Combined Training of Recurrent Neural Networks with Particle Swarm Optimization and Backpropagation Algorithms for Impedance Identification
This document has been relocated to http://scholarsmine.mst.edu/ele_comeng_facwork/925
There were 23 downloads as of 27 Jun 2016.
A recurrent neural network (RNN) trained with a combination of particle swarm optimization (PSO) and backpropagation (BP) algorithms is proposed in this paper. The network is used as a dynamic system modeling tool to identify the frequency-dependent impedances of power electronic systems such as rectifiers, inverters, and DC-DC converters. As a category of supervised learning methods, the various backpropagation training algorithms developed for recurrent neural networks use gradient descent information to guide their search for optimal weights solutions that minimize the output errors. While they prove to be very robust and effective in training many types of network structures, they suffer from some serious drawbacks such as slow convergence and being trapped at local minima. In this paper, a modified particle swarm optimization technique is used in combination with the backpropagation algorithm to traverse in a much larger search space for the optimal solution. The combined method preserves the advantages of both techniques and avoids their drawbacks. The method is implemented to train a RNN that successfully identifies the impedance characteristics of a three-phase inverter system. The performance of the proposed method is compared to those of both BP and PSO when used separately to solve the problem, demonstrating its superiority