To tackle the problem of reduced accuracy of model prediction current control (MPCC) response of the permanent magnet synchronous motor (PMSM) caused by parameter mismatch, this paper proposes an MPCC method with improved duty cycle. First, the duty cycle is added to the single-vector MPCC, and a duty cycle control model is established; secondly, a motor parameter identification algorithm is designed based on the Adaline neural network and least mean square (LMS) algorithm; and finally, identification parameters are applied to the duty cycle control to achieve the real-time update of model parameters. Through the system simulation verification, the algorithm reduces the current fluctuation and steady-state error when the motor parameters change, and effectively improves the current response accuracy.