基于模糊扩展卡尔曼的直线感应电机无速度传感器控制

Speed sensorless vector control of linear induction motors based on fuzzy extended Kalman filter

直线感应电机(linear induction motor,LIM)可实现无接触、无摩擦的直线运动,具有爬坡强、横断面小、低噪音等优势,在中低速磁悬浮与城市轨道交通中被广泛应用。其中,基于观测器的电机无速度传感器矢量控制方法,是解决速度传感器成本高、可靠性低、维护难等问题的有效途径。针对LIM速度辨识受系统内外扰动影响导致观测精度下降问题,提出基于模糊扩张卡尔曼滤波(Fuzzy Extended Kalman Filter,FEKF)的速度观测方法,提高观测器鲁棒性。首先在αβ坐标系下建立考虑动态边端效应的LIM数学模型,利用LIM的实时电流和电压作为输入信号,以电流、磁链和速度作为状态变量,推导出基于扩展卡尔曼滤波(extended kalman filter,EKF)观测器电机状态空间方程的离散模型,利用EKF对LIM的速度和磁链进行在线观测,并用于实现LIM的矢量控制。其次,为提高EKF的噪声协方差矩阵Q对LIM系统内外扰动的鲁棒性,引入模糊控制方法对矩阵Q进行自适应调整,实现速度与磁链的估计精度。具体通过计算系统理论残差与实际残差的偏离程度得到噪声调节因子,实时调整噪声协方差矩阵Q,并在下一周期更新模糊扩展卡尔曼的反馈增益矩阵,提高对速度与磁链的估计精度。之后对系统的能观性与FEKF的收敛性进行分析,探明矩阵Q中各参数的边界及整定规律。最后通过仿真和硬件在环实验验证所提FEKF优化算法的有效性,结果表明FEKF能有效提高对LIM速度辨识准确性,实现LIM无速度传感器矢量控制。

Linear induction motor(LIM)has been actively applied in middle-low speed maglev and urban rail transit due to its strong climbing ability,simple structure and low maintenance cost.Among them,the observer based motor speed sensorless vector control method is an effective way to solve the problems of high cost,low reliability,and difficult maintenance of speed sensors.To address the problem that LIM velocity discrimination was affected by the internal and external disturbances of the system,the Fuzzy Extended Kalman Filter(FEKF)based velocity observation method was proposed to improve the robustness of the observer.Firstly,a mathematical model of LIM was developed in the coordinate system to take into account the dynamic side effects.The extended kalman filter(EKF)was used for online observation of the speed and magnetic chain of the LIM and for vector control of the LIM.Secondly,in order to improve the robustness of the noise covariance matrix Q of the EKF to perturbations inside and outside the LIM system,a fuzzy control method was introduced to adaptively adjust the Q matrix to achieve the estimation accuracy of the speed and magnetic chain.Specifically,the noise adjustment factor was obtained by calculating the degree of deviation between the theoretical and actual residuals of the system.So that the noise covariance matrix Q can be adjusted in real time,and the feedback gain matrix of the fuzzy extended kalman filter(FEKF)was updated in the next cycle to improve the estimation accuracy of the speed and magnetic chain.Afterwards,the energy observability of the system and the convergence property of the FEKF were analysed to explore the bounds of each coefficient in the matrix Q.Finally,the validity of the proposed FEKF algorithm was verified through simulations and hardware-inthe-loop experiments.The results show that the FEKF can effectively improve the accuracy of LIM speed identification and implement LIM speed sensorless vector control.