Data-Driven H∞ Controller Synthesis of A Seeker-Integrated Inertially Stabilized Platform

Abstract

In this thesis, a new method is proposed that allows the synthesis of controllers without the need for any parametric model of an inertially stabilized platform. It is a known fact that the stability and performance of model-based controllers are highly related to the accuracy of the developed model. In this context, it was planned to design a robust controller with the "Data-Driven Control" method and to test the relevant controller on the stabilized gimbal platform inside the infrared seeker. The general behavior of the open-loop system required for the data-driven design was observed by looking at the frequency-based input-output relationship. Then, the controller was designed according to the determined performance weights. During the determination/design of the weighting functions and also the controller; a convex optimization method was used in accordance with the Nyquist stability criterion in order to provide robust performance and stability. In addition to the non-parametric design, after creating a state space model, a model-based LQG design including a discrete LQR controller and a Kalman Filter were also implemented. The unknown parameters of the model were obtained through system identification tests and implemented into the model. Finally, in order to evaluate the iv applicability of the study and to use it in a real-time system, the results of the robust controller created with data-driven synthesis were compared with the model-based LQG controller; frequency and time-based performance analysis were performed. As a result of the comparisons, it was observed that the data-driven controller is significantly superior to the other controllers in the speed loop bandwidth performance, which is critical for stabilized platforms. In addition, the reaction time to the angular velocity commands transmitted by the guidance was measured to be shorter than the model-based controller. In this context, considering the total time and complexity spent on the design, quite successful results were obtained with the data-driven robust controller in terms of performance and feasibility.