A Trained Network Solution for Multi-State Structural Awareness
Abstract
This project considered a rapid condition assessment test using transmissibility measurements in vibration tests. A hypothesis test is established that detects changes in transmissibility estimations over time that indicate target failure mode occurrence; in this project, fastener preload loss was the first considered failure mode for simple representations of satellite structures, and later consideration was given to cracking in beam-like rotor structures. Inherent to this procedure (or any structural health monitoring application) is the quantification of uncertainty in order to obtain maximal probability of detection while minimizing the probability of false alarms. Within the specific applications just cited, a set of novel statistical models were developed which resulted in actual probability density function derivations for transmissibility measurements (considering both estimation error and extraneous environmental noise-induced error). These probability density functions could then be used to establish quantitative thresholds that helped establish rigorous probability-of-detectionbased hypothesis testing that suggest optimal testing procedures for detecting damage when using transmissibility as a feature.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 12, 2012
- Accession Number
- ADA557291
Entities
People
- Michael D Todd
- Zhu Mao
Organizations
- University of California Regents