Temperature Drift Compensation For FOG And Research On Testing Technique For Fiber Optic Coil

Fiber optic gyro has special advantage and is used more and more extensive. For the application of FOG, it has to be promised to have the widly operating temperature range, but the FOG system is very sensitive to the temperature variation in reality. From the curve of the temperature on measured, it can be seen that the zero point drift of FOG is obviously intensified.In this paper, after building up the model of the temperature drift error and the further research on the engineering application of temperature drift compensation technology, it is analyzed and tested to the character of important units in FOG system, such as the optic fiber coil.Firstly, it is given the source, the research background and significance of the topics, then analyzed the research situation and development trend, the mechanism of how and why the temperature could have the impact on the componts of FOG system is drawn theoretically.Secondly, the major error induced by the the optic fiber coil of the fiber optic gyro system was researched, and the nonreciprocal phase drift because of thermal effect in the QAD method was analyzed. The optic path for the test was designed and constructed and the characteristic of optic fiber coil was tested as well. The feasibility of the testing method was validated through the tests.Thirdly, there is usually three aspects which are influenced by the the start-up drift and the temperature variation in FOG system. A lot of tests to the sefl-made gyro in the lab has accomplished and the linear regression temperature model is set up with the suitable model parameters after data processing. It can be said that the model is correct and effective.Finally, through the theoretical analysis based on RBF neural network model of the gyro temperature, using a wavelet analysis of the trend of extraction of gyro drift as model parameters, on the basis of a large number of experments, the model network is trained with the test data and the the network hidden layer nodes and the threshold value is determined, then a good network which has been trained is used to compensate the gyro temperature drift to verify the correctness and effectiveness of the model.