Transmission network topology control offers cheap flexibility to system operators for mitigating grid congestion. However, finding the optimal sequence of topology actions remains a challenge due to the large number of possible actions. Although reinforcement learning (RL) approaches have attracted interest for long-term planning in large combinatorial action spaces, they encounter challenges such as training stability, sample efficiency, and unforeseen consequences of RL actions. Addressing these challenges, this paper proposes a hybrid curriculum-trained RL and Monte Carlo tree search (MCTS) approach to determine sequential topological actions for mitigating grid congestion. The curriculum-based approach stabilizes training by first pre-training a policy network through supervised imitation learning, followed by RL training. The policy network guides the MCTS to simulate promising future trajectories, mitigating unforeseen consequences and identifying long-term strategies to improve grid security. Moreover, the MCTS-verified actions are used for RL training, enhancing sample efficiency and training time. A distance factor is added to the MCTS, which improves convergence by prioritizing actions closer to congestion. Numerical results on the IEEE 118-bus system show that the proposed hybrid approach improves the timesteps survived by 30% compared to a standard RL approach, and by 5% compared to a brute-force baseline. Additionally, the inclusion of the distance factor increases the timesteps survived by 15%. These results highlight the potential of the proposed method for real-world applications of using sequential topological actions to effectively relieve grid congestion.

The increasing share of uncertain renewable energy sources (RES) in power systems necessitates new efficient approaches for the two-stage stochastic multi-period AC optimal power flow (St-MP-OPF) optimization. The computational complexity of St-MP-OPF, particularly with AC constraints, grows exponentially with the number of uncertainty scenarios and the time horizon. This complexity poses significant challenges for large-scale transmission systems that require numerous scenarios to capture RES stochasticities. This paper introduces a scenario-based decomposition of the St-MP-OPF based on the alternating direction method of multipliers (ADMM). Additionally, this paper proposes a machine learning-accelerated ADMM approach (ADMM-ML), facilitating rapid and parallel computations of numerous scenarios with extended time horizons. Within this approach, a recurrent neural network approximates the ADMM sub-problem optimization and predicts wait-and-see decisions for uncertainty scenarios, while a master optimization determines here-and-now decisions. Additionally, we develop a hybrid approach that uses ML predictions to warm-start the ADMM algorithm, combining the computational efficiency of ML with the feasibility and optimality guarantees of optimization methods. The numerical results on the 118-bus and 1354-bus system show that the proposed ADMM-ML approach solves the St-MP-OPF with 3-4 orders of magnitude speed-ups, while the hybrid approach provides a balance between speed-ups and optimality.

The privatization of distribution systems has resulted in the development of multiple-microgrid (multiple-MG) systems where each microgrid independently operates its local resources. Moreover, the high integration of independent distributed energy sources could lead to operational issues such as grid congestion in future distribution systems. Therefore, this paper provides a transactive-based energy management framework to operate multiple-MG distribution systems; while, alleviating grid congestion in a decentralized manner. In this respect, alternating direction method of multipliers (ADMM) is considered to develop an operational framework that copes with distributed nature of multiple-MG systems. In this context, a novel procedure in the context of ADMM is proposed to distributedly determine transactive coordinator signals which address energy prices as well as power losses and grid congestions. Furthermore, each MG takes into account stochastic programming and the conditional value-at-risk index to handle the uncertainty of its operational scheduling. At last, the proposed framework is applied on IEEE 37-bus and 123-bus test grids to investigate its efficacy in distributed energy management of multiple-MG systems.

Presence of microgrids (MGs) and renewable energy sources (RESs) in distribution power systems could result in dramatic changes in operation and planning of these systems. In this regard, operational procedures employed by utilities should take into account the intermittent nature of RESs, while coping with the independent operation of MGs. In this perspective, this paper develops a distributed power management framework based on alternating direction method of multipliers (ADMM) for distribution networks with multi-MG (MMG) structures. The proposed approach develops a transactive signal in the context of ADMM to coordinate operation of MGs in a distributed manner. In this context, the configuration of MMG distribution systems is devised in a way that facilitates modern distributed control of the system by eliminating the role of a central coordinator in traditional distribution systems. Moreover, robust optimization technique is deployed in the MG operational scheduling to handle the uncertainty associated with RESs. Finally, through a numerical case study, the efficiency of the proposed framework to schedule local resources for MMG distribution systems is analyzed.

Distribution networks are undergoing a fundamental transition due to the expansion of flexible resources as well as renewable energy sources in the system. In this regard, multi-agent structures are developed in modern distribution systems to facilitate the independent operation of local resources. Nevertheless, the non-coordinated operation of independent agents could result in a deviation between the real-time power purchased from transmission network and the day-ahead scheduling. Consequently, this paper aims to provide a novel framework that enables the decentralized management of multi-agent distribution systems, while coordinating the real-time power request and the day-ahead scheduling. In this regard, the alternating direction method of multipliers (ADMM) is taken into account to facilitate the decentralized operation of the multi-agent systems. Furthermore, transactive control signals are employed to exploit the real-time operational scheduling of independent agents in order to minimize the deviation of real-time power exchange and the day-ahead scheduling. Finally, the developed methodology is implemented on the IEEE 37-bus test system in order to analyze the effectiveness of the proposed approach for the operational management of multi-agent distribution systems.

In this paper, a two-stage charging management framework is proposed to exploit the potential flexibility of Electric Vehicles (EVs). The aim is to reduce drastic variations in distribution system net load caused by integration of intermittent renewable distributed generation (DG) units. In the first stage, a home-based charging method is formulated in which the desired charging schedule of EVs is obtained by minimizing the cost of energy taking into account owners' preferences. The attained charging schedules are then announced to EVs Coordinator Agent (ECA) which, as the second stage, applies an LP optimization to reduce the cost of ramp provision. Moreover, an incentive scheme is considered to motivate EV owners to participate in the controlled charging program which allows the ECA to modify their desired schedules in exchange for financial bonuses. The simulation results show that the proposed framework can efficiently reduce the system net load ramp and associated costs without disturbing individual convenience.