چكيده به لاتين
Abstract
In this thesis, the categories of data-driven controllers, entitled Data-Driven Adaptive Sliding Mode Control (DDASMC), are considered for a class of nonlinear and nonaffine Multi-Input Multi-Output (MIMO) systems. The data-driven controllers operate based on input-output measured data. Hence, they do not depend on the mathematical model of the system. This type of controller is classified as a subset of indirect data-driven and model-free controllers that are designed based on the dynamic linearization method. In this thesis, first, the DDASMC with adaptive switching gain and PJM observer was designed. To evaluate the performance of the controller, the proposed method is applied to three-tank and 2-DOF robot manipulator systems. According to the results, the main advantages of the proposed controller are the ease of adjusting the coefficients and suitable performance for the systems with slow dynamics. Moreover, the proposed controller is able to reject disturbance and reduce chattering phenomenon. However, the proposed controller does not apply to systems with fast dynamics such as 2-DOF robotic manipulators. Therefore, the proposed controller was combined with optimal and constrained methods. By considering the saturation phenomenon in the input and obtaining the optimal control input, the proposed controller can increase the speed of convergence and be applied to systems with fast dynamics. The controller proposed in this part was applied to a 2-DOF robotic manipulator. Results show that the proposed controller has a good disturbance rejection ability and high tracking accuracy. On the other hand, a large amount of calculations and the difficult setting of the coefficients are among its main problems that force the user to utilize the powerful hardware. Hence, at the end of this thesis, the DDASMC with Neural Network Disturbance Observer (NNDOB) is proposed. The main advantages of the proposed controller are the use of an NNDOB to compensate for missing data and increase the speed of convergence by using a terminal sliding surface. Furthermore, by using the designed DOB, no prior knowledge about the upper bound of disturbances is needed. The presented controller significantly increases the sampling time and as a result, requires simpler hardware for implementation. However, it is difficult to adjust the controller parameters. This controller is implemented on a control board and is applied to the 2-DOF laboratory manipulator. The experimental and simulation results of this thesis demonstrate that the proposed controller compared to the other data-driven methods has appropriate performance in the presence of external disturbances and measurement noise. To recapitulate, with the help of the presented controller in this thesis, the chattering phenomenon was significantly reduced, stability proofs were done more easily and the conservatisms were decreased. In addition, disturbance rejection was performed well, tracking ability was enhanced, and the sampling time was significantly increased.
Keywords: Nonlinear MIMO systems, Data-Driven Sliding Mode controller, Optimal Sliding Mode, Terminal Sliding Mode, Adaptive Sliding Mode Observer, Disturbance Observer.
