科研之友
微信
新浪微博
Facebook
分享链接
摘要
The graph theory is an important method to achieve conceptual design for mechanism. During the process of kinematic structures enumeration using graph theory, isomorphism identification of graphs is a NP-complete problem. It is very important to improve isomorphism identification efficiency and reliability. To solve the problem, a genetic algorithm model and a Hopfield neural networks model are presented respectively. In the meantime, some improved operators are proposed to prevent premature convergence. By comparing the two models used in graph isomorphism problem base on simulation, the advantages and limitations of the two approaches are discussed. A complex model based on Neural-Genetic algorithm is proposed. Numerical experiments demonstrate the performance of the complex algorithm is more successful compared with the approach applying genetic algorithm or Hopfield neural network simply. Some examples confirm the validity of the model. The work builds a reliable isomorphism identification algorithm for intelligent CAD of mechanism.
