Considering that the speed control system of the suspended permanent magnetic maglev train is more complicated and the parameters are more unstable than those of other trains, the traditional speedtracking algorithm has large tracking errors, frequent controller output changes, high energy consumption, and decreasing the passengers' riding comfort. To improve the shortcomings of the traditional automatic train operation (ATO) control algorithm, this paper proposes a predictive fuzzy proportional-integralderivative control algorithm with weights (WM-F-PID). The main contribution of this work is to propose a cascaded predictive fuzzy PID (F-PID) control algorithm architecture with weights and use an improved steepest descent method to calculate online the weight of the F-PID controller input occupied by the predictive controller output. Compared with the proportional-integral-derivative (PID), F-PID, model predictive control (MPC), and simple cascade predictive fuzzy PID (M-F-PID) control algorithms, this control algorithm effectively improves train tracking accuracy and comfort and reduces train energy consumption and stopping errors.INDEX TERMS Suspended permanent magnetic maglev train, WM-F-PID control algorithm, Online optimization algorithm, Speed-tracking.

In this study, an adaptive linear active disturbance rejection control is proposed to achieve steady levitation of the magnetic levitation ball.The proposed algorithm was designed and its convergence was proven via derivation. It can address the difficulty in parameter tuning of the controller and realize the real-time self-adaptive optimization of parameters. Besides,to verify the effectiveness of the proposed parameter tuning strategy,we analysed the anti-interference ability and tracking effects of SMC, LADRC and A-LADRC in different cases. Finally, the results showed that the anti-interference ability of A-LADRC is stronger than SMC and LADRC. In addition,the anti-interference effect intensifies when the interference increases. The designed controller also performs the best tracking performance among these three controllers. Therefore, A-LADRC exhibits better robustness and dynamic performance.