The frequency regulation accuracy of electric bus directly affects power system stability and security. In this study, the physical model of electric busbar is established, and the PSO-ANFIS control model is constructed by combining adaptive neuro-fuzzy inference system (ANFIS) and improved particle swarm optimization algorithm (PSO). The model adopts the Sugeno fuzzy system as the basic framework, extracts the complex mapping relations in the training samples using subtractive clustering method, and optimizes the ANFIS parameters by the improved PSO algorithm to improve the model generalization and accuracy. Experimental validation shows that the method can realize the precise adjustment of output frequency accuracy in the order of 0.8E-12 when the frequency control word variation is 192315 times the minimum step value. Comparative analysis of multi-region simulation shows that compared with the traditional PI control and centralized model predictive control, the PSO-ANFIS model improves the frequency deviation regulation level in region 1 by 56.85% and 26.54%, and improves the regional control deviation (ACE) regulation by 50.56% and 40.54%, respectively. In addition, when coordinating fixed-frequency and variable-frequency electric buses to participate in regulation together, the maximum frequency deviation of the system is reduced to -0.1236 Hz, and the power deviation of the contact line is reduced to 0.0267 p.u., which is significantly better than that of a single type of electric bus participating in regulation. The study shows that the proposed adaptive fuzzy control algorithm can effectively improve the regulation accuracy of the electric bus frequency, accelerate the frequency recovery speed, reduce the overshooting amount of the FM, and better coordinate the contact line power fluctuation of the multi-area power system.
