| A new systematic design method is developed to implement a classical linear continuous-data controller by using the sliding mode technique. For this continuous-data controller, the sliding variable is formulated by utilizing both the input and the output of the plant. This sliding mode controller is designed without having to use the state variables of the plant. One main advantage of this approach is that controllers with improper transfer function can be implemented without noise enhancement. The choice of switching gain is studied so that the sliding condition can be satisfied.; When the system is contaminated with noise, the sliding on the average concept is presented in order to avoid choosing a very large switching gain. The effect of noise on choosing the switching gain is also studied.; For the case when the sliding conditions are not fulfilled, this dissertation presents other conditions under which the system can still be BIBO stable. This is analyzed when there is no noise, and also when noise is existing. The conditions are related to the Popov plot of a linear system.; Finally this new technique is successfully used to design a sliding mode controller for an inertially stabilized platform. This platform is subjected to torque disturbance, gyro measurement noise, parameter and structural uncertainties. The objective is to reduce the deviations of the angular position from its steady state value. |