Vibration based fault detection and identification in an aircraft skeleton structure via a stochastic functional model based method

• Published in: Mechanical systems and signal processing Vol. 22; no. 3; pp. 557 - 573
• Main Authors: Sakellariou, J.S., Fassois, S.D.
• Format: Journal Article
• Language: English
• Published: London Elsevier Ltd 01.04.2008; Elsevier
• Subjects: Aircraft skeleton structure; Exact sciences and technology; Fault detection; Fault estimation; Fault identification (localization); Functionally pooled models; Fundamental areas of phenomenology (including applications); Measurement and testing methods; Physics; Solid mechanics; Statistical methods; Structural and continuum mechanics; Vibration based methods; Vibration, mechanical wave, dynamic stability (aeroelasticity, vibration control...); Fault estimation; Fault detection; Fault identification (localization); Statistical methods; Vibration based methods; Aircraft skeleton structure; Functionally pooled models; Vibration; Probabilistic approach; Statistical estimation; Stochastic method; Telemetry; Modeling; Remote sensing; Statistical method; Statistical test; Added mass; Signal processing; Skeleton; Localization; Aircraft; Defect detection; Non contact measurement
• ISSN: 0888-3270; 1096-1216
• DOI: 10.1016/j.ymssp.2007.09.002

The problem of vibration based fault detection, identification (localization) and estimation in a scale aircraft skeleton structure is considered via a stochastic functional model based method (FMBM). The method is based on the novel class of stochastic Functionally Pooled models, which are capable of accurately representing the structure in a faulty state for the state's continuum of fault magnitudes, as well as interval estimation and formal statistical hypothesis testing procedures. The faults considered consist of small masses attached to the structure. The method is capable of operating even on single-excitation single-response signals, and is shown to achieve effective fault detection and identification, as well as remarkable accuracy in estimating the exact fault magnitude. This is so even for “unmodelled” faults, or faults monitored by remote sensors.