Date of Award
Open Access Thesis
The non-linear elastodynamics of a flat plate subjected to low velocity foreign body impacts is studied, resembling the space debris impacts on the space structure. The work is based on a central hypothesis that in addition to identifying the impact locations, the material properties of the foreign objects can also be classified using acousto-ultrasonic signals (AUS). Simultaneous localization of impact point and classification of impact object is quite challenging using existing state-of-the-art structural health monitoring (SHM) approaches. Available techniques seek to report the exact location of impact on the structure, however, the reported information is likely to have errors from nonlinearity and variability in the AUS signals due to materials, geometry, boundary conditions, wave dispersion, environmental conditions, sensor and hardware calibration etc. It is found that the frequency and speed of the guided wave generated in the plate can be quantized based on the impactor's relationship with the plate (i.e. the wave speed and the impactor's mechanical properties are coupled). In this work, in order to characterize the impact location and mechanical properties of imapctors, nonlinear transient phenomenon is empirically studied to decouple the understanding using the dominant frequency band (DFB) and Lag Index (LI) of the acousto-ultrasonic signals. Next the understanding was correlated with the elastic modulus of the impactor to predict transmitted force histories.
The proposed method presented in this thesis is especially applicable for SHM where sensors cannot be widely or randomly distributed. Thus a strategic organization and localization of the sensors is achieved by implementing the geometric configuration of Theodorous Spiral Sensor Cluster (TSSC). The performance of TSSC in characterizing the impactor types are compared with other conventional sensor clusters (e.g. square, circular, random etc.) and it is shown that the TSSC is advantageous over conventional localized sensor clusters. It was found that the TSSC provides unbiased sensor voting that boosts sensitivity towards classification of impact events. To prove the concept, a coupled field (multiphysics) finite element model (CFFEM) is developed and a series of experiments were performed. The dominant frequency band (DBF) along with a Lag Index (LI) feature extraction technique was found to be suitable for classifying the impactors. Results show that TSSC with DBF features increase the sensitivity of impactor's elastic modulus, if the covariance of the AUS from the TSSC and other conventional sensor clusters are compared. It is observe that for the impact velocity, geometric and mechanical properties studied herein, longitudinal and flexural waves are excited, and there are quantifiable differences in the Lamb wave signatures excited for different impactor materials. It is found that such differences are distinguishable only by the proposed TSSC, but not by other state-of-the-art sensor configurations used in SHM. This study will be useful for modeling an inverse problem needed for classifying impactor materials and the subsequent reconstruction of force histories via neural network or artificial intelligence.
Finally an alternative novel approach is proposed to describe the Probability Map of Impact (PMOI) over the entire structure. PMOI could serve as a read-out tool for simultaneously identifying the impact location and the type of the impactor that has impacted the structure. PMOI is intended to provide high risk areas of the space structures where the incipient damage could exist (e.g. area with PMOI > 95%) after an impact.
Agbasi, C.(2014). Classification of Low Velocity Impactors Using Spiral Sensing of Acousto-Ultrasonic Waves. (Master's thesis). Retrieved from http://scholarcommons.sc.edu/etd/2660