Publications / 1985 Proceedings of the 2nd ISARC, Pittsburgh, USA

Automation of Condition and Deterioration Surveys Using Knowledge-Based Signal Processing

K. Maser, D. Smit
Pages 189-206 (1985 Proceedings of the 2nd ISARC, Pittsburgh, USA, ISSN 2413-5844)
Abstract:

"Condition and deterioration surveys using state-of-the-art sensory techniques generate far more data than can be interpreted by conventional methods. As a result, powerful sensory techniques such as ground penetrating radar and infrared thermography have yet to realize their full potential in civil engineering and construction applications. This paper proposes a method to automate the interpretation of large quantities of sensory data. This method combines conventional digital signal processing with encoded judgement and experience taken from the sensory, materials and structural domains. Automated radar data analysis to detect deterioration in reinforced concrete bridge decks has been selected as an illustrative example. In this application, a digital processor would produce a signature for each radar position based on the amplitudes and arrival times of radar waveform peaks. A knowledge-based processor would interpret these signatures using encoded knowledge of radar, concrete deterioration, and bridge engineering. The interpretation may conclude that deterioration is unlikely, that the environment is conducive for deterioration, and/or that deterioration has actually begun. Each conclusion would have an associated certainty factor. An analysis of several signatures has been carried out using mini-MYCIN for the knowledge-based processing. Mini-MYCIN is an expert system “shell” based on the MYCIN system developed for medical diagnosis. The signatures for each waveform were artificially generated to illustrate several conditions associated with deterioration. We have concluded from this work that the Mini-MYCIN shell is not well-studied to spatial reasoning because of its context three; nor is it well-studied to extensive data entry because of its interactive nature. We have also concluded that future work should seek to keep knowledge from different domains (e.g., radar, concrete, bridges) separate in the knowledge base. A better rationale for assigning certainty factors should also be developed. "

Keywords: No keywords