Research on nonlinear optical noise suppression in dual-differential detection optical path based on deep learning...

Area of Science:

• Quantum Precision Measurement
• Atomic Physics
• Optical Sensing

Background:

• Spin-exchange relaxation-free (SERF) comagnetometers are essential for high-precision angular velocity measurements.
• Long-term stability is limited by polarization noise (temperature drifts) and common-mode noise (photodiode mismatches).
• Current mitigation techniques offer only partial solutions to these noise sources.

Purpose of the Study:

• To develop a robust framework for mitigating nonlinear photonic noise in SERF comagnetometers.
• To enhance the long-term stability and precision of angular velocity measurements.
• To address polarization and common-mode noise challenges.

Main Methods:

• A dual differential compensation framework integrating deep learning techniques.
• Utilizing a Jones matrix model to analyze nonlinear noise coupling.
• Employing a hybrid Temporal Convolutional Network (TCN)-Long Short-Term Memory (LSTM) network for dynamic photodiode response calibration.

Main Results:

• Achieved a one-order-of-magnitude improvement in comagnetometer stability over a 2-hour period.
• Reduced the standard deviation of the differential output from 4.38 × 10⁻³ to 2.00 × 10⁻⁴.
• Demonstrated effective mitigation of nonlinear photonic noise.

Conclusions:

• The proposed dual differential compensation framework offers a robust solution for enhancing SERF comagnetometer stability.
• Deep learning integration provides dynamic calibration capabilities for improved noise reduction.
• This advancement is critical for ultra-high precision quantum sensing applications.