Impacts of Modeling Errors and Randomness on Topology Identification of Electric Distribution Network

Abstract

A few models have been established to study the relationship between SCADA voltages in the cyber layer and the topology of electric distribution networks in the physical layer. While these models could identify the topology of distribution networks under some assumptions, one of the issues that havent been deeply studied is whether the topology of distribution networks could be identified when violating these assumptions. We define the violation of these assumptions as the impact of modeling errors which may have different characteristics comparing to other errors. When the model of using SCADA voltage correlation to identify the topology of distribution networks is constructed, a few assumptions are made such as uniformed $L$ / $R$ ratio and the uncorrelated injective power which is a very strict restriction in practice. This paper focuses on understanding the impact of the identification result when violating some of these assumptions that used to established the model. Specifically, 4 cases are presented in this paper while each case contains whether violates the assumption of uniformed $L$ / $R$ ratio and uncorrelated injective power or not. The analysis results show that violating each individual assumptions could cause inaccurate identification result. Also, the results show that the errors could be decreased by increasing the sample size of the SCADA voltage. Understanding the impact of errors is useful to better understand weakness of the model and find the way of decreasing errors.