Connection Phase Estimation of Pole Mounted Distribution Transformers by Integer Form of Population Based Incremental Learning Considering Measurement Errors and Outliers by Correntropy

This paper proposes connection phase estimation of pole mounted distribution transformers by integer form of population based incremental learning considering measurement errors and outliers by correntropy. In electric power distribution systems, since it is difficult for electric power utilities to manage connection phases of pole mounted distribution transformers, a connection phase estimation method is required. Connection phase estimation can be formulated as a combinatorial optimization problem using measurement data of measurement points. Conventionally, various connection phase estimation methods have been developed such as statistic, branch and bound, and tabu search based methods. However, when measurement errors and outliers occur in the measurement data, the conventional methods cannot handle them or maintain estimation accuracy. Moreover, since the branch and bound based method dose not utilize power flow calculation, accurate electric circuit calculation, for calculating an objective function value, a solution may not be evaluated accurately. Regarding the tabu search based method, since various evolutionary computation methods have been developed in recently years, estimation accuracy using the tabu search based method may be improved by applying other evolutionary computation methods. The proposed method is applied to a distribution model system based on a JST-CREST126 distribution lines model. Simulation results indicate correct connection phases can be estimated by applying the correntropy to the connection phase estimation problem even if measurement errors occur. Moreover, the proposed method can improve the estimation accuracy than the conventional method.