Impedance-Estimation Driven Voltage Management Framework for Unbalanced Inverter Based Resource Integrated Distribution Networks Through Coordinated Control

Abstract

The growing penetration of distributed energy resources (DERs), advanced power electronics, and dynamically reconfigurable switching operations has transformed modern electric distribution systems into highly nonlinear, unbalanced, and data-rich networks. Conventional analysis tools are highly dependent on fixed system parameters, static topology information, and balanced approximations that are no longer valid under high DER variability, frequent switching, and unbalanced loading. Consequently, utilities face limited observability, inaccurate state estimation, and degraded effectiveness of legacy voltage control devices such as capacitor banks and on-load tap changers (OLTCs). This dissertation Addresses these challenges through an integrated, physics-informed, data-driven framework for three-phase unbalanced distribution systems, encompassing real-time parameter estimation, topology identification, and coordinated voltage regulation. The first contribution is a hybrid Kalman filter estimation algorithm for real-time estimation of line conductance (G) and susceptance (B). The framework decouples the linear prediction step from a sigma-point nonlinear update step, leveraging a physics-informed path matrix to achieve numerical stability, phase-selective estimation, and fast convergence under load changes, topology switching, shunt capacitor banks, and high DER penetration. Validation on the IEEE 13-bus and 123-bus test feeders demonstrates superior accuracy (MAE, RMSE, R2), and computational efficiency relative to standalone EKF and UKF methods under partial observability and measurement noise. The second contribution is a Thevenin impedance-based topology estimation frame work that infers switch open/closed status across multiple feasible radial configurations without exhaustive search or full network observability. By exploiting deviations in estimated Thevenin impedance signatures at switch locations, the method reliably distinguishes among five feasible radial topologies of the IEEE 123-bus feeder, even under unbalanced voltages and measurement sparsity. The third contribution is a data-driven, Z-Bus-aided voltage regulation framework that uses estimated parameters and identified topology to compute real-time phase-wise Thevenin voltage sensitivities for tap decision-making. The framework reconstructs primary-side regulator voltages under partial observability, including switch-adjacent regulator locations, and executes prediction-before-action tap control to avoid oscillatory behavior. Validated across static, dynamic, and topology-transition scenarios, the controller restores voltage profiles across all three phases with fewer tap operations and faster correction than conventional OLTC strategies, particularly during simultaneous load fluctuations and topology changes. Collectively, these three components form a unified situational-awareness and control framework that improves distribution system observability, enables accurate real-time modeling, and delivers reliable voltage regulation under high DER penetration. The results directly support the development of next-generation distribution management systems (DMS), providing utilities with scalable, model-informed, data-driven tool for operating modern unbalanced distribution networks