|Real-time Inverse Model Estimation by a Recursive Least Squares Method for Disturbance Observer-based Control Systems: Balancing Control of a Single-wheel Robot
Sang-Deok Lee and Seul Jung*
International Journal of Control, Automation, and Systems, vol. 17, no. 8, pp.1911-1920, 2019
Abstract : This article proposes a real-time identification and control technique for disturbance observer (DOB) – based control systems to improve the balancing control performance of a single-wheel robot. In a DOB-based control configuration, the inverse model of the system is obtained by inverting the identified forward model to extract the disturbance. However, a problem arises when the stability of the inverse model is not guaranteed. The Q filter design may solve the problem, but the process is time consuming. Therefore, we propose a stable inverse model identification technique with a minimal effort of designing Q filters in DOB schemes. The proposed scheme follows three steps: a recursive least square (RLS) method is used for updating the parameters of a second order model. Then the stability is checked. The last step is to make the model stabilized depending upon the stability through the all-pass-filter technique. After these steps, the transformed inverse model becomes stable and finally can be used for DOB control schemes. Experimental studies on balancing control of a single-wheel robot are conducted and their performances are compared to verify the proposal.
DOB, inverse model estimation, recursive least square, single-wheel robot, stability.