On-line identification of synchronous generator using Self Recurrent Wavelet Neural Networks via Adaptive Learning Rates

In this paper, the Self-Recurrent Wavelet Neural Network (SRWNN) is used as a model predictor for identify a synchronous generator. Further, a hybrid algorithm combining Chaotic Global Search (CGS) algorithm with Back-Propagation (BP) algorithm, referred to as CGS-BP algorithm, is proposed to train the weights of SRWNN-Identifier (SRWNNI). And also, the gradient-descent method using Adaptive Learning Rates (ALRs) is applied to train all weights of the SRWNNI, in on-line mode. The ALRs are derived from discrete Lyapunov stability theorem. Finally, the proposed SRWNNI are evaluated on a single machine infinite bus power system under different operating conditions and disturbances to demonstrate their effectiveness and robustness. Also, the SRWNNI is compared with Wavelet Neural Network Identifier (WNNI) and Multi-Layer Perceptron Identifier (MLPI).