Topology Transformation Based ANN Approach for Optimal PMU Placement for Monitoring and Control of Power System

Abstract

Efficient placement of Phasor Measurement Units (PMUs) is crucial to minimize their number while ensuring full power system observability. A power system is deemed observable when the voltage at every Bus is known. This study proposes a topology transformation approach combined with Artificial Neural Networks (ANN) for optimal PMU placement. The approach incorporates Zero-Injection Buses (ZIBs), where a ZIB is merged with one of its neighboring buses to reduce the number of required PMUs. The selection of the neighboring Bus significantly affects the merging outcome. Three selection principles are applied to identify the optimal Bus for merging with the ZIB. These principles are designed to minimize the number of PMUs necessary for complete system observability. ANN further enhances the method by training the network on historical data to predict optimal PMU placement patterns, thereby improving placement accuracy and reducing computational time. Additionally, the research explores scenarios involving power flow measurements and formulating the placement problem using Integer Linear Programming (ILP) to determine the minimal PMU set. Simulations are conducted in MATLAB on various IEEE bus systems, with a detailed case study provided for the IEEE 14-bus system. The results show that the combined topology transformation and ANN approach produces competitive results compared to existing techniques. The study also compares the IEEE 14-bus and IEEE 30-bus systems by inducing faults at specific buses to verify if PMUs at optimal locations provide complete observability. The comparison is extended to cost analysis, evaluating the financial implications of PMU placement at these optimal locations. The inclusion of ANN aids in reducing costs while maintaining the system's observability and fault coverage.