基于多智能体系统的微电网安稳控制策略

随着新能源渗透率不断提高，微电网在分布式能源(distributed energy resources, DERs)运行控制中的作用日益凸显，但分布式能源的接入会给微电网的运行带来扰动。为此，本文提出了一种基于多智能体系统 (multi-agent system，MAS) 的两级分散协调控制方案来保证微电网的高安全性和稳定性。上层智能体基于Petri网(petri-net, PN)，针对各类能源的不同工作模式，构建DERs的PN模型，并以PN模型设计协调逻辑控制指令集，通过网格自适应直接搜索过滤算法 (mesh adaptive direct search filter algorithm, MADSFA) 选择最佳的协调逻辑控制指令。下层智能体负责局部连续控制，针对不同功率特性DERs的逆变器，分别采用不同的下垂控制策略。为保证暂稳态下微电网的稳定性，在逆变器和下垂控制器之间增设预设性能控制器，以确保逆变器的输出功率误差无论是在暂态还是稳态的情况下均能稳定在预设误差范围以内。同时为提高微电网系统的稳定性，两层之间的交互采用主从和无固定主从相结合的通信模式，这种交互机制提高了MAS在系统的实时监控过程中的时效性和灵活性。最后，仿真结果验证了基于MAS两级分散协调控制策略的有效性。

Stability Control Strategy of Microgrid Based on Multi-agent System

With the increasing penetration of new energy, the role of microgrids in the operation control of distributed energy resources (DERs) has become increasingly prominent, but the access of distributed energy will also bring disturbance to the operation of microgrids. Therefore, this paper proposes a two-level decentralized coordinated control scheme based on multi-agent system (MAS) to ensure the high security and stability of microgrid. The upper agent is based on Petri-net (PN). According to the different working modes of various energy sources, the PN model of DERs is constructed, and the coordination logic control instruction set is designed based on the PN model. The optimal coordination logic control instruction is selected by mesh adaptive direct search filter algorithm (MADSFA). The lower agent is responsible for local continuous control, and different droop control strategies are adopted for inverters with different power characteristics DERs. In order to ensure the stability of the microgrid under transient and steady states, a preset performance controller is added between the inverter and the droop controller to ensure that the output power error of the inverter can be stabilized within the preset error range whether in transient or steady state. At the same time, in order to improve the stability of the microgrid system, the interaction between the two layers adopts the communication mode of master-slave and non-fixed master-slave combination. This interaction mechanism improves the timeliness and flexibility of MAS in the real-time monitoring process of the system. Finally, the simulation results verify the effectiveness of the two-level decentralized coordinated control strategy based on MAS.