Certification in Structural Health Monitoring Systems

Despite advances in health monitoring systems, the certification process for automated SHM systems remain rooted in the probability of detection (POD). POD is a fundamental part of the certification process, however, it is not the only measure of system validity. The process to certify a system must contain measures of accuracy and reliability. Disclosing the accuracy and reliability at the time of system design allows the end user to weigh differences among various SHM system architectures. In this manner, end users may identify the architecture that meets their specific overall system requirements. Further, it creates a benchmark from which system redesigns may be assessed and compared in order to improve overall system performance. In order to determine the appropriate measures of accuracy and reliability for the SHM system, first, the necessary output of such a system must be defined. Typically, the system is built in order to identify and assess structural damage. The appearance of damage is a detection problem whereas the extent of damage (crack length and location) is an estimation problem. Thus, the proper certification of a SHM system must incorporate measures of accuracy and reliability for both the detection and estimation of structural damage. These include probabilities related to detection such as the probabilities of true and false alarms as well as positive and negative predictive values. Confidence intervals must be defined as measures of the reliability of these probability estimates. Measures for the accuracy of the location and extent of structural damage include summary measures of the estimation process or model as well as the predicted location or crack length and its associated confidence interval. Establishment of the certification process as related to SHM will lead to better quantification of system performance for both detection and estimation and ultimately better system designs. These criteria may then act as a springboard for certification in subsequent prediction of structural failure, time remaining until structural failure, and other general aspects of risk analysis related to structural health. _____________ Christine M Schubert Kabban, Ph.D., MBA, Assistant Professor, Department of Mathematics and Statistics, AFIT/ENC, Air Force Institute of Technology, 2950 Hobson Way, WPAFB, OH 45433-7765 (937) 255-3636 x4549, [email protected] Mark M. Derriso, Technical Advisor, Integrated Systems Branch, Air Force Research Laboratory, 2130 8 Street, WPAFB, OH 45433 Certification in Structural Health Monitoring Systems C. M. S. KABBAN1 and M. M. DERRISO2 Report Documentation Page Form Approved