A Novel Gravity Compensation Method for High Precision Free-INS Based on "Extreme Learning Machine".

Area of Science:

• Geodesy and Geophysics
• Navigation and Control Systems

Background:

• High-precision inertial sensors have highlighted gravity disturbance as a key factor impacting inertial navigation system (INS) accuracy.
• Existing gravity compensation methods may not adequately address the complexities of gravity variations in diverse terrains.

Purpose of the Study:

• To investigate the effect of gravity disturbance on INS performance.
• To propose and validate a novel gravity compensation method for high-precision INS.
• To enhance the estimation and compensation of gravity disturbances for improved navigation accuracy.

Main Methods:

• Utilized extreme learning machine (ELM) to estimate gravity disturbance based on geoid gravity data.
• Applied upward continuation to process gravity disturbance data to the INS operating height.
• Integrated compensated gravity disturbance into INS error equations to mitigate error propagation.

Main Results:

• The ELM method demonstrated improved gravity disturbance estimation accuracy, with root mean square error (RMSE) reductions of 23% (plain) and 44% (mountain) compared to bilinear interpolation.
• Field experiments confirmed significant positioning accuracy improvements: 13% in a plain area and 29% in a mountain area over 2-hour trials.
• The proposed method effectively restrains INS error propagation caused by gravity disturbances.

Conclusions:

• The novel ELM-based gravity compensation method offers a substantial advancement for high-precision INS.
• The method proves effective in diverse geographical conditions, enhancing navigation system reliability and accuracy.
• This research provides a robust solution for mitigating gravity-induced errors in INS applications.