一种自适应扩展卡尔曼滤波的永磁同步电机无位置传感器矢量控制

Sensorless vector control of permanent magnet synchronous motor based on adaptive extended Kalman filter

在粗差干扰或噪声统计偏差情况下,扩展卡尔曼滤波(EKF)对永磁同步电机(PMSM)的速度和转子位置估计存在精度下降问题,为此提出一种基于新息序列的自适应扩展卡尔曼滤波算法(AEKF)。首先,将粗差干扰加入系统观测方程中,分析粗差干扰对系统观测精度的影响。其次,为增强算法的抗扰性能,在新息协方差计算中设置加权系数,通过调整临近时刻的新息协方差阵权重,计算出新息协方差值,并更新到卡尔曼增益的计算。最后,建立AEKF数学模型,并对比粗差干扰与噪声统计出现偏差情况下,AEKF与EKF两种策略的观测性能。仿真和实验结果表明,在粗差干扰或噪声统计信息出现偏差情况下,AEKF算法对永磁同步电机转速的观测具备更强的鲁棒性及更高的预测精度。

In the case of gross error interference or noise statistical deviation,the accuracy of extended Kalman filter(EKF)in speed estimation and rotor position prediction of permanent magnet synchronous motor(PMSM)is decreased.An adaptive Kalman filter algorithm based on innovation sequence was proposed.First,the gross error interference was added to the system observation equation,and its influence on the observation accuracy was analyzed.Secondly,in order to strengthen immunity of above system,the weighting coefficient was set in the innovation covariance calculation to complete the calculation of innovation covariance matrix by adjusting the weight of the innovation covariance matrix at the adjacent time,and import the value into the Kalman gain calculation.Finally,the AEKF mathematical model was established based on the above steps,and the observation performance of AEKF and EKF was compared under the condition of deviation of gross interference and noise statistics.Simulation and experimental results show that AEKF algorithm has stronger robustness and higher prediction accuracy for PMSM speed under the interference of gross error or noise statistics.