Optimization design method for robustness measurement and equilibrium for different kinds of multi-target performances of mechanical structure

Abstract

The invention discloses an optimization design method for robustness measurement and equilibrium for different kinds of multi-target performances of a mechanical structure. The method comprises the following steps: considering uncertain factors within an interval, establishing a robust optimization design model containing different kinds of target performances such as a cost type, a fixed type, anincome type, and a deviation type for the mechanical structure; nesting an inner layer of a genetic algorithm into double layers and calculating left and right boundaries of all structures of performance indices; performing feasibility discrimination on design vectors in an outer layer of the genetic algorithm; classifying robust equilibrium of a feasible solution on the basis of a robust equilibrium coefficient of an interval constraint performance and a robust equilibrium coefficient of an interval different-kind target performance, sequencing design vectors on the basis of the overall distance of different-kind target performance robustness. Therefore, robust equilibrium optimization of different kinds of multi-target performances for the mechanical structure is achieved. With the method, unified modeling is achieved in order to resolve a robust optimization problem of different kinds of multi-target performances. High robustness level of the constraint performance is ensured. Theoptimization result is objective. All kinds of structures are robust and balanced in overall performances.

Description

technical field [0001] The invention belongs to the field of mechanical structure optimization design, and relates to a robustness measurement and equilibrium optimization design method for heterogeneous multi-objective performance of mechanical structures. Background technique [0002] As the performance requirements of complex equipment continue to increase, designers need to consider more and more performance indicators when optimizing the structure of key components. Due to the influence of uncertain factors and the requirements of robust design, the mean value and fluctuation of structural performance indicators need to be considered during the optimization process, which increases the number of objectives and constraints to be optimized, and these objectives and constraints often conflict and restrict each other. . However, domestic and foreign scholars usually only consider the robustness of single-objective multi-constraint performance in the research on the multi-p...

AI Technical Summary

Problems solved by technology

However, domestic and foreign scholars usually only consider the robustness of single-objective multi-constraint performance in the research on the multi-performance robust optimization design of mechanical structures using interval numbers to represent uncertain factors, and do not consider the robust optimization problem of multi-objective performance balance. Guarantee the overall robustness of mechanical structure performance

[0003] A large number of existing studies have shown that for the multi-objective optimization design of mechanical structure performance, the processing method is often to transform the multi-objective optimization problem into a single-objective optimization problem by taking regularization factors and weighting factors. The selection of regularization factors and weighting factors requires a lot of experience , and its different values ​​will lead to different optimization results, with great uncertainty

At the same time, the existing multi-objective optimization research is mainly the optimization form of "minimization", "maximization" or the combination of the two, but there are usually four kinds of heterogeneous objective performance in actual engineering, including cost type, fixed type, income type and deviation type. type, the existing research lacks a unified modeling method for the performance of these four types of objectives and a solution algorithm for the corresponding heterogeneous multi-objective models

Images