Open Access Open Access  Restricted Access Subscription or Fee Access

Generating Viable Data to Accurately Quantify the Performance of SHM Systems



Reliable structural health monitoring (SHM) systems can automatically process data, assess structural condition and signal the need for human intervention. There is a significant need for formal SHM technology validation and quantitative performance assessment processes to uniformly and comprehensively support the evolution and adoption of SHM systems. In recent years, the SHM community has made significant advances in its efforts to evolve statistical methods for analyzing data from in-situ sensors. Several statistical approaches have been demonstrated using real data from multiple SHM technologies to produce Probability of Detection (POD) performance measures. Furthermore, limited comparisons of these methods - utilizing different simplification assumptions and data types - have shown them to produce similar POD values. Given these encouraging results, it is important to understand the circumstances under which the data was acquired. Thus far, the statistical analyses have assumed the viability of the data outright and focused on the performance quantification process once acceptable data has been compiled. This paper will address the array of parameters that must be considered when conducting tests to acquire representative SHM data. For some SHM applications, it may not be possible to simulate all environments in one single test. All relevant parameters must be identified and considered by properly merging results from multiple tests. Laboratory tests, for example, may have separate fatigue and environmental response components. Flight tests, which will likely not include statistically-relevant damage detection opportunities, will still play an important role in assessing overall SHM system performance under an aircraft operator’s control. One statistical method, the One-Sided Tolerance Interval (OSTI) approach, will be discussed along with the test methods used to acquire the data. Finally, prospects for streamlining the deployment of SHM solutions will be considered by comparing SHM data needs during what is now an introductory phase of SHM usage with future data needs after a substantial database of SHM data and usage history has been compiled.


Full Text: