A detection method for abnormal harmonic current monitoring data using three-parameter Weibull distribution

A handful of abnormal data exist in harmonic current monitoring data, affecting the accuracy of the harmonic source localization and harmonic contribution assessment. In this paper, the distribution model of harmonic current monitoring data is established using three-parameter Weibull distribution according to the statistical characteristics of the harmonic current data. The least squares method is used to estimate the shape parameter, scale parameter and location parameter of Weibull distribution. Based on the parameter estimation of three-parameter Weibull distribution, the range of the majority sample is obtained in order to determine the threshold for identifying the abnormal data in harmonic current. The simulation results show that the three-parameter Weibull distribution has better adaptability for harmonic current characteristic, the method proposed can identify the abnormal data effectively.