A novel model free adaptive high-order sliding mode control strategy based on data-driven extended sliding mode observer is proposed to address the problem of PMSM drive system over reliance on accurate models and poor robustness in the face of load disturbances. Firstly, convert the motor motion equation into a discrete partial form dynamic linearization model. Secondly, a new controller is constructed that integrates the advantages of partial format model free adaptive control and discrete-time high-order sliding mode control; Simultaneously design a data-driven extended nonsingular discrete terminal sliding mode observer to observe disturbances in real time and input them into the controller to compensate for tracking errors. Then, based on the motor output speed and input reference current data within the sliding time window, an improved pseudo ladder real-time estimation algorithm is constructed to enhance the tracking ability of time-varying parameters, and achieve data-driven control based on the second-order partial range model. Finally, through simulation and experimental comparison with traditional methods under sudden changes in operating conditions, the results show that this method can shorten the convergence time of the motor by 35% and reduce the average waveform distortion caused by load disturbances by 18.4%, effectively ensuring the stable and efficient operation of PMSM and verifying the reliability and superiority of the proposed method.