Research on Multi-objective Parameter Optimization Based on the Experimental Design and ANN-GA in the Digital Environment

Aiming at the multi-objective parameter optimization problem of black box system in the digital design/simulation environment, a multi-objective parameter optimization model based on the orthogonal design/uniform design and artificial neural network-genetic algorithms ANN-GA is established. In this method, the principle of experimental design is used to arrange a test or virtual test project. The data modifying and data collecting in the virtual test are accomplished applying the characteristic of parameterization in the environment of digital simulation. At last, the neural network and Pareto genetic algorithm are adopted to optimize multi-objective parameters. A Pareto-optimal set of digital model can be found in specified region.