NN-based GA for engineering optimization

For many engineering optimization problems, there are no explicitly known forms of objective functions in terms of design variables, or it only by complicated analysis or time-consuming simulation to obtain the performance of solution. Aiming at such kind of problems, this paper proposes a neural network (NN)-based genetic algorithm (GA), where the good approximation performance of NN and effective and robust evolutionary searching ability of GA are applied in hybrid sense. That is, NNs are employed in predicting the objective value, while GA is adopted in searching optimal designs based on the predicted performance. Simulation results and comparisons based on a well-known pressure vessel design problem demonstrate the feasibility and effectiveness of the strategy, and much better results are achieved than some existed literature results. In addition, the consistency and statistical quality of the resulted solutions can be improved by applying multiple neural networks.