No description available.

<p>To control the exhaust gas recirculation (EGR) and the exhaust brake, the position of a butterfly valve connected to a piston inside a pneumatic cylinder is controlled by altering the pressure inside the cylinder. This thesis evaluates the possibility to do this with pulse width modulation (PWM) controlled On/Off valves. The whole electro-pneumatic actuator system is built out of two On/Off valves and a cylinder.</p><p>A mathematical model of the system is constructed. The complete system model on state space form consists of nine states and is nonlinear. The model captures the dynamics of the system. The statics of the system is not captured as accurately. The model is still good enough to be used as aid when developing control strategies, since position feedback is available.</p><p>Automatic control strategies for the system are first developed and tested in simulation. The first approach is PID control. Because of the nonlinear properties of the system the results from a PID with a constant proportional part is unsatisfactory. To cope with the nonlinearities, a fuzzy controller is constructed; the results prove somewhat better, but not as good as expected due to implementation difficulties.</p><p>In a test bench the system is controlled by a P controller with feedforward from position. The feedforward strongly reduces the nonlinear behavior of the system. With this implementation the results that were hoped for with the fuzzy controller are reached.</p>

To control the exhaust gas recirculation (EGR) and the exhaust brake, the position of a butterfly valve connected to a piston inside a pneumatic cylinder is controlled by altering the pressure inside the cylinder. This thesis evaluates the possibility to do this with pulse width modulation (PWM) controlled On/Off valves. The whole electro-pneumatic actuator system is built out of two On/Off valves and a cylinder. A mathematical model of the system is constructed. The complete system model on state space form consists of nine states and is nonlinear. The model captures the dynamics of the system. The statics of the system is not captured as accurately. The model is still good enough to be used as aid when developing control strategies, since position feedback is available. Automatic control strategies for the system are first developed and tested in simulation. The first approach is PID control. Because of the nonlinear properties of the system the results from a PID with a constant proportional part is unsatisfactory. To cope with the nonlinearities, a fuzzy controller is constructed; the results prove somewhat better, but not as good as expected due to implementation difficulties. In a test bench the system is controlled by a P controller with feedforward from position. The feedforward strongly reduces the nonlinear behavior of the system. With this implementation the results that were hoped for with the fuzzy controller are reached.