The Study On The Tracking Control System Identification Technology

Tracking control systems have been widely used in areas such as industry, agriculture, defense, astronomy and so on, and the system’s mathematical model of controlled object is the basis for control.In general, the controlled object model is often unknown or semi-unknown.Because when the internal mechanism of the system is so complicated that the model structure and the corresponding parameters of the controlled object is difficult to use existing knowledge and methods to determine. Even when the environment where the system stays changes or the system itself fails,the controlled object model structure and parameters will change. So it is very important to obtain tracking control system mathematical model of controlled object timely and accurately.Currently, the main method for tracking control system to obtain mathematical model of controlled object is firstly using dynamic signal analyzer to obtain the frequency characteristics of controlled object at 30 k Hz sampling frequency and then using MATLAB toolbox to identification the transfer function of it.This method which with the help of special instruments is not conducive to tracking controlsystem’s integration and also not making full use of the system’s hardware and software facilities. The purpose of this study is to use hardware and software of tracking control system itself to identification the controlled objecttransfer function model at low sampling frequency.Firstly, this paper analyzes the composition of the tracking control system, and on this basis,establishes controlled object of precisiontracking system the fast-mirror system’s mathematical model. Analyze the importance of designing system identificationexperiment and set forth requirements to identificationexperiment in selecting identification input data, sampling period andpreprocessing data.Next, study the basic theory of time-domain identification, and take it as a basis forstudy and compare M-sequence and Invert-Repeated M-Sequence. Studies have shown that Reverse-M-sequencebecause of its self-correlation function does not include the DC component is more suitable as identificationexcitation signal.Study the principles of the relevant identification method as taking Invert-Repeated M-Sequence as excitation signal and analyze Reverse-M-sequence’s parameters selection.Study the principle of Hankel matrix method which can convert impulse response into transfer function model.Simulate time-domain identification method based on Reverse-M-sequence and verify the validity and feasibility of the method.Then, this paper study the significance of the frequency-domain identification and frequency characteristics, on this basis, research frequency characteristics identification method based on steady-state sinusoidal and swept sine and carry them out to identification simulation.Simulation results show that the method based on swept sinedue to its low sampling frequency, a small amount of computation and high recognition accuracy is more suitable as an integrated identification method in tracking control system.Then study four methods to obtain the transfer function model from the frequency characteristics: basic Levy method, fixed Levy method, iterative Levy method and MATLAB GUI method, and simulation. The simulation shows that fixed Levy methodwith high identify accuracy, no request for setting identify parameter’s initial values, general algorithm complexity is an ideal method for tracking control system.Finally,build a faster-steer-mirror system experimental platform based on PC/104 and program in the controller with Vx Works environment. Firstly, take experiment on the time-domain identification method based on Invert-Repeated M-Sequence and verify validity and anti-jamming capability of this method. Then take experiment on frequency-domain identification method based on swept sine and analyze the impact of sensor errors for identification results.Experimentresults show that the time-domain method based on Invert-Repeated M-Sequence which has strong noise suppression capability, high identify precision and short identify time is an ideal identification method for tracking control system.