Performance-based Adaptation Termination for Preventing Parameter Drift in Adaptive Vibration Suppression
Parameter drift remains a practical limitation of adaptive vibration control systems, particularly when persistence of excitation diminishes after disturbance attenuation or when measurement disturbances render the adaptation problem ill-conditioned. This issue is especially relevant in flexible structures, where adaptive controllers may continue updating parameters even after satisfactory vibration suppression has been achieved. This paper proposes a computationally efficient root-mean-square (RMS)-based stopping criterion that mitigates parameter drift by freezing adaptation once satisfactory vibration attenuation has been sustained over a prescribed interval. The criterion is implemented by recursively computing an exponentially weighted moving RMS of the performance variable and requires negligible additional computational effort. The proposed mechanism is integrated with retrospective cost adaptive control (RCAC) and validated through both numerical simulations and closed-loop experiments on a cantilever-beam vibration-suppression platform with a noncollocated actuator configuration. Without the proposed stopping criterion, continued adaptation after disturbance rejection leads to gradual parameter drift and degradation of vibration suppression performance. When the RMS-based threshold is enabled, controller parameters remain bounded, and the achieved vibration attenuation is preserved. The results demonstrate that a simple performance-based monitoring mechanism can effectively prevent parameter drift in adaptive vibration control, retain the transient benefits of adaptation, and incur minimal computational overhead.