Multi-objective Evolutionary Algorithm And Its Application In Testability Design

In the field of large-scale system fault diagnosis and test design problems,with the increasing system complexity,the requirements for test design are becoming more and more complicated.For example,the single-operation testability design evolves into a multi-operational mode test design,single target The optimization problem is promoted to multi-objective optimization problems and so on.How to design a test design with high efficiency and accuracy is a problem that needs to be solved.In this paper,a multi-objective evolutionary algorithm for grouping synthesis is designed to meet the needs of current test design.In the research field of multi-objective evolutionary algorithms,there are already many excellent algorithms.How to improve them to achieve better results is currently a hot topic.This article uses the features of MOEA/D specific direction search,combined with NSGA-III type Algorithm,designed a three-stage system-level multi-objective evolutionary algorithm.This paper first introduces the test design field and the research background and development status of multi-objective evolutionary algorithms.The difficulties and development bottlenecks of the current multi-objective evolutionary algorithms are introduced.For example,in the case of high-dimension multi-objective optimization problems,there are problems such as failure of dominant rules,high time complexity,and difficulty in population visualization.The importance and development status of the testability design problem of multi-operation mode in the field of test design are introduced.Multi-objective optimization problems and mathematical definitions of Pareto-related terms are given.According to whether the algorithm has an elite mechanism for classification,such as non-elite MEGA,NSGA,elite mechanism NSGA-II,SPEA,SPEA-II,PESA,and PESA-II.Among them,the NSGA-III type was highlighted,and its detailed pseudo code was given and analyzed.Finally,the algorithmic time complexity of the main steps is given.The relevant mathematic definitions of operational mode testability design and the practical significance of the relevant parameters are given.The steps to solve the problem using traditional multi-objective evolutionary algorithms are introduced.The group-integrated multi-objective evolutionary algorithm and its solution to the multi-operational mode test design problems are introduced.step.Finally,the experimental results of the two methods are given,and the results are compared and analyzed.The characteristics of MOEA/D and NSGA-III are analyzed in detail,and it is pointed out that MOEA/D has the characteristics that it can set a specific direction search,and the characteristics of iterative evolution of populations in NSGA-III.Try to set the initial population in the NSGA-III stage,use MOEA/D to set the search method for the specific direction,and set the optimal initial solution.Finally,conclusions are drawn through comparative analysis of experiments.This paper mainly studies the adaptation of the multi-objective evolutionary algorithm to the specific problem of multi-operating mode testability design,and improves NSGA-III.These studies provide the basis for algorithmic implementation in the future to implement the test design platform software.