High-end applications demand higher dynamic response and disturbance rejection in permanent magnet synchronous motors. However

harsh environments

complex conditions

and nonlinear parameter variations often cause model mismatch

affecting speed response. To address this

this paper proposes a model-free adaptive sliding mode control method based on an improved Kalman prediction and update algorithm. The method initiates by transforming the PMSM model into a compact-form dynamic linearization data model

thereby mitigating reliance on precise mathematical modeling. Concurrently

an improved Kalman algorithm is employed to perform real-time prediction and updating of the pseudo partial derivative

resolving its estimation dependency and compensating for data model inaccuracies. Building upon this

the control law is designed by integrating the updated data model with a global fast integral terminal sliding mode function based on a variable power reaching law. This ensures rapid state convergence. Simulation and experimental results demonstrate that

compared to PI control and conventional model-free adaptive control

the proposed method significantly enhances convergence speed

reduces chattering

and achieves efficient data-driven control.