一种基于神经网络的馈线自适应继电保护技术

针对常规保护的自适应能力及检测某些复杂故障方面存在一定局限性，利用前馈神经网络，导出了常规保护与神经网络参数间的显式关系，证明了神经网络继电保护可实现常规馈线保护的功能。

A Neural Network Based Adaptive Feeder Relay Protection

Considering the limitations of conventional protections in adaptability and in the detecting some complex failures, the explicit relationship between the conventional protections and the neural network (NN) parameters is derived by use of a simple feedforward neural network (FNN). The NN based protective relay is proved to be able to perform the protective functions of conventional feeder protective relays, and a new NN based adaptive feeder protective relaying approach is proposed. Because neural networks have a very strong ability in pattern recognition, an NN classifier is used in the multi parameter recognition of fault signals. Simulation results show that it is an effective adaptive protection.