Utility harmonic impedance estimation based on time series clustering and improved Pettitt method

• Harmonic data are filtered by a time series clustering method to reduce the influence of background harmonic fluctuations. • An improved Pettitt method is proposed to detect the change points of the harmonic impedance. • The self-born weighted least square method is used to calculate the harmonic impedance by adaptively weighting the data. • The capacitive voltage transformer measurement error is corrected by a parameter identification method. Utility harmonic impedance estimation is critical for power quality assessment and improvement. Noninvasive methods without injecting harmonics are widely adopted to estimate utility harmonic impedance using natural load variations. However, background harmonic voltage fluctuations and abrupt harmonic impedance changes can lead to significant errors in utility harmonic impedance estimation. In this paper, a new noninvasive method is proposed to solve the above problem. To overcome the errors in harmonic impedance estimation caused by background harmonic voltage fluctuations, a time series clustering (TSC) method based on the cross-correlation principle is proposed to filter harmonic data. Moreover, an improved Pettitt method is proposed to identify the change points of harmonic impedance. Finally, the self-born weighted least squares (SBWLS) method is used to calculate harmonic impedance by iteratively weighting the anomalous data to weaken its influence, thereby improving the accuracy of the utility harmonic impedance. Simulation and field results validate the proposed method.