This research investigates a pneumatic suspension system containing an air spring, air flow valve, and an accumulator, where the spring and damping functions are combined into one package. The spring and accumulator provide the spring characteristics, and the computer controlled adjustable valve provides the damping characteristics by automatically adjusting the air flow between the air spring and the accumulator.

An extensive analysis and investigation of the plant dynamics is performed. A dynamic plant model is developed and tuned to experimental data. The plant model is then used in the design of a semi-active control system. A detailed description of the model tuning procedure is provided.

Based upon the insights gained through analysis and system identification, a semi-active control methodology is developed, which exploits certain unique features of the system. Three potential controllers are developed and compared, where each controller uses different measurement feedback signals. However, all three controllers measure direct force generation through a pressure feedback signal. Both experimental and simulation data for the controllers is provided.

The first controller uses an LQI (Linear Quadratic Impulse) optimal solution, based on Covariance Control Theory, to generate an optimal "active" damping control force, along with a Set-Point plus PI tracking controller to adjust the valve opening to cause the system to track this desired force during a switching event or control window of opportunity.

The second controller uses a Modified Skyhook solution to generate the ideal tracking signal, along with a Set-Point plus PI tracking controller. The LQI controller is used in simulation (offline) to aid in setting the skyhook gain on the Modified Skyhook controller.

The third controller uses a Relative Displacement solution to generate the ideal tracking signal, along with a Set-Point plus PI tracking controller. The LQI controller is used (offline) to aid in setting the gain on the Relative Displacement controller. This controller is probably the most useful for vehicular applications, since only relative coordinates and a pressure are required for feedback.

It was found that all three controllers could track an optimally generated "active" signal during the switching event, provided the proper gains were chosen.