogate model and its application in MDO are studied,and the Optimal Latin Hypercube Sampling（OLHS）is used to construct BP surrogate model.Under the influence of random-fuzzy uncertainty,the BP neural network urrogate model is used as the approximate function of discipline analysis,and RBMDOis carried out for the high-speed railway train gear transmission system under uncertain conditions.