3D architecture facade optimization based on genetic algorithm and neural network

The design of 3D scene should follow the rules of architecture's organization. At present, the 3D scene design are usually carried out by art designer who lack the knowledge of architecture. A method is proposed in this paper to solve the problem. We improved the interactive genetic algorithm to obtain the best adaptive features, and combined the ART1 network to simulate the behavior of users to evaluate the individuals. This method can solve the problem that manual evaluation operation may cause errors by the user's fatigue during the evaluation process, and it can increase the number of generations to obtain more information. We improved the ART1 net work based on the principle of experimental psychology to simulate the hierarchical structure of human brain's memory mode, and increased the memory capacity and compute efficiency. In this way, we can obtain the more accuracy adaptive values of 3D facade's features and improved the3D architecture facade's evolution process. This method can reduce the tedious work in art design, and effectively guide the design of 3D scene scheme. The disadvantage of this method is that it is not do enough works to deeply explore the relationship between some kinds of implicit aesthetic indexes in aesthetic personality, and lack of successful exploring in establishing a reasonable approximate model to express the implicit aesthetic characteristics. In future we should study more deeply and solve the problems above. This method is suitable for batching optimizing the 3D architecture facade, and has a positive meaning for 3D architectural design, landsape design, 3D game scene design and virtual reality.