The estimation of the harmonic impedance based on linear regression is one of the commonly used estimation methods. In order to ensure the correlation between real part and imaginary part of the harmonic impedance estimation

the strict mathematical derivation and simplified solution procedure of the single dependent variable complex domain partial least square method is presented. In view of the responsibility distinction for multiple harmonic sources at the public connection point

based on the actual recorded data

by analyzing the relationship between the current fluctuation of each feeder line

the index factor has been obtained

which is used to get the dominant fluctuation filtering to screen the data between voltage and current with good linearity. Then the harmonic impedance of the equivalent system is estimated by the partial least squares regression in the complex field. Finally

the quantitative responsibility distinction of each harmonic source is realized by virtue of the equivalent method proposed

its feasibility and effectiveness are verified by simulation and experimental results. This work is supported by Technology Research and Development Program of China Railway Coperation(No. 2015J005-A).