基于MARS的无速度传感器矢量控制系统研究

Speed Sensorless Vector Control Control Based on MRAS

应用模型参考自适应方法,对无速度传感器矢量控制系统的转速估计进行研究,针对常规速度辨识器中的基准模型易受积分初值和漂移问题的影响,造成辨识结果不准确的问题,对传统的MRAS方法进行了改进研究,并且以此为基础,采用基于神经网络的模型参考自适应辨识电机转速。MATLAB仿真结果表明,应用这种估计算法计算量小,收敛速度快,估计转速能很好地跟踪实际转速,而且具有很好的鲁棒性。

This paper discusses the sensorless vector control with a method of model reference adaptive system. The normal model can always be affected by the initial value and drift of integral calculation in the conventional speed identifier. This paper presents a new normal model and an adjustable model, and the motor speed is identified by using the neural network. The MATLAB results show that the convergence of the estimation is fast, and the speed is estimated accurately.