Risk-Constrained Energy Management Optimization for Multi-Microgrids under Abnormal Conditions

نوع: Type: Thesis

مقطع: Segment: masters

عنوان: Title: Risk-Constrained Energy Management Optimization for Multi-Microgrids under Abnormal Conditions

ارائه دهنده: Provider: hossein homayoun

اساتید راهنما: Supervisors: alireza hatami

اساتید مشاور: Advisory Professors:

اساتید ممتحن یا داور: Examining professors or referees: Dr.saleh razini and Dr.mohammad mehdi shahbazi

زمان و تاریخ ارائه: Time and date of presentation: 2026

مکان ارائه: Place of presentation:

چکیده: Abstract: With the increasing penetration of renewable energy sources and the expansion of microgrids, coordinated energy management in multi-microgrid systems, particularly under uncertainty and abnormal operating conditions, has become one of the major challenges in the operation of distribution systems. This study presents a risk-constrained energy management framework for multi-microgrid systems with the objectives of reducing operating costs, minimizing energy not supplied, and enhancing system resilience. To model the uncertainty associated with renewable energy generation, load demand, and energy prices, the Conditional Value-at-Risk (CVaR) criterion is incorporated into the objective function, and the energy management problem is formulated as a Markov Decision Process. In this framework, the reinforcement learning agent observes the system state and receives rewards to learn an appropriate operating policy, while the Proximal Policy Optimization (PPO) algorithm is employed for decision-making; additionally, Soft Actor-Critic (SAC) and Multi-Agent Deep Deterministic Policy Gradient (MADDPG) are considered as alternative reinforcement learning frameworks. To evaluate the performance of the proposed method, the standard IEEE 33-bus distribution network, consisting of 33 buses and 32 distribution lines with a nominal voltage of 12.66 kV, is utilized and extended to include multiple microgrids, distributed generation resources, energy storage systems, and different types of loads. The performance of the model is evaluated under a fault scenario in which the upstream grid is disconnected from hour 10 to hour 16, accompanied by a 40% reduction in renewable energy generation. The results demonstrate that the proposed method is capable of supplying 100% of the critical loads throughout the six-hour fault period and reducing the energy not supplied to 3.05 MWh; furthermore, the system resilience index reaches 0.889. The total system recovery time is also reduced to 1.2 hours using the intelligent PPO-based strategy, while no voltage fluctuations outside the permissible range are observed during the recovery process. In addition, the online decision-making time of the proposed method is only 0.04 seconds, compared with 85.20 seconds for PSO, 120.40 seconds for GA, and 12.50 seconds for MILP, demonstrating the suitability of the proposed method for fast and online operation. In the final comparison, the proposed method achieves an operating cost of 21304.8 USD and a resilience index of 0.889, demonstrating its effectiveness in balancing economic operation, load supply, and resilience under critical and uncertain conditions.