Hybrid ANNs-GAs Strategy Based Constrained Optimization and Application in Sheet Metal Flanging Forming

Abstract

In order to provide a general purpose method to search optimum solution for complex constrained engineering problems without explicit system model, a hybrid optimization strategy based on artificial neural networks (ANNs) and genetic algorithms (GAs) is proposed in this paper. This strategy combines the strong nonlinearity mapping abilities of ANNs and effective and robust evolutionary searching ability of GAs. Firstly, ANNs are utilized to model the un-known system using inputs and outputs of system. Then the direct comparison approach based improved GAs are employed to search optimal solution in the constrained design space, using the trained ANNs as the function generator of system outputs. This strategy is implemented in optimization of design variables for sheet metal flanging process. The verification results of numerical simulation and the experiments demonstrate the feasibility and effectiveness of the strategy.