Integrated dynamic spiking neural P systems for fault line selection in distribution network

Abstract

Due to the compensating function of neutral grounded arc suppression coil, fault line selection in distribution network is still facing challenges: the classical models have insufficient learning ability in extracting fault features, and there is an imbalance in the original data used, resulting in low accuracy in fault line selection. In order to address this issue, this paper proposes a novel variant of spiking neural P (SNP) systems called integrated dynamic SNP (IDSNP) systems, which consist of gated neurons, rule neurons, and weighed neurons with different designed rules. Furthermore, according to the IDSNP systems, an IDSNP(FL) model is developed for fault line selection in distribution network, where the number of layers for transmitting weighted neuron spiking information could be dynamically changeable depending on the complexity of the number of stations in the power system. Finally, the proposed model is evaluated on a real-time dispatch dataset of a real power system. The experimental results show that the IDSNP(FL) model achieves the best performance compared with several classical models in deep learning, verifying the effectiveness of the proposed model for fault line selection tasks in distribution network.