| This dissertation presents development of an information architecture which formalizes, represents and deploys the necessary knowledge for structural health monitoring operation, in particular, for damage identification, location and assessment with diagnosis, in instrumented structural systems.; The features of the developed information architecture are described as follows. This information architecture: (1) contains both physical and behavioral information about the structural system and its principal components, (2) contains sensor, signal and intelligent signal processing information and capabilities, (3) supports on-line autonomous signal processing to handle the avalanche of signals produced by a multi-channel instrumented civil structure, and (4) supports health monitoring operation at various abstraction levels.; A hybrid scheme, amalgamating object-oriented knowledge representation and logic-based implementation, is developed in order to support various aspects of the structural health monitoring operation.; Neural networks are investigated and incorporated into this hybrid scheme as a non-modal signal monitoring tool which is suited for damage-sensitive and noise-tolerant autonomous on-line processing of the generated signals. Among a number of neural network algorithms, backpropagation networks are selected, trained and tested for damage detection, location and severity determination using both actual sensor signals and simulated signals.; A case study which involves health monitoring was conducted using the vibration signals from a series of experiments performed on an instrumented scale model bridge in order to evaluate performance, adequacy and extensibility of the developed system.; It is concluded that the proposed information architecture and resulting computational environment constitute a promising potential foundation of an eventual smart structural system based on its reasonably good performance.; Further study is suggested in the areas of instrumentation, signal processing, diagnosis, knowledge engineering, and integrated software engineering. |
