Adjustment Techniques Of High-precision Antenna Sub-reflector

The design of high frequency and large diameter become an inevitable tendency of the radio telescope's development. For large-diameter antenna, different antenna elevation will have different effects on the deformation of antenna panel due to the gravity. In order to overcome the impacts on antenna electrical properties caused by the panel deformation, it needs to adjust the antenna sub-reflector for counteracting this effect. Thus, the research on the motion precision of antenna's sub-reflector adjustment mechanism is a key technology to improve the antenna panel pointing accuracy.This thesis research object is a high-accuracy deputy surface adjustment mechanisms. As to improve the movement-posed accuracy of this six degrees of freedom(DOF in brief) parallel mechanisms, the motion-attitude errors will be analyzed and summarized which is cooperated with the specific experiment situation based on at least the theories of parallel mechanism calibration and forward and reverse kinematics. Two categories of affecting the mechanism-attitude errors are concluded: the relative certain errors caused by the hinge calibration errors and the uncertain non-linear errors caused by the hinge gap, panel deformation, measurement errors and etc.First of all, the thesis proposes the inverse solution of hinge calibration algorithm based on the six DOF parallel mechanisms according to the hinge calibration errors. Meanwhile it calculates the errors between the hinge central coordinates of the original calibration and the actual one. What's more, it simulates the required experimental data to validate whether the algorithm is effective. The basic principle is to use measured kinematic parameters of the six DOF parallel mechanisms to construct the hinge's error function, and solves the mechanisms' real hinge coordinates whose object is the minimization of the error function.Secondly, the thesis analyzes the non-linear relationship between the extension and the compensation of outrigger when mechanisms move, aiming to the non-linear uncertain errors caused by hinge gap, panel deformation, measurement error and etc. Since this non-linear relation is not one-to-one correspondence but a multi-value one, it brings out the direct compensation method based on the mixture density neural network and simulates the experimental data needed which verifies the effectiveness of the method. The fundamental principles are to train the non-linear and multi-value corresponding relation between outrigger's extension and compensation which exploits the characteristics of the neural network excellent non-linear fitting capabilities and mixture density model probability selection according to multi-value mapping.Finally, for the adjustment experiments of high precision adjustment mechanisms, it develops and improves the control interface based on DELPHI language. And different method procedures are designed to adapt to the required data characteristic of the two upper algorithms, Also the real experimental data are recorded and collected to validate each compensation algorithm's effectiveness.