Supervised Learning Framework for Identification of Inverter Control Mode and Estimation of Inverter-Based-Resource & Load Parameters

Modern distribution networks are becoming increasingly active in nature owing to the rapid proliferation of Inverter-based-Resources (IBRs). Since these IBRs are capable of operating in multiple control modes, it is important for the utility operator to accurately identify the control modes and also estimate the associated parameters. This paper presents a three-pronged data-driven approach for the real-time control mode identification of inverter-based resources (IBRs) and parameter estimation of IBRs and aggregated load. The proposed approach utilizes voltages, and power data at micro-PMU reporting rate at the node under observation for IBR mode identification and parameter estimation. A real-time implementation framework for the three-pronged approach having an adaptive window selection algorithm is developed to track the control mode along with parameters in real-time. The identified windows from an adaptive window selection algorithm are utilized by the three-pronged approach for i) identification of control modes, ii) parameter estimation of IBRs, and iii) aggregated load parameter estimation. The paper also performs robustness analysis by considering different noise levels in the PMU measurements. Random forest, K-nearest neighbors, and Decision tree algorithms are utilized for each prong of the three-pronged approach with RF outperforming the other algorithms. The proposed approach developed in this work is tested in detail on the IEEE 13-node distribution test system to demonstrate its effectiveness.