Frequency response data based disturbance observer design applicable to non-minimum phase systems

Bandwidth Maximization of Disturbance Observer Based on Experimental Frequency Response Data

“…In regard to the aforementioned works, we integrated the frequency response data-based method into DOB optimization design, and the preliminary work which designed a second order Q filter was presented in [16]. In this paper, we confirm the previous findings and enhance its generality and performance.…”

“…Third, ω p and ω t set lower and upper bounds for the Q filter bandwidth frequency, and a 1 and a 2 should all be positive to guarantee the stability of the Q filter itself. While in our previous study [16], the damping factor of the Q filter ( a 1 2 √ a 2 ) was limited to [0.5, 1], we managed to remove this constraint in this paper, which adds more freedom in tuning parameter. In summary, the optimization problem can be formulated into the following form:…”

“…According to Eq. The optimized result of our previous study [16] is selected as an initial condition for optimization which is a 1 = 0.0838, a 2 = 1.76 × 10 −3 , ω p = 9.9 rad/s for both cases. The convex constraints are established using our proposed method, followed by optimization of the said problem using off the shelf toolboxes, Yalmip [21] and Mosek [22] in Matlab.…”

“…3. The requirement for the nominal plant and the damping factor of the Q filter have been mitigated compared with [16]. The proposed method is versatile for both minimum phase and non-minimum phase plants.…”

“…The selected initial disturbance observer [16] and the feedback controller C f b work in the system together.…”