Research On The Wavelet De-noising And Error Modeling Of Gyro Drift Signal

With the development of inertia technology, the required precision of the gyroscope has been much higher. Gyro drift is an important target that can affect the precision of gyroscope, so the research on gyro drift filtering and modeling is much more important. In this thesis, on the basis of wavelet and multi-wavelet de-noising theory and time series analyzing method, the wavelet de-noising method, which is applied to the gyro drift signal processing, is studied, and the time series model and the dynamics model of the gyro drift signal are built. The experiment results of the gyro signal proved that the methods proposed in this paper have acquired inspiring results in improving the precision of the north-finding system.Firstly, based on the selection of the wavelet bases of gyro drift signal, bi-orthogonal wavelet and redundant wavelet are used in gyro signal processing systematically. Compared with the classic Butterworth filter, the results of wavelet filtering are much better. As for the pulse noise in gyro output signals that are collected from bad background, a median filter based on wavelet is designed. Based on these, further more research is presented in this paper. Multi-wavelet theory is introduced in gyro signal de-noising in the thesis, in which the virtues and drawbacks are given through analyzing and comparing between the above methods and the multi-wavelet de-noising of gyro drift signal.Secondly, systematic and comprehensive statistic analysis of the gyro drift signal is presented in the paper, in which the application of trend terms elimination of wavelet in the gyro drift error modeling is also included. The time series model of gyro drift signal using the smooth wavelet and multi-wavelet to remove the trend terms is built in this thesis. Based on the above research extended adaptive Kalman filter is further studied on the gyro signal, a dynamic error model of gyro signal is presented.In the end, a visualized software system, which is developed under the environment of Visual C++ 6.0 based on the research results, has implemented the de-noising of gyro signal. The system consists of three main modules, such as wavelet de-noising, wavelet analyzing and the time series modeling of gyro signal and its north calculation of north-finding system. The kind graphic interface of the system can show the filter results, which provides the software platform for the same type of task research in the future with perfect functions.