Parameters identification in measurement of distribution lines with mutual inductance

With the appearance of high voltage parallel distribution lines in many cities, their mutual inductance measurement becomes one of the difficulties in research. In the existing on-line measurement methods such as incremental method, differential equation method and integral equation method, parameters are all worked out by solving the over-determined equations acquired in different tests. Focusing on the parameters identification of those methods, principles of least square(LS) estimation and total least square(TLS) estimation are analyzed. Digital simulations were made to compare their regression fitting optimization indexes under different noise. It shows that LS estimation can not obtain the optimal solution when coefficient matrix and constant vector both have noise while TLS estimation can. Laboratory tests using coils with mutual inductance as power distribution lines are carried out to verify the conclusion.