Recurrent Iteration Model Predictive Current Control for Linear Motor Drives in CNC Machine Tools Based on Cartesian Distance Minimization

With the increasing demand for high-speed and high-precision motion control in CNC machine tools, permanent magnet linear synchronous motors (PMLSMs) have been widely adopted in feed drive systems due to their excellent dynamic performance and positioning accuracy. However, traditional model predictive current control (MPCC) methods suffer from high computational burden and strong dependency on accurate motor models, which limit their practical applications in industrial environments. To address these issues, a recurrent iteration MPCC for PMLSM drives based on Cartesian distance minimization principle is proposed in this paper. First, a recurrent iteration MPCC inspired by idea of recurrent neural network(RNN) is introduced into PMLSM control through cyclically using last sampled current and predicting next period current, which reduces dependence on inductance. The voltage vector deviations between inverter voltages and machine steady state voltage vectors are analyzed. The dependence on resistance can be reduced by analyzing the correlation between ideal current vector and voltage vector deviation. Secondly, the selection of optimal voltage vector is transformed to the minimization of Cartesian distance between reference voltage vector and calculated 19 ideal voltage deviations in the proposed virtual voltage vectors hexagon. In order to decrease the computation burden, the selected voltage vectors deviation are divided into 8 regions and chosen from only 2 derived voltage vector deviations. Finally, experimental results show that the proposed method can effectively improve control performance under the parameter mismatches and reduce the computation burden.