This paper proposes a real-time speed identification method by using a symmetric strong tracking extended Kalman filter (SSTEKF) for induction motor sensorless drive. In SSTEKF, the residual sequences are forced orthogonal to each other, and the gain matrix is tuned in real-time by introducing fading factors into the covariance matrix of the predicted state. The modeling error is reduced, and the mutational state is tracked rapidly based on SSTEKF. Simultaneously, the Cholesky triangular decomposition is used to change the working way of the multiple fading factor matrix in the error covariance matrix. The application of the Cholesky triangular decomposition guarantees that the error covariance matrix is symmetric in the process of iteration, and the stability of the algorithm is enhanced. Therefore, the estimation accuracy, the tracking speed, and the noise suppression of the proposed method are better than the EKF. The correctness and effectiveness of the proposed method are verified by experimental results.

中文翻译：

---

基于对称强跟踪扩展卡尔曼滤波器的感应电机驱动无传感器控制，以减少建模误差


本文提出了一种利用对称强跟踪扩展卡尔曼滤波器（SSTEKF）的感应电机无传感器驱动实时速度识别方法。在SSTEKF中，残差序列被强制彼此正交，并且通过将衰落因子引入到预测状态的协方差矩阵中来实时调整增益矩阵。基于SSTEKF，减少了建模误差，快速跟踪突变状态。同时，利用Cholesky三角分解来改变误差协方差矩阵中多重衰落因子矩阵的工作方式。 Cholesky三角分解的应用保证了迭代过程中误差协方差矩阵的对称性，增强了算法的稳定性。因此，该方法的估计精度、跟踪速度和噪声抑制均优于EKF。实验结果验证了该方法的正确性和有效性。