A modified adaptive notch filter for resonance suppression in flexible mechatronic systems

Abstract In the mechatronic systems, the mechanical resonance caused by the flexibility will reduce the control accuracy and reliability. To suppress the resonance and improve system's robustness, a modified adaptive notch filter (ANF) is proposed in this paper. Considering the controller's saturation nonlinearity, different limit cycle oscillation phenomena are analyzed firstly, which can be distinguished into two resonance cases. To cope with different resonance cases, the proposed ANF detects the oscillation and estimates the frequency of the oscillating velocity signal to iteratively tune the notch filter online. As the frequency and amplitude of the oscillating signal vary in a large range, a gain adjust factor is added to modify the conventional frequency estimator guaranteeing a fixed settling time regardless of the initial condition and the frequency, amplitude of input. The simulation and experiment results verify that the modified frequency estimator has better robustness compared to the conventional method and the proposed ANF can cope with different resonance cases.