Data-Based Active Vibration Control Strategy for Ultralow Frequency Isolation System Considering Sensor Noise Measurements

The measurement accuracy of ultra-precision instruments such as large-scale scanning electron microscopes (SEM) is significantly affected by ground micro-vibrations (typically at the VC-D or VC-E level). The ultra-low micro-vibration levels and low signal-to-noise ratios impede the accuracy of practical models, posing significant challenges for control system implementation. This paper presents a data-based hybrid control strategy for ultra-low frequency vibration isolation to address the sensor noise and model dependency issues that affect the precision of control signals. To reduce the noise proportion in the measurement reference signals, a system state observer is developed to recursively update state estimates using sensor measurement information. A closed loop feedback controller based on model predictive control is designed that incorporates the system’s relative displacement, balancing resonance suppression and stability. To further suppress in-band vibrations, a data-driven feedforward control strategy was constructed, which incorporates a concise and stable generalized FIR filter for online parameter identification and leverages the rapid convergence of the Gauss-Newton iterative algorithm, thereby effectively over-coming the reliance on accurate prior parameters. Simulation and experimental results demonstrate that the proposed hybrid control strategy effectively suppresses vibrations at the system’s natural frequency, constrains relative displacement within 5 μm, and significantly reduces the vibration transmissibility in the 0.8-20 Hz frequency range. This research shows tremendous potential for engineering applications in micro-vibration isolation for ultra-precision microscope measurement equipment.