چكيده به لاتين
In this thesis, an Adaptive Fuzzy Extended State Observer (AFESO) for single-input single-output nonlinear affine systems in the presence of output constraint is proposed to estimate the states of the system and external disturbances. Extended State Observers (ESOs) have been used for their efficiency in the presence of disturbance and uncertainty, but it should be noted that it has some limitations, such as matching conditions, dependency on the bandwidth of the observer, the inability to remove the effect of the time-varying disturbance variable, peaking phenomenon, and high sensitivity to measurement noise. To overcome this limitation, in this thesis, disturbance is considered time-varying and applied to the states of the system. The purpose of the proposed fuzzy-adaptive observer is to overcome the limitations of the ESO and enhance the system’s performance in the presence of external disturbances and output constraints. The ESO is a class of high gain observers that requires a large bandwidth for optimal performance. Since the ESO is considered linear with constant gains, the bandwidth of the observer is limited, which drives the system away from the optimal performance. One of the solutions to overcome the bandwidth limitation is to design an Adaptive Extended State Observer (AESO). In addition, the observer gains are considered time-varying and adjusted with adaptation laws to eliminate the effect of external disturbances and perturbation parameters. First, the matched disturbance and linear observer function are considered and a fuzzy system is employed to estimate the nonlinearities of the system. Next, matched and mismatched disturbances are applied to systems. Moreover, the class of the system is extended to the pure feedback and the observer function is designed in a nonlinear form. Although designing nonlinear observers cause more computational time and complexity in the design procedure, compared to a linear observer, it has much better performance in a transient and steady state in the presence of disturbance and uncertainty. Moreover, to guarantee the convergence of the tracking error as well as satisfying the output constraints of the system, a command-filtered backstepping method based on the barrier Lyapunov function method is designed in stability analysis. To improve the performance of the system, the observer parameters and filter bandwidth are adjusted using the Mamdani fuzzy system. Simulating example of an inverted pendulum and a single link flexible-joint manipulator are considered. Furthermore, the performance of the proposed method has been compared with a recently published scheme in the related literature in terms of convergence rate, overshoot, and fluctuations in the system response. Stability analysis and convergence of estimation errors have been studied in the design procedure.
