Dynamic Intelligent Method for Voltage Violation Management in High-Renewable-Penetration Distribution Networks
TL;DR
The method employs a dual-agent architecture: DERMS_Agent coordinates task scheduling, data management, and computational resource allocation, while Solution_Agent performs three-phase unbalanced power flow calculation and MIQP-based voltage violation joint optimization.
Abstract
This paper proposes a dynamic intelligent method for voltage violation management in high-renewable-penetration distribution networks. The method employs a dual-agent architecture: DERMS_Agent coordinates task scheduling, data management, and computational resource allocation, while Solution_Agent performs three-phase unbalanced power flow calculation and MIQP-based voltage violation joint optimization. Four key technical contributions are presented. (i) An asymmetric nodal admittance matrix is developed to incorporate transformer tap-phase-shift and capacitor branches within a unified formulation. (ii) Five categories of analytical sensitivities are systematically derived, covering transformer tap, phase shift, and capacitor compensation effects for both voltage regulation and harmonic suppression. (iii) A three-parameter MIQP joint optimization model is constructed with voltage deviation minimization as the objective and three-phase unbalance and resonance avoidance as constraints. (iv) A two-stage hybrid solution strategy combining Ipopt continuous relaxation with Gurobi neighborhood enumeration is designed to achieve real-time solvability. Validation on a real 10 kV feeder with 91 transformer areas over 768 time sections (8 days) demonstrates a 95.6% voltage violation resolution rate within the first three polling cycles and an average single-section solution time of 0.83 s, satisfying the real-time requirements of 15 min operational control cycles.