基于神经网络和准同步采样算法的电力谐波分析

Power Harmonic Analysis via Neural Network and Quasi-synchronous Sampling Method

针对非同步采样的电力系统谐波分析精度问题,文章提出了一种准同步采样算法和神经网络相结合的谐波分析算法。其以准同步采样算法为基础,给出了常用情况下的准同步窗系数的函数表达式,为神经网络谐波分析算法提供了基波频率估计值;然后,基于最速下降法,确定了迭代方向上的适应性最优步长,该步长使神经网络算法收敛于全局最小值。在非同步采样情况下,该算法迭代次数均为10余次,幅值检测的相对误差精度可达1×10-10%,相量检测的相对误差精度可达1×10^(-8)%;并且在信噪比为30 dB时,幅值检测的相对误差基本低于1×10^(-2)%。仿真结果表明,该算法检测速度快且精度高,具有较好的应用价值。

Aiming at the accuracy of power system harmonic analysis with asynchronous sampling,a harmonic analysis algorithm based on quasi synchronous sampling algorithm and neural network is proposed in this paper.Based on the quasi synchronous sampling algorithm,a functional expression of quasi synchronous window coefficient in common cases is given,which provides a fundamental frequency estimation for neural network harmonic analysis algorithm.Then,based on the steepest descent method,the adaptive optimal step size in the iterative direction is determined,which makes the neural network algorithm converge to the global minimum value.In the case of asynchronous sampling,the iteration number of the algorithm is more than 10 times,the relative error accuracy of amplitude detection can reach 1×10-10%,and the relative error accuracy of phasor detection can reach 1×10-8%.When the signal-to-noise ratio is 30 dB,the relative error of amplitude detection is basically less than 1×10-2%.Simulation results show that the algorithm has fast detection speed and high precision,and has good application value.