Open Access

ARTICLE

Collaborative Optimization of Active and Reactive Power in AC/DC Hybrid Distribution Networks Based on Improved Particle Swarm Optimization Algorithm

Longbo Luo¹, Hongming Cai¹, Zili Chen², Meng Ye¹, Min Liu^2,*
1 Guangzhou Power Supply Bureau, Guangdong Power Grid Co., Ltd., CSG, Guangzhou, China
2 Energy Electricity Research Center, Jinan University, Zhuhai, China
* Corresponding Author: Min Liu. Email:

Energy Engineering https://doi.org/10.32604/ee.2026.079667

Received 26 January 2026; Accepted 16 March 2026; Published online 09 April 2026

Abstract

Driven by the national “dual carbon” strategy, the large-scale integration of distributed renewable energy and direct current loads has introduced significant challenges to traditional alternating current distribution networks, including reduced operational efficiency, decreased voltage stability, and increased complexity in grid integration. To address these issues, research investigates the optimal dispatch of alternating current/direct current (AC/DC) hybrid distribution networks under high renewable energy penetration and proposes a coordinated active and reactive power scheduling method. A multi-objective optimization model is formulated to minimize total network losses and total voltage deviations across a 24-h scheduling horizon. The model incorporates a comprehensive suite of decision variables, such as transformer tap positions, reactive compensation capacities from static var compensators and capacitor banks, energy storage system (ESS) charge and discharge outputs, and the active and reactive power setpoints of voltage source converters (VSCs) and distributed generators (DGs). To ensure physical feasibility, the formulation strictly respects network constraints, including power flow equations, inverter apparent power limits, and ESS state-of-charge dynamics. To solve highly non-linear and multi-modal problem, an improved particle swarm optimization (PSO) algorithm is developed. The algorithm integrates a linearly decreasing inertia weight with a non-linear arccosine learning factor adjustment. This dual-mechanism approach effectively balances global search breadth with local convergence precision, preventing premature stagnation in the complex solution space of the hybrid network. Simulation conducted on a modified typical AC/DC hybrid distribution network demonstrate the efficacy of the proposed method. A sensitivity analysis using a normalized Pareto front confirms that a balanced weighting scheme achieves an optimal equilibrium, reducing network losses while maintaining nodal voltages well within regulatory limits, specifically between 0.95 and 1.05 per unit, even under worst-case peak load conditions. The results indicate that the coordinated dispatch strategy significantly improves system economy and reliability. Research provides a mathematically justified and robust solution for the efficient operation of modern hybrid networks, supporting the transition toward high-penetration renewable energy systems under carbon neutrality goals.

---

Keywords

AC/DC hybrid distribution network; multi-objective optimization; reactive power optimization

---