An adaptive neural networks model for isomorphism discernment of large-scale kinematic structure

Graphs isomorphism discernment is one of the most important and difficult issues in graphs theory based structures design. To solve the problem, a Hopfield Neural Networks (HNN) model is presented. The solution of HNN is design as a permutation matrix of two graphs, and some operators are improved to prevent premature convergence. The convergence properties of HNN model and the improved HNN model are studied by analysing the search process. The computation times of the HNN model will not be affected greatly by enhancing the number of nodes in graph and the algorithm is efficient for large-scale graphs isomorphism problem.