<p>To improve the dynamics of the linear permanent magnet synchronous motor (LPMSM) with complex inductance characteristics, a novel predictive current control method for the linear motor (LM-PCC) is proposed in this article. The position-dependent parameter variations of the LPMSM, particularly the inductance matrix, cause performance degradation in conventional predictive control methods. To overcome this, a state-space equation incorporating the inductance matrix is first established, and the effects of self-inductance and mutual inductance mismatch on transient performance and stability are theoretically analyzed. Secondly, an inductance matrix estimation strategy that combines voltage injection and recursive least squares is proposed. By injecting voltage pulses that do not affect the motor’s operation and combining with the theoretically derived inductance model, the rank deficiency in the multi-parameter estimation is resolved, and real-time tracking of the inductance matrix is achieved. Furthermore, mutual inductance is considered in the ultra-local model to improve transient performance and suppress mutual coupling. Simulation and experimental validation on the LPMSM platform confirm the effectiveness of the proposed method.</p>