基于改进容积卡尔曼的PMLSM无传感控制研究

Research on PMLSM Sensorless Control Based on Improved Cubature Kalman Filter

在永磁直线同步电机(PMLSM)的无传感器控制系统中,需要对动子的位置和速度进行实时状态估计。针对标准的容积卡尔曼滤波算法在PMLSM无传感控制中存在状态协方差矩阵易失去非负定性,以及噪声统计特性未知时变导致的滤波精度降低甚至发散的问题,提出一种容积卡尔曼滤波的改进算法。该算法结合平方根滤波和改进的渐消型记忆时变噪声估值器特点,能够保证滤波过程中状态协方差阵的非负定性,同时具有应对噪声变化的自适应能力。在永磁同步直线电机的无传感控制仿真实验中,改进的CKF算法能够明显提高标准CKF的滤波精度,在速度跟踪性能上,负载突变前、后的最大跟踪误差百分比分别为0.428 6%、0.146 8%,稳定跟踪后的跟踪误差百分比稳定在0.045 7%。

In the sensorless control system of permanent magnet linear synchronous motor(PMLSM), it is necessary to estimate the position and speed of the rotor in real time. Focusing on the problem that the covariance matrix tends to lose positive definiteness and noise statistical characteristics are inaccurate, which results in inaccurate filtering or even filter divergence in the PMLSM sensorless control system, an improved CKF algorithm is proposed. The algorithm combines the characteristics of square root filter and improved fading memory time-varying noise estimator to ensure the non-negative of the state covariance matrix and has ability to adapt to the noise changes. Simulation results show that compared with the CKF in the PMLSM sensorless control system the proposed improved CKF algorithm can improve the filtering accuracy obviously. In terms of speed tracking performance, the max tracking error percentages before and after the load mutation are 0.428 6% and 0.146 8% respectively. The tracking error percentage after stable tracking is around 0.0457%.