基于电力大数据的分布式电网异常负荷动态检测方法

A Dynamic Detection Method of Abnormal Load in Distributed Power Grid Based on Power Big Data

电网负荷数据基数大、分布范围广,且其异常状态检测复杂度高。提出一种基于电力大数据分布的电网异常负荷动态检测方法。采用非线性回归方程,估计中心负荷权重,并分割动态检测区域。采用状态估计法结合参数平滑对异常的负荷数据进行状态估计,并利用自回归滤波(extended Kalman filter,EKF)剔除噪声数据。计算负荷数据的近相似系数,划分异常数据域,设定分布概率较高的数据为异常负荷数据。通过观测负荷数据与异常域中心之间的关联性,判断负荷是否存在异常问题。仿真实验结果表明:高信噪比环境下,该方法检测异常负荷数据的最大特征量为160条;低信噪比环境下,异常负荷数据的最大特征量为158条,且峰值出现在节点51—节点120的位置。检测出的负荷数量均超过电网最大阈值,说明所提方法能够精确检测出异常负荷,且能够完全包含真实阈值,检测的全面性可高达100%。

Power grid load data is characteristic of huge amount,wide distribution range and high complexity of abnormal state detection.Therefore,a dynamic detection method of abnormal load based on power big data distribution is proposed.The center load weight is estimated by nonlinear regression equation and the dynamic detection region is segmented.The state estimation method combined with parameter smoothing is used to estimate the abnormal load data,and the Extended Kalman Filter(EKF)is used to eliminate the noise data.The near similarity coefficient of the load data is calculated,abnormal data domain is divided,and the data with high distribution probability is set as abnormal load data.Whether abnormal load exists is judged according to the correlation between observed load data and abnormal domain center.The simulation results show that under the high SNR environment,the maximum feature quantity of abnormal load data detected by the proposed method is 160;in the low SNR environment,the maximum feature quantity of abnormal load data detected is 158,and the peak value occurs at the position of Node 51-Node 120.The number of loads detected exceeds the maximum threshold of the power grid,indicating that the proposed method can accurately detect abnormal loads,and can fully contain the true threshold,and the detection comprehensiveness can reach 100%.