Research And Application Of Evolutionary Structural Optimization Method Based On Control For Evolutionary Step Length

Evolutionary structural optimization(ESO) method is one of the most important structural topology optimization methods. ESO method has a simple basic idea, no middle density in the process of optimization, and a clear the topology form, which is suitable for engineering application, but is easily fall into local optimal. Therefore, in order to improve the ability of global optimization for ESO method, Control model of evolutionary step length of element density and an improved ESO method are developed analyzing evolutionary step length of element density. The improved ESO method is based on bi-directional evolutionary structural optimization method, sensitivity redistribution technology and judgment criteria of singular element. The application of the improved ESO method is also investigated.The main contents are given as following:1. Determination for control model of evolutionary step length of element density. All units are divided to the main units, the secondary units and general units according to element density of each unit impacting on structure performance, and the computational method of units’ weight coefficient is determined. Control requirement of evolutionary step length of element density is analyzed according to the influence rule of units on structure performance, and some kinds of control models of evolutionary step length of element density are builted in line with the requirement to control the rate of elements being deleted. A control model of evolutionary step length of element density has better comprehensive performance and is identified through analyzing and comparing structure optimization results of several control models in two numerical examples.2. Development of evolutionary structural optimization method based on control for evolutionary step length of element density. To avoid the failure of optimization in ESO caused by the mistaken delete units, the operation for add element of bi-directional evolutionary structural optimization method is proposed. Checkerboard pattern is controlled by using sensitivity redistribution technology and can obtain smooth optimal structure. Deleting some units may lead to the problem of structural compliance mutation in structure optimization. To improve the ability of global optimization for ESO, mistaken deleted units are restored by bring in judgment criteria of singular elements at the early stage of the optimization. Then, the effectiveness of presented method is verified by numerical examples.3. Realization of the prototype system for evolutionary structural optimization method based on control for evolutionary step length of element density. The prototype system for evolutionary structural optimization method based on control for evolutionary step length of element density is carried out with the platform of Matlab. Engineering practicability of the presented method is demonstrated by using the prototype system to optimize bearing isolator and spring lug isolator in a company is studied.