Automated Interpretation for Sensing in Situ Conditions

Abstract

A concept for automating the interpretation of sensor signals used to characterize in situ conditions is proposed, developed, implemented, and applied. The concept, called AISD, automates the application of encoded knowledge to the interpretation of signals generated by sensor devices used to detect in situ properties of materials and structures. The knowledge is structured according to four interpretive disciplines. The incoming signals are algorithmically processed into signatures that, along with contextual data provided by the user, are recognized by and operated upon by the knowledge base. A review of potential applications of the AISD concept supports those involving condition assessment of extensive facilities. The interpretation of ground penetrating radar for bridge deck evaluation is selected as a model for implementation of a prototype system. A prototype knowledge base and supportive software have been developed and applied to synthetic data. Results of computational experiments demonstrate the mutually supportive combination of sensor and context data in producing the desired condition assessment. Future developments for this specific application are proposed. More general issues involving the use of quantitative sensor data in an heuristic reasoning environment are raised for future consideration.