Multi-agent Deep Reinforcement Learning-based (MADRL) optimal dispatching for active distribution networks (ADNs) represents a future trend, necessitating robust security measures due to the sensitivity of MADRL. However, limited research addresses this issue. This paper introduces Multi-Agent Directed Shift Attack (MADSA), a powerful attack strategy for MADRL in ADNs, which targets single and full agents. The single-agent attack maximizes the deviation of a single agent’s strategy before and after the attack, leading to ADN degradation. And the full-agent attack unifies the action direction of all agents to achieve maximum degradation. To counter MADSA, we propose an adversarial defense method named Multi-Agent Gradient Leveling Defense (MAGLD) with Gradient Leveling Regularization, which enhances the robustness of the defense strategy. Case studies show MADSA can degrade the ADN under continuous and single-step attacks, with even the small attack amplitudes significantly increasing power losses and causing overvoltage, surpassing existing methods, such as fast gradient sign method and noise attacks. The proposed defense strategy effectively mitigates these attacks, offering forward-looking considerations for the security of artificial intelligence based control in ADNs.

As the number of electric vehicles (EVs) increases, massive numbers of EVs have started to gather in commercial parking lots to charge and discharge, which may significantly impact the operation of the grid. There may also be a deviation in the departure time of charged and discharged EVs in commercial parking lots. This deviation can lead to insufficient battery energy when the EVs leave the parking lot. This study uses the simulation software AnyLogic to build a commercial parking lot multi-agent simulation model, and the agent-based model can fully reflect the autonomy of individual EVs. Based on this simulation model, we propose an EV scheduling algorithm. The algorithm contains two main agents. The first is the power distribution center agent (PDCA), which is used to coordinate the energy output of photovoltaic (PV), energy storage system (ESS), and distribution station (DS) to solve the problem of grid overload. The second is the scheduling center agent (SCA), which is used to solve the insufficient battery energy problem due to EVs’ random departures. The SCA includes two stages. In the first stage, a priority scheduling algorithm is proposed to emphasize the fairness of EV charging. In the second stage, a genetic algorithm is used to accurately determine the time interval between charging and discharging to ensure the maximum benefit of EV owner. Finally, simulation experiments are conducted in AnyLogic, and the results demonstrate the superiority of the algorithm over the classical algorithm.