Design of A Multi-Agent System for Electric Vehicle Cluster Regulation Based on Collaborative Filtering Algorithm

The large-scale integration of electric vehicles (EVs) has brought opportunities and challenges to the scheduling of integrated energy systems. The large-scale integration of EVs into the power grid will lead to an increase in the peak load curve of the distribution network due to disorderly charging, posing a threat to the safe operation of the power system and reducing the economic efficiency of system operation. The integration of large-scale EVs into the power grid has certain potential for regulation at various time periods, and the participation of EV clusters (EVCs) in grid demand response can be regulated through optimized charging and discharging operations to improve the adequacy level of the distribution network. Collaborative filtering algorithm (CF) is a case of data mining (DM) and the most popular machine learning algorithm. A multi-agent system for EVC regulation based on DM and CF is proposed to address the problem of existing charging station planning neglecting the upper limit of distribution network capacity and uneven distribution network load caused by the large number of EVs connected. At the same time, considering the preferences of EV users in practical scenarios. By prioritizing the calculation of reactive power resource regulation near nodes, and when the regulation capacity is insufficient, the goal of improving regional node voltage is achieved by adjusting the EVC charging power. The experimental results show that the proposed system has certain feasibility and can optimize the partitioning results of the distribution network, improving the safety and stability of the distribution network to a certain extent.