| Seeker system is a kind of high precision servo system. Due to the uncertain factors of the surroundings, the parameters of servo system structure are changeable, which can lead to the decline of the overall performance.This paper presents a method of offline detection and model modification used in seeker system.This paper analyzed the two-axis stabilized platform using dynamic analysis and established the state space expression of the system combining electromechanical control system. Then this paper analyzed the effect of the change of the parameters on the control system based on it. This paper will establish the precision model of the system through system identification because the changeable model is unknown. The identification mainly contains two parts:(1) This paper identified the seeker system using DFNN, and analyzed the structure, mathematical description and learning algorithm of it. The EKF algorithm is used to adjust the premise parameters, and the linear least square algorithm is used to update conclusion parameters. So this paper improved the learning algorithm of the parameters.(2)This paper established the related testing system based on the single-axis turntable and finished the testing experiment of the system. Then DFNN and improved DFNN were used to the identification simulation of the system. The result showed that the performance of the improved DFNN is better than the basic DFNN, and the obtained model using it can approximately describe the characteristics of the seeker system,such as the nonlinearity and the linearity. The precision of the obtained model is in the scope of the requirements.The end of this paper provided an optimization method of controller parameters. It combines the model reference adaptive control and the improved GA. The result of the simulation proved that this method can modify the control system model and make the performance of the whole system close to the reference model. Thus the optimization purpose is achieved. |
PDF Full Text Request