摘要
输电线路故障行波频谱与故障距离之间存在数学关系,故利用故障行波频谱可以实现故障测距,且高压直流输电线路故障暂态过程具有更强的固有频率信号。鉴此,提取固有频率的幅值和频率作为样本属性,提出一种基于神经网络的高压直流输电线路故障测距算法。采用粒子群算法对BP神经网络的权值和阈值进行优化,提高了网络的训练效率,使其收敛速度加快。PSCAD和MATLAB仿真结果表明,该故障测距算法具有较高的可靠性和精确性。
The spectra of fault induced traveling wave is related to the fault distance,therefore the spectra of fault induced traveling wave can be utilized for fault location. And the transient process of HVDC transmission line has more stronger inherent frequency signal. The amplitude frequency and frequency of the inherent frequency are cho-sen as the characteristics of artificial neutral network (ANN). A novel fault-location method for HVDC transmis-sion lines based on ANN is presented. It can improve the training efficiency,speed up convergence by using PSO algorithm to optimize the weights and thresholds of BP neural network. Simulation results by PSCAD and MATLAB show that accurate fault location method of HVDC transmission lines can be achieved with the proposed method.
出处
《电力科学与工程》
2014年第3期45-49,共5页
Electric Power Science and Engineering