永磁同步电机多步模型预测电流控制球形编码研究

Multi-step model predictive current control of permanent magnet synchronous motor based on sphere decoding

中文关键词  永磁同步电机  有限控制集  多步预测  球形编码  事件触发

英文关键词  permanent magnet synchronous motor  finite control set  multi-step prediction  sphere decoding algorithm  event triggering

中文摘要

      传统多步模型预测电流控制遍历所有开关序列, 并寻求成本函数最小, 使得运算量较大. 球形编码算法将成本函数转换为开关序列对应的矩阵二范数平方, 并采用事件触发机制动态精简计算量. 仿真结果表明: 基于球形编码的多步模型预测电流控制性能良好, 与传统方法完全等价, 控制效果相当. 基于STM32H743单片机平台, 球形编码和传统方法单控制周期执行时间实验结果表明: 对于多步预测, 球形编码算法可减少单控制周期执行时间, 2步预测减小至96.78%, 3步预测减小至87.99%, 4步预测减小至73.41%, 5步预测减少至63.63%, 在控制性能与传统方法相当的条件下, 提高系统实时性能

英文摘要

      Traditional multi-step model predictive current control (MPCC) of permanent magnet synchronous motor (PMSM) traverses all switch sequences and seeks the minimum cost function, resulting in large calculation. The sphere decoding algorithm transforms cost function to the square of the matrix two norm corresponding to the switch sequence and uses the event triggering mechanism to simplify calculation. Simulation results show that under the control of MPCC based on sphere decoding, the PMSM works properly. The sphere decoding is equivalent to the traditional MPCC. The single control cycle execution time experiment of sphere decoding and traditional MPCC is verified by the STM32H743 micro control unit platform. And real-time experimental results show that for multi-step prediction, the sphere decoding algorithm can reduce the execution time of single control cycle. For two-step, it is reduced to 96.78%, for three-step to 87.99%, for four-step to 73.41% and for five-step to 63.63%. Therefore, the sphere decoding can improve the real-time performance of the system with almost the same control performance as the traditional method