Improved Multidisciplinary Design Optimization Based on Genetic Algorithm and Artificial Neural Networks and Its Application

摘要

Multidisciplinary Design Optimization (MDO) is an algorithm widely used in the engineering field currently. However, traditional MDO often leads to the failure of convergence or local optimum problems caused by convergence. In such cases, a multidisciplinary design optimization basedon genetic algorithm (GA) and artificial neural networks (ANN) (GA-ANN-MDO) is presented in the paper. Under the thought of parallel distribution of traditional MDO, the real sub-disciplinary model is replaced by a highly precise ANN model dependent on the Latin Hypercube experimental designmethod in the GA-ANN-MDO, so as to reduce the computational cost and smooth the value noise. The GA optimization system level is applied to decline the possibility of partial solution involved in the optimization. As shown from the optimization results of two classic mathematical examples,GA-ANN-MDO is presented good robustness, which could quickly and effectively converge to the global optimal solution. In addition, a project example was employed finally to verify the feasibility of GA-ANN-MDO in the engineering.