Online identification of generator dynamics in a multimachine power system with a spiking neural network

This paper presents the application of a spiking neural network for online identification of generator dynamics in a multimachine power system. An integrate and fire model of a spiking neuron is used in this paper where the information is communicated through the interspike intervals. A network of spiking neurons is trained online based on a gradient descent algorithm. Speed and terminal voltage deviations of a generator in the IEEE 10-machine 39-bus New England power system are predicted one time step ahead by a spiking neural network. Two different training conditions are considered, namely, forced and natural perturbations. The simulation results show that a spiking neural network can successfully estimate the speed and terminal voltage deviations for both small and large perturbations applied to a power network.