Abstract: The electric vehicle charging system needs to cope with complex charging needs, which suffer from significant power fluctuations and redundant charging interaction states, resulting in limited charging effectiveness and high average losses for interactive mobile electric vehicles. Therefore, based on semi supervised learning algorithms, an interactive mobile electric vehicle charging system is designed. Reconfigure the hardware structure in the two core parts of the intelligent charging node and central control gateway, connect the LoRa wireless transmission module, and achieve charging command data transmission. The optimization objective is to maximize the satisfaction of electric vehicle charging needs, construct an objective function, introduce label propagation algorithm, aggregate label update results, reduce the redundancy of charging interaction states, combine integer linear programming solver, solve the objective function, generate the optimal scheduling plan, train the objective function by predicting label iteration updates, and prioritize the charging of the scheduling plan, Output the optimal charging plan. The experimental results show that after the application of the system, charging and discharging operations can be achieved with small battery power fluctuations, and the average loss of the charging pile is lower, which has a relatively ideal charging scheduling effect.