Study For Multi-Disciplinary Time-Dependent Reliability Design Optimization Methods Based On Surrogate Model

Multidisciplinary Design Optimization(MDO)is a method system for design optimization of large-scale complex systems.The conventional MDO is a deterministic method.However the time-varying uncertainties in real projects can threaten the performance of complex mechanical systems.In addition,there are often complex coupling characteristics in MDO,which make the calculation of time-dependent uncertainties complicated in complex mechanical systems.Time-dependent MDO generally includes two aspects of time-dependent reliability analysis and design optimization.Because the surrogate model method has advantages in reducing computation,this paper uses the agent model method to analyze the time-dependent reliability.Because of the multidisciplinary design optimization involved,this paper has also done some preliminary research on the MDO optimization algorithm.The main content of this article is:(1)In view of the complexity of the MDO problem,the conventional method is not easy to quickly locate the target function.This paper proposes an MDO method based on the improved Forbearing Stratified Sequencing Method.This method regards the MDO system optimization layer as the main optimization goal,the subject coupling coordination as the secondary optimization goal,and then under the binary search frame,it searches for the optimal objective function value satisfying the coupling relation stably and efficiently.In this paper,the flow of solving the improved cascading sequence method is given in detail,and the effectiveness of the proposed method is verified by using a mathematical example and a heart dipole example.(2)Due to the complexity of the time-dependent uncertainty factors,the efficiency of conventional time-dependent reliability analysis is relatively low.This paper uses time-dependent response extreme value method to transform the time-dependent reliability into the traditional reliability analysis problem with the function as implicit.Then,a time-dependent reliability analysis method is established using the quadratic polynomial response surface.The quadratic response surface method calculates the time-varying reliability by several times approximately establishing a local high-precision quadratic polynomial surrogate model,which can effectively improve the efficiency of time-dependent reliability analysis.This paper verifies the validity of the method through two engineering examples.(3)Since the multi-disciplinary framework includes the coupling characteristics,the time-varying uncertainty factors will be propagated to the various sub-disciplines along with the coupling relationship,thus making it difficult to perform time-dependent reliability analysis in a multi-disciplinary framework.Firstly,this paper analyzes the propagation characteristics of a single sample and generalizes it to obtain the propagation characteristics of the entire time-dependent system.Then,based on the decoupling strategies of traditional disciplines,this paper proposes a dual-cycle analysis mechanism to search for functional functions containing state functions.The time-dependent response exteme value,combined with the method of the surrogate model,establishes the time-dependent reliability analysis and calculation model under the multi-disciplinary framework,and realizes the time-dependent reliability analysis under the multi-disciplinary framework.This paper verifies the validity of the method with a numerical example with obvious coupling characteristics.(4)In order to improve the performance of complex mechanical systems under the influence of time-dependent uncertainties,multidisciplinary time-dependent reliability design optimization must be performed for complex mechanical systems.Because the inverse reliability analysis method can effectively improve the efficiency of reliability design optimization,based on the Kriging model,the response model of the target constraint response extreme value is established,and the inverse reliability analysis of the time-dependent function function is implemented.Then,Based on the Sequential Optimization and Reliability Assessment method,a framework for solving multi-disciplinary time-dependent reliability design optimization problems is established.In the end,the validity of the method is verified by a mathematical example and a reducer design example.