Abstract:

To solve the problem of slow convergent speed of ACA (Ant Colony Algorithm), an improved method is proposed. One kind of convergence factor is a
dded to ACA to guarantee that the best route produced in each iteration would be given additional pheromone increment during the pheromone global updating procedure. Therefore, the possibility of algorithm trapped in local optimal solution is reduced. The convergence of the improved ACA is proved and its optimizing ability is tested. The simulation results show that the improved ACA has higher optimizing ability and faster convergent speed in contrast with the basic ACA. The improved ACA is used to optimize neural network, and the optimized neural network is applied in the diagnosis of transformer fault. The results show that the optimized neural network based on the improved ACA has higher convergent speed and diagnostic accuracy in contrast with BP neural network.

Key words: ant colony algorithm, neural network, transformer, fault diagnosis