This paper investigated a hybrid excitation flux switching linear magnetic suspension motor (HEFSLMSM), which is used in maglev trains. The extended state observer of the original active disturbance rejection control (ADRC) algorithm, designed using the traditional fal function, often leads to issues such as chattering, slow convergence speed and excessive overshoot. To improve the performance of suspension system, an improved ADRC strategy was proposed. Based on the special principle and operation mechanism of HEFSLMSM, the mathematical model of the system was derived, including the motor excitation circuit voltage equation, the magnetic suspension force equation and motion equation. A new sfal function, which met the criteria of “small error, large gain, large error, small gain, continuous smoothness, differentiability everywhere, symmetry at the origin,” was designed to replace the original fal function. An improved ADRC simulation model was established and compared with ADRC and proportional integral devivative controllers through simulation experiments. The simulation results showed that the HEFSLMSM maglev system using the improved ADRC exhibited significant advantages in dynamic performance, effectively suppressing various uncertain disturbances and ensuring the system stability and accuracy.