Residual-triggered adaptive strong tracking Kalman filter for acousto-optic modulator-based laser power stabilization

High laser power stability is critical for precision interferometry and laser frequency stabilization. Acousto-optic modulators (AOMs) are widely used for laser power control due to their fast response and high linearity; however, their diffraction efficiency is sensitive to environmental temperature variations and self-heating, limiting long-term stability. This work presents a closed-loop laser power stabilization scheme based on a residual-triggered adaptive strong tracking Kalman filter (RT-ASTKF), operating without temperature-control hardware. The algorithm employs a fixed process-noise covariance together with adaptive measurement-noise estimation, enabling real-time compensation of AOM thermal drift and circuit noise. The system is implemented on an STM32H743 platform for synchronous acquisition and RF gain control. In a 13.9-h experiment, the closed-loop system achieves an RMS stability of 0.0615% and a peak-to-peak fluctuation of 0.391% (following 3σ outlier removal), while the Allan deviation remains at the 10 −8 –10 –9 level for integration times of 10 2 –10 4 s. The open-loop periodic drift (4.363% peak-to-peak, 0.3780% RMS) is suppressed to 0.391% peak-to-peak and 0.0615% RMS, corresponding to 11.16-fold peak-to-peak and 6.15-fold RMS reduction (approximately 20.9 dB and 15.8 dB, respectively). These results demonstrate effective suppression of compound disturbances without additional thermal-control hardware.