Self-sensing electromagnetic shunt damper for adaptive vibration control

Abstract The self-sensing damper is an effective vibration suppression actuator, particularly in scenarios where installing additional sensors is prohibited. This study proposes a self-sensing electromagnetic shunt damper to improve the linear model-based velocity prediction accuracy and extricate the heavy dependence on external sensors. By introducing a gradient boosting regression (GBR) model with feature-engineered voltage signals, lagged derivatives, and polarity information for data-driven velocity inference, the prediction error is reduced by 14% versus the traditional GBR model. The model is optimized and pruned via cross-validated grid search to fit 32 KB microcontroller flash memory, enabling real-time computation. An integrated balance logic algorithm is then adopted for energy-efficient adaptive control with the adjusted load resistance based on the predicted velocity. Moreover, the experimental results with swept sinusoidal excitation confirm the high-precision velocity prediction accuracy and effective vibration suppression performance. This sensor-free and low-cost solution simplifies system architecture, reduces installation complexity, thereby holding great promise for broad applications in civil engineering, automotive engineering, and precision machinery.