Research On Application Of Wavelet Analysis In Vehicle Positioning And Orienting System Using Laser Gyro

The dissertation mainly investigates the application of wavelet transformation in ALV positioning and orienting system. Combined with forward linear prediction and Kalman filtering, wavelet transformation was used to preprocess the outputs of dither laser gyroscopes and accelerometers and further applied to initial alignment of SINS and GPS/DR integrated navigation.The main content of the dissertation is following:(1) The problems of signal multiscale de-noising and dynamic system multiscale estimation were investigated. An improved recursive de-noising method using threshold was raised reckoned on the state-of-art of present algorithms. According to the theory of wavelet multiscale analysis, a multiscale forward direction linear prediction algorithm based on the convolutional wavelet transform has been put forward. Based on the theory of multiscale Kalman estimation for the dynamic system, a multiscale Kalman filtering scheme with real-time decomposition and estimation has been designed, and a method based on wavelet transform has been proposed to preprocess measurement noise.(2) With respect to dither laser gyroscopes outputs in various environments, the selection of wavelet bases and the best scale for decomposition was studied.(3) According to different properties of the wavelet coefficients between noise and signal, a detection algorithm based on products of multiscale wavelet coefficients was proposed. Consequently, the above results were applied to the two problems of ALV initial alignment and GPS/DR integrated navigation. The main work consists of the following aspects:(1) The wavelet detection algorithm was applied to initial alignment of SINS on the stationary base and the experiment results on ALV showed that wavelet detection algorithm can detect jamming commendably. Multiscale FLP and multiscale Kalman filtering were applied to preprocess dither laser gyroscope signals. The experiment results on the three-axis table and ALV showed that both multiscale FLP and multiscale Kalman filtering can depress the noise efficiently. Multiscale Kalman filtering was applied to fine alignment, experiment results on ALV showed that it can further improve the performance of azimuthal estimation.(2) Multiscale Kalman filtering was applied to DR autonomous navigation and GPS/DR integrated navigation and the algorithm based on multiscale Kalman filtering was designed for dead reckoning of autonomous and integrated navigation. Results on hardware-in-the-loop simulation and ALV experiment showed that the multiscale Kalman filtering further can improve the precision of dead reckoning.(3) This dissertation suggested a method for DR error estimation based on wavelet multiscale analysis. Experiment results on ALV demonstrated that substantial improvement in DR accuracy could be obtained by applying wavelet multiscale analysis.