Evolutionary Diagonal Recurrent Neural Network for Nonlinear Dynamic System Identification
Mu Yuqiang, Sheng Andong, Guo Zhi · 2008
Conventional training methods for diagonal recurrent neural network identifier are limited because its structure is fixed by previous experiences and the weights are local optimal. In this paper, a novel identifier based on evolutionary diagonal recurrent neural network (EDRNN) is proposed. Compared with conventional methods, it has prominent advantage in identifying nonlinear dynamic systems because the structure and weight of EDRNN can be evolved simultaneously. Experimental results with the classical nonlinear systems confirm that EDRNN-based method is a promising tool for identifier.