Academic literature on the topic 'Nonlinear Controller'

Bibliography: leaves 74-[76.]<br>A novel control method, based on interval analysis, that optimises the control surface (or u-surface) for sampled systems with output disturbances is demonstrated on a driven pendulum with actuator constraints. The fitness function to be maximized is the probability of each state of the system being controlled to the setpoint without being perturbed to regions that are more iterations away from the setpoint. The u-surface is designed by finding all the states that could go to the setpoint in an interval and optimising these states. This process is repeated (backwards in time) by optimising states that go to the previously optimised states until no more states that have not been optimised are found. The proposed control method has been applied to the problem of swinging up a driven pendulum from rest to the inverted position with constraints on the torque of the motor. This method is computationally intensive and time constraints limit its current application to systems of low order.

Control design for nonlinear systems remains an open problem in control theory despite the recent increase in research attention. This PhD work is motivated by this fact, addressing the constructive observer design approach, the output regulation problem, minimum entropy control, fault tolerant control (FTC), and iterative FTC for nonlinear systems.

In this thesis, digital controller design methods for sampled-data nonlinear systems are considered. Although sampled-data nonlinear control has attracted much attention in recent years, the controller design methods for sampled-data nonlinear systems are still limited. Therefore, a range of controller design methods for sampled-data nonlinear systems are developed such as backstepping, adaptive and robust backstepping, reduced-order observer-based output feedback controller design methods based on the Euler approximate model. These controllers are designed to compensate the effects of the discrepancy between the Euler approximate model and exact discrete time model, parameter estimation error in adaptive control and observer error in output feedback control which behave as disturbance. A dual-rate control scheme is presented for output-feedback stabilization of sampled-data nonlinear systems. It is shown that the designed controllers semiglobally practically asymptotically (SPA) stabilize the closed-loop sampled-data nonlinear system. Moreover, various applications of these methods are given and their performances are analyzed with simulations.

De nos jours, la théorie de contrôle joue un rôle significatif dans presque tous les domaine de la science et de l'ingénierie. Les contrôleurs linéaires PID sont les applications principales de la théorie de contrôle, et ils se basent sur les systèmes de contrôle simples. Mais beaucoup de vrais systèmes possèdent des caractéristiques non-linéaires. Dans la pratique, il est nécessaire de faire beaucoup de linéarisations. Quand nous employons le contrôleur classique dans un système non-linéaire fortement complexe, les difficultés augmentent exponentiellement. Pour éviter les imperfections, on peut employer des contrôleurs flous. Le contrôleurs flous se basent sur le système de connaisance. Ce sont des outils importants dans le domaine de l'automatique. Ils possèdent beaucoup plus d'avantages que les contrôleurs classiques"PID", mais ils ont besoin d'experts pour concevoir les règles de base. La limite principale des contrôleurs flous est la difficulté d'établir les règles de base. Maintenant, beaucoup de recherches sont consacrées à la fusion des réseaux de neurones et de systèmes flous dans une nouvelle structure (les réseaux de neuro-floue). Cette approche combine les avantages de deux paradigmes puissants dans une capsule simple, et fournit un cadre puissant pour extraire des règles floues des données numériques. Cependant, cette technologie n'est pas parfaite. Il reste quelques difficultés: beaucoup de règles floues sont nécessaires, les algorithmes sont complexes et la fiabilité est basse (Par exemple, pour un même modèle ou fonction, les résultats dépendent des ensembles d'apprentissage). Pour éviter les difficultés, ce mémoire présente une nouvelle méthode, appelée"inférence neuro-floue de haute-dimension". L'idée fondamentale de cette méthode proposé est de considérer chaque donné dans ce système comme point avec la haute dimension. Chaque dimension d'entrée sera traitée en même temps dans les mêmes sous-ensembles de haute dimension. L'algorithme proposé a été examiné sur différentes applications, et les résultats ont été comparés aux données éditées sur trois problèmes de repère. Cet algorithme est simple à employer, et les résultats expérimentaux prouvent que le nombre de faisceaux exigés est inférieur à ceux rapportés dans la littérature. L'exactitude de rendement est bonne dans beaucoup d'applications.

To design a robust controller for active suspension systems is very important for guaranteeing the riding comfort for passengers and road handling quality for a vehicle. In this thesis, the mathematical model of full vehicle nonlinear active suspension systems with hydraulic actuators is derived to take into account all the motions of the vehicle and the nonlinearity behaviours of the active suspension system and hydraulic actuators. Four robust control types are designed and the comparisons among the robustness of those controllers against different disturbance types are investigated to select the best controller among them. The MATLAB SIMULINK toolboxes are used to simulate the proposed controllers with the controlled model and to display the responses of the controlled model under different types of disturbance. The results show that the neurofuzzy controller is more effective and robust than the other controller types. The implementation of the neurofuzzy controller using FPGA boards has been investigated in this work. The Xilinx ISE program is employed to synthesis the VHDL codes that describe the operation of the neurofuzzy controller and to generate the configuration file used to program the FPGA. The ModelSim program is used to simulate the operation of the VHDL codes and to obtain the expected output data of the FPGA boards. To confirm that FPGA the board used as the neurofuzzy controller system operated as expected, a MATLAB script file is used to compare the set of data obtained from the ModelSim program and the set of data obtained from the MATLAB SIMULINK model. The results show that the FPGA board is effective to be used as a neurofuzzy controller for full vehicle nonlinear active suspension systems. The active suspension system has a great performance for vibration isolation. However the main drawback of the active suspension is that it is high energy consumptive. Therefore, to use this suspension system in the proposed model, this drawback should be solved. Electromagnetic actuators are used to convert the vibration energy that arises from the rough road to useful electrical energy to reduce the energy consumption by the active suspension systems. The results show that the electromagnetic devices act as a power generator, i.e. the vibration energy excited by the rough road surface has been converted to a useful electrical energy supply for the actuators. Furthermore, when the nonlinear damper models are replaced by the electromagnetic actuators, riding comfort and the road handling quality are improved. As a result, two targets have been achieved by using hydraulic actuators with electromagnetic suspension systems: increasing fuel economy and improving the vehicle performance.