基于分段迭代的电力谐波神经网络分析方法

发布时间：2018-03-10 21:55

本文选题：神经网络　切入点：谐波　出处：《电测与仪表》2014年23期 　论文类型：期刊论文

【摘要】：为了提高谐波和间谐波分析的速度、精度和抗噪性能,将基于分段迭代的增强型Adaline神经网络应用于电力系统谐波和间谐波分析。该网络在加汉宁窗双谱线插值FFT算法的基础上,将采样数据按采样时间分段,依次用各段对应的误差信息来调整增强型Adaline神经网络的参数。该方法结合一点迭代法和全部点迭代法的优点,既将各时段内的误差进行平均,减少噪声对参数调整的影响,又充分保留误差中包含的谐波和间谐波信息,提高网络的精度。另外,根据参数估计误差和频率对误差函数一阶偏导之间的关系,提出修正信号幅值最大分量对应的频率调整量的处理方式,提高网络的实时性和精度。仿真结果验证了分析结论的正确性。
[Abstract]:In order to improve the speed, accuracy and anti-noise performance of harmonic and interharmonic analysis, an enhanced Adaline neural network based on piecewise iteration is applied to harmonic and interharmonic analysis of power system. The network is based on the FFT algorithm with double spectral lines interpolation. The parameters of the enhanced Adaline neural network are adjusted with the corresponding error information of each segment according to the sampling time segment. The method combines the advantages of the one-point iterative method and the all-point iterative method, and the errors in each period are averaged. The effect of noise on parameter adjustment is reduced, and the harmonic and interharmonic information contained in the error is fully preserved to improve the accuracy of the network. In addition, according to the relationship between the parameter estimation error and frequency to the first order partial derivative of the error function, In order to improve the real-time and accuracy of the network, the processing method of correcting the frequency adjuster corresponding to the maximum signal amplitude component is proposed. The simulation results verify the correctness of the analysis conclusion.
【作者单位】：国网浙江建德市供电公司;重庆大学电气工程学院输配电装备及系统安全与新技术国家重点实验室;国网浙江杭州市余杭区供电公司;
【基金】：重庆市科技攻关资助项目(CSTC,2011AB3003) 国家“111”计划资助项目(B08036)
【分类号】：TM711

【参考文献】