Centralized Hybrid Voltage Control of Active Distribution Network Using Recursive Least Squares Method for Sensitivity Estimation

This paper presents a hybrid centralized, optimal, adaptive, and data-driven Volt-VAR and Volt-Watt control framework for voltage regulation in distribution systems. The proposed methodology coordinates on-load tap changers (OLTC), distributed generation (DG), battery energy storage systems (BESS), capacitor banks, and demand response systems within a unified optimization framework. The proposed algorithm uses real-time measurements to continuously update voltage-control sensitivity matrix, significantly reducing computational cost. Mixed-integer programming optimizes both discrete OLTC and capacitor operations, and continuous controller setpoints while respecting operational constraints, power factor and battery state-of-charge requirements. Validation on a modified, unbalanced IEEE 34-bus distribution system demonstrates effectiveness under normal and extreme operating conditions. For typical disturbances, the model resolves voltage deviations within 1.2 seconds per iteration. Under severe conditions with load changes up to three times nominal values, voltage deviations are progressively resolved within 6 seconds maximum. Results show effective coordination among all controllers, with the recursive sensitivity adaptation enabling learning from system disturbances to improve subsequent control actions. The proposed approach offers a computationally efficient and practically implementable solution for modern smart grid applications with diverse distributed energy resources.