Adaptive Strong Tracking Alman Filter Application In Gyro-stabilized Platform

Gyro stabilized platform (GSP) is a core equipment in the inertial navigation, guidance and measurement system based on the core inertial component-the gyroscope, mounted on a moving carrier, and can isolate the carrier disturbance and ensure the pointing of the visual axis (LOS) of the device mounted on the platform stable. Due to the impact of the external environment and other reasons such as the structure errors, as an angular rate sensor, the gyroscope in the platform can bring drift and noise in measurements, which can affect the visual axis stability and accuracy. So the design of the filter for gyroscope output signal is one of the major tasks in gyro-stabilized platform studies. Kalman filter is the optimal estimation theory which describes dynamic system by state equation. The estimation and filtering are in the form of prediction and correction. The linear minimum variance estimation criteria is adopted to estimate the system state, which has the advantages of real-time recursion, small storage need, simple and easy design.The research of this dissertation is to apply adaptive strong tracking Kalman filter, which has a stronger robustness to the uncertainties, to filter the gyroscope output signal in the velocity loop of gyro-stabilized platform. The velocity loop control accuracy is much improved. The research work includes the following two aspects:1) Based on of the nonlinear friction model of velocity loop in gyro-stabilized platform, an adaptive strong tracking Kalman filter is designed to filter the gyroscope output signal by using a first order linear nominal model obtained after the full feedforward compensation of the nonlinear friction. Considering the actual velocity loop has different model parameters when working forward and in reverse respectively, the parameters of the nominal model take the geometric mean values of the parameters when the velocity loop works forward and in reverse, which simplifies the algorithm design. The error between the actual model and the nominal model is considered as the process disturbance, whose statistic is estimated by the Sage-Husa time-varying process disturbance statistic estimator for the following state estimation, so as to improve the state estimation accuracy. Moreover, in order to enhance the algorithm stability and further improve the algorithm’s robustness to systematic errors, the strong tracking filter algorithm is also combined to correct the prediction variance in real time to minimize the autocorrelation function of the output error sequence. Finally, simulation experiments are performed and the results are analyzed.2) In both the cases of full and half feed-forward compensations of non-linear friction of velocity loop in gyro-stabilized platform, the adaptive strong tracking Kalman filter designed is combined with the two control schemes of PI control and model reference adaptive control (MRAC) for the velocity loop control system. Considering the gyroscope output noise, comparison of the performance of the two control systems before and after filtering is made. Analysis via comparative simulation experiments show that the model reference adaptive control plus strong tracking adaptive Kalman filter is more effective to further improve the velocity loop control accuracy.