Abstract
Based on the test data of a fiber optic gyroscope(FOG) in full temperature and angular rate range, it is demonstrated that the characteristic of FOG temperature drift is nonlinear. A novel comprehensive compensation approach using Gaussian process regression is proposed to improve the full-temperature-range adaptability of FOG. This approach establishes directly the mapping relationship between the observable information (containing the temperature and the actual output) and the ideal linear output of FOG by choosing a suitable covariance function and optimizing its hyperparameters. Our approach overcomes the shortage of the traditional method which introduces twice compensation errors caused by modeling the zero bias and the scale factor respectively, so the compensation precision is improved and the compensation process is simplified. Compared with the traditional method and the least square support vector regression, the experimental results demonstrate that the model established in the current paper can reflect the characteristic of the temperature drift of FOG more accurately, and can obtain better generalization ability and higher compensation precision. The predicted root mean square error of the new model is less than 0.003(o)/s and the nonlinearity error of FOG is reduced from 10 to 10, which means the new model can effectively improve the insensitivity of FOG output to the temperature..
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call