جستجوی مقالات مرتبط با کلیدواژه « فروشنده دوره گرد » در نشریات گروه « برق »

This paper proposes a synergetic combination of chaotic neural network with self-feedback, Lyapunov exponent, and simulated annealing for combinatorial optimization problems such as travelling salesman problem (TSP). Unlike conventional neural networks only with point attractors, the chaotic neural network has more flexible dynamics, so that it can be expected to have higher ability of searching for optimal or near-optimal global solutions. One of the most important problems related to conventional neural networks is becomies trapped into the local minimums. Although chaotic neural networks can solve this problem, but they have difficulty due to convergence towards the equilibrium point. Therefore, we have tried to add the Lyapunov exponent and gradual cooling factor as a simulated annealing process, until network converges to the global optimal solutions quickly. In order to evaluate the proposed approach, TSP with different cities is used. Numerical experiments of the propsed approach on 10-TSP are shown that it has high efficiency to converge to global optimal solutions.