Robust phase detection in distribution systems

This paper proposes an on-line algorithm to detect the phase connection for end users in a power distribution system. In distribution systems, feeder switching often changes the phase connection information of end users in the real-time operation. Recently, Advanced Metering infrastructure (AMI) are being installed in distribution systems. They enable utilities to record end-point voltages in defined intervals. This paper first presents a method to find phases belonging to same phase using synchronized data and fixed topology of distribution grid. Then, this paper presents a method to clean the noisy data through Artificial Neural Network (ANN) trained by the historical data. The proposed methods is tested using modified 13-bus IEEE distribution test system on the low-voltage side.