Development of an autonomous continuous monitoring system for mechanical damage detection.
- Hoon
- David W.
- Charles R.
The primary objective of damage identification is to ascertain the existence of damage within a mechanical system. This study applies the Sequential Probability Ratio Test (SPRT) to examine if damage is present or not. In the original formulation of the SPRT, the distribution of data is assumed Gaussian and thresholds for monitoring are set focusing on the center mass properties of the distribution. Decision-making for damage identification is, however, often sensitive to the tails of the distribution and the tails may not necessarily be governed by Gaussian characteristics. By modeling the tails using the technique of Extreme Value Statistics (EVS), the thresholds for the SPRT may be set more accurately avoiding the unnecessary normality assumption. The proposed combination of the SPRT and the EVS is demonstrated using experimental data collected from a three-story frame structure with bolted connections. The primary goal of structural health monitoring is simply to identify from measured data if a structure has deviated from a normal operational condition. Particularly, vibration-based damage detection techniques assume that changes of the structure's integrity affect characteristics of the measured vibration signals enabling one to detect damage. Many current approaches to this problem involve methods that leave much to the interpretation of analysts. These methods may enable a trained eye to discern and locate damage but are not easily automated or objective. In an attempt to automate the damage identification procedure, the SPRT is employed for the decision-making procedure. The original SPRT assumes that the extracted features have a Gaussian distribution. This normality assumption, however, may place misleading constraints on the tails of the distribution. As the problem of damage detection specifically focuses attention on the tails, the assumption of normality is likely to lead the analysis astray. To overcome this difficulty, the performance of the SPRT is improved by integrating the EVS, which specifically models behavior in the tails of the distribution of interest, into the SPRT.
- Research Organization:
- Los Alamos National Laboratory
- Sponsoring Organization:
- DOE
- OSTI ID:
- 976160
- Report Number(s):
- LA-UR-02-2429
- Country of Publication:
- United States
- Language:
- English
Similar Records
Utilizing the sequential probability ratio test for building joint monitoring
Damage detection in mechanical structures using extreme value statistic.