AR-based identification and control approach in vibration suppression

Abstract

In vibration suppression control with a wide bandwidth for a resonant plant, it is a requisite to identify the plant with high accuracy. However, the more complicated the plant is, the more difficult its system identification becomes. This paper proposes the identification of a modified plant instead of the original plant. The modified plant is a minor control loop (MCL), considering the original plant, when one of the output values is the control feedback. The MCL is designed in order to obtain an optimal damping factor of the modified plant. Moreover, the modified plant is identified based on the autoregressive exogenous (ARX) model and the least-squares method. Based on these techniques, it is possible to specify an exact uncertainty between the nominal and identified parameters of the ARX model. Therefore, a robust vibration suppression control system, which has a wide frequency band, can be systematically designed. The advantages of the proposed design method for a two-mass resonant system are demonstrated through simulations and experiments in the cases of the position and speed control.