Sensor Head Temperature Distribution Reconstruction of High-Precision Gravitational Reference Sensors with Machine...

Area of Science:

• Physics
• Engineering
• Computer Science

Background:

• Temperature fluctuations critically impact high-precision gravitational reference sensor performance.
• Direct temperature measurement of sensor heads is often infeasible due to space constraints and sensor complexity.

Purpose of the Study:

• To develop and evaluate a high-accuracy interpolation method for reconstructing sensor head surface temperatures.
• To assess the performance of the XGBoost-LSTM method in ground-based and on-orbit scenarios.

Main Methods:

• Utilized a hybrid XGBoost-LSTM model for sensor head temperature reconstruction.
• Simulated and analyzed method performance under ground-based and on-orbit conditions.

Main Results:

• Achieved temperature reconstruction precision two orders of magnitude higher than conventional methods.
• Demonstrated one order of magnitude improvement over a Backpropagation (BP) neural network.
• Exhibited remarkable stability and robustness in simulations.

Conclusions:

• The XGBoost-LSTM method meets the Taiji Program's payload temperature control precision requirements.
• Provides essential data support for thermal noise modeling and subtraction in gravitational reference sensors.