基于PSO-ENN算法的高压直流输电线路故障测距

Fault Location on HVDC Transmission Line Based on PSO-ENN Algorithm

高压直流输电线路故障测距对保障输电线路安全稳定的运行具有十分重要的意义。为了提高故障测距的精度,利用故障频谱对线路故障进行测距分析,并且故障行波在传播过程中具有强烈的固有频率信号。因此,文中将固有频率主频以及2倍频的幅值和频率作为输入训练样本,将故障距离作为输出训练样本,提出一种基于Elman动态神经网络的高压直流输电线路故障测距算法,并采用粒子群(PSO)算法优化Elman神经网络的初始权值和阈值。最后,采用MATLAB软件进行仿真,结果表明该算法具有较高的收敛速度与测距精度,可以为高压直流输电线路故障测距提供理论支持。

To improve the accuracy of fault location on high voltage direct current（HVDC） transmission line, fault spectrum is used to analyze the fault location. The fault traveling wave in propagation process contains strong natural frequency signal. Therefore, taking the natural frequency and two-octave frequency as training sample, an HVDC transmission line fault location algorithm based on dynamic Elman neural network is proposed, and the particle swarm optimization algorithm is used to optimize the initial weights and threshold. Simulation via MATLAB shows that the proposed algorithm possesses higher convergence speed and accuracy, and it can be applied to fault location on HVDC transmission line.