Current protection based on SPDS neural network
Xu Zi · Guangdong Electric Power · 2002
To improve the study efficiency and reduce the training time, a new method of current protection based on SPDS Neural Network is put forward. The network is a three layer SPDS neural network model and consists of three parts: the subsidiary network ANN1 of identifying the faults type and faults phase, the subsidiary network ANN2 of identifying the faults direction and the subsidiary network ANN3 of distinguishing swing and faults. The model test results for all types of faults showed that this model is feasible and the speed of training is quicker than that of the BP neural network.