基于神经网络的自适应距离保护研究

STUDY ON NEURAL NETWORK BASED ADAPTIVE DISTANCE PROTECTION

超高压输电线路的距离保护必须具备足够的对过渡电阻的反应能力。在双侧电源条件下，过渡电阻可能造成保护范围的伸长或缩短。提出了神经网络距离保护算法，通过对样本的学习来获得保护的动作特性，具有内在的对过渡电阻和系统运行方式的自适应性。该算法已在一实际的500kV系统输电线路上仿真实现，考虑单相接地故障和I段保护，在设计的90％保护范围内具有稳定良好的保护性能。

Distance protection for EHV transmission lines is demanded to have the ability to deal with fault resistance. In condition of two terminal power sources. however. conventional distance protection may underreach or overreach protection range by the influence of fault resistance and the remote source infeed. In this paper. a novel distance protection scheme for EHV transmission lines using neural networks is proposed. which obtains operating characteristics by learning of samples and achieves adaptability to fault resistance and the remote source infeed. The protection technique in zone I for a one-line to ground fault is described in detail and tested with a realistic. complicated 500 kV simulation model system. A reliable 90% of line length for zone I is realized.