Effects of neural network procedures for improving genetic algorithms

摘要

In the combining strategy proposed by the authors of this paper, the model of multilayer neural network was applied to map the relationship between the parent individuals and the corresponding elite offspring, and was incorporated into the evolutionary procedure to improve genetic algorithms. This paper is involved in the investigation on the procedures of neural network applied in the combing strategy. The model of neural network for improving genetic algorithms is introduced briefly at first. The optimization problems of ten-design-variable function and seventy-two-bar structure are applied to investigate the impact of the patterns on the model of neural network then. The numerical experiment of the influence of the training procedure of the patterns on the model of neural network is made finally.