A Harmonic Impedance Identification Method of Traction Network Based on Data Evolution Mechanism

In railway electrification systems, the harmonic impedance of the traction network is of great value for avoiding harmonic resonance and electrical matching of impedance parameters between trains and traction networks. Therefore, harmonic impedance identification is beneficial to suppress harmonics and improve the power quality of the traction network. As a result of the coupling characteristics of the traction power supply system, the identification results of harmonic impedance may be inaccurate and controversial. In this context, an identification method based on a data evolution mechanism is proposed. At first, a harmonic impedance model is established and the equivalent circuit of the traction network is established. According to the harmonic impedance model, the proposed method eliminates the outliers of the measured data from trains by the Grubbs criterion and calculates the harmonic impedance by partial least squares regression. Then, the data evolution mechanism based on the sample coefficient of determination is introduced to estimate the reliability of the identification results and to divide results into several reliability levels. Furthermore, in the data evolution mechanism through adding new harmonic data, the low-reliability results can be replaced by the new results with high reliability and, finally, the high-reliability results can cover all frequencies. Moreover, the identification results based on the simulation data show the higher reliability results are more accurate than the lower reliability results. The measured data verify that the the data evolution mechanism can improve accuracy and reliability, and their results prove the feasibility and validation of the proposed method.