Abstract:

This thesis is directed toward the design and implementation of a distributed actuation and monitoring system for research in diagnosis and control of embedded systems. The core of the actuation and monitoring system are smart transducers, a Network Capable Application Processor (NCAP) and a Smart Transducer Interface Module (STIM) pair, whose interaction is defined by the IEEE 1451.2 standard. The physical system under consideration is a typical three-tank configuration with a pump and a set of valves that control the flow of liquid into and out of the tanks. The first task in this thesis work was to redesign the physical system so that the tank system could be operated in a variety of different configurations. The second task was to design and implement the STIM to provide the actuation and monitoring functions.

To design the system to run model-based diagnosis and fault-adaptive control experiments, an accurate model of the system was derived that could predict system behavior under a variety of operating modes. Suitable experiments were designed to estimate the model parameters in a step by step fashion, and the data collected by running the experiments was used to derive the model parameters using least square estimation methods available in the Matlab environment. Further experiments were then conducted to demonstrate that the model accurately predicts the physical system dynamics.

Download: Lyons2004_thesis.pdf (749 KB)