Structural performance optimum design method under non-probability conditions

Abstract

The invention discloses a structural performance optimum design method under non-probability conditions. The method comprises following steps: an interval optimum design model considering structural reliability requirements is established; sample points are obtained by means of Latin hypercube sampling and collaborative simulation; a Kriging surrogate model for predicting objective functions and constraint functions is established; the interval optimum design model is solved by means of a double-layer nested genetic algorithm, wherein in the inner layer of the genetic algorithm, the left and right boundary of the interval values of the objective functions and the constraint functions are calculated, in the outer layer of the genetic algorithm, the interval reliability and reliability violation degree of each constraint function according to an unified formula are calculated to obtain the total reliability violation degree of a design vector, and the feasibility is determined; superiority ordering is performed on all design vectors according to a superior relation criterion based on the total reliability violation degree; the optimal solution of the interval optimum design model is output when maximum evolution algebra or convergence threshold is reached. By means of the method, structural performance optimum design method under non-probability conditions can be realized.

Description

technical field [0001] The invention relates to a structural performance optimization design method under non-probability conditions. technical background [0002] Uncertain factors, such as material properties, load environment and geometric dimensions, generally exist in the optimal design of engineering structures. Therefore, the structural optimization design process must consider the influence of uncertain factors. Optimal design based on probability model is one of the effective ways to deal with uncertain factors, and it has been fully researched in both method and application. However, probabilistic optimization design requires a large amount of sample data to obtain accurate probability distribution information about uncertain quantities. However, only very limited sample data can be obtained in engineering, and the probability optimization design may be sensitive to the distribution information of random parameters, that is, a small error in the probability model...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problem of uncertain structural optimization design that needs to consider structural performance reliability requirements in actual engineering, the present invention provides a structural performance optimization design method under non-probability conditions, and establishes an interval optimization model that considers structural performance reliability requirements. A unified formula for calculating interval reliability in constraint functions and a direct solution algorithm for interval optimization models are proposed, thereby avoiding the tedious process of converting interval optimization design models into deterministic models and the loss of information during the conversion process, and without introducing various Model conversion parameters greatly simplify the solution process, and obtain high reliability and high precision calculation results while maintaining high calculation efficiency

Images