基于波形识别的永久性和瞬时性故障的判别

RESEARCH OF SELF_ADAPTIVE RECLOSE BASED ON WAVE IDENTIFICATION

利用人工神经网络的波形识别功能解决了自适应重合闸中永久性故障和瞬时性故障的判别问题，并且提出“循环训练法”以改善神经网络的收敛特性，同时在提高网络的泛化能力方面也做了有益的探索。通过有导师的学习，结合超高压输电线进行仿真计算及数据训练，其结果表明：经充分训练的网络可正确而快速识别永久性故障和瞬时性故障，且不受故障点的位置、故障时的初相角、过渡电阻和系统运行方式的影响。

This paper deals with the problem of identification between permanent fault and transient fault in self_adaptive reclose using the wave identification function of artificial neuron network, and brings forward a circular_training method in order to improve the convergency feature of neuron network. At the same time, it makes a meaningful probe in improving the generalization feature of neuron network. After the supervised study, the digital simulation and data training of superhigh voltage transmission line, the results indicate that the well_trained neuron network is able to identify permanent fault and transient fault correctly and quickly. Furthermore, it isn′t effected by fault location, initial_phase angle at the time of fault′s outbreak, transition resistance and system operation mode.