Study On Servo System Modeling Of Large Antennas Based On The Subspace Method

This project comes from a research on anti-gust-interference technology of large antennas. Due to the fact that large antennas are big in size, and susceptible to gust disturbance, conventional PID controllers could not satisfy the requirements of control accuracy.Hence, instead of conventional PID controller, LQG controller is adopted in this project,whose design is based on the precise model of system. Aiming for modeling difficulties of antennas due to their structural complexity, this thesis adopt a scheme to model large antennas based on the subspace identification method. A problem is deep discussed that how to reduce the noise effect in system identification modeling. On this basis, the antenna servo control experimental platform is built to further identify the precision of subspace modeling method from the perspective of engineering. The main work and achievement are as follows:Firstly, by comparing the difference between subspace method and conventional identification methods, considering large antenna features, subspace identification method is adopted to model large antenna in this thesis.Then the thesis describes the main factors that have effects on identification in identification experiments in detail, and gives the related principles of designing identification experiments. A software related to subspace identification method is developed under MATLAB environment. For the example of antenna model, the design of system identification experiment is completed, as well as data acquisition, data preprocessing, and modeling identification in the process of subspace identification modeling. Test is carried on with some effective methods, and simulation result shows availability of subspace identification method on large antenna modeling.Next, the effects of noise on system identification are analyzed in the thesis and also the identification performances of subspace identification method are further studied with different SNRs, i.e. signal to noise ratios. The simulation datum are preprocessed through the correlation method, which reduces the influence of noise on identification process effectively and improves the system identification effect greatly.At last,the design process of an experiment platform for an antenna servo system arepresented in this thesis, and the development of data acquisition algorithm and identification software are completed. The position loop model is obtained through experiments, and the verification work of identification model is completed through two aspects including the perspective of predictive outputting ability of model and practical control performance of controller, which demonstrates availability of scheme in this thesis.The identification software developed in the experiment can be applied to the large antenna modeling in practise only through a simple migration.