A Comprehensive Study on Fitness Approximation Techniques in Shape Optimization of Aerofoil Forging Preform Tools

Abstract

In the process of complex engineering designs or optimizations, a large number of physical experiments or numerical simulations are required to evaluate certain performance qualities before a satisfactory result can be obtained. In both cases, constructing an approximate model is often necessary to provide a reliable response as an alternative to experiments or simulations. In this paper, three types of approximation models were developed and applied in a shape design of an aerofoil forging preform tool. Their modeling techniques are presented in detail. An optimal Latin hypercube technique was employed for the design of the experiment and sampling with the expected coverage of parameter space. Finite element (FE) simulations of multistep forging processes were implemented to acquire the objective function values for evaluating the forging performance. By a parametric study, the effects of design variables on objective responses and correlations were investigated for a clear insight into their functional nature. Comprehensive analyses and comparisons between different approximate models have been carried out. Finally, an optimization design of a preform tool was successfully achieved based on a particle swarm (PSO) algorithm combined with the proposed approximate model.