Application issues of virtual reference feedback tuning and non-iterative correlation-based tuning

Virtual reference feedback tuning method and Non-iterative correlation-based tuning method both belong to the data-driven methods and aim to get the controller parameters from input and output data. This paper proposed to raise some practical issues we meet when using these methods. If the input signal includes noise and an ideal second order transfer function is used as a reference model, we should carefully choose the input signal, the reference model and the discrete method for this reference model. In addition, we proposed to apply the least square method with constrains to the virtual reference feedback tuning method to avoid getting parameters which are without practical meaning or out of range. Finally, we will simulate the results in Matlab and apply them to a motion control system driven by a linear motor.