This Partnerships for Innovation (PFI) project is a Type III (A:C) partnership between Villanova University, an NSF PFI grantee (0332490), and Bucknell University, an institution new to the PFI Program (defined as one that has never been a PFI grantee). The project applies core and emerging research in acoustic signal processing, aiming at increasing acoustic system performance and enhancing ultrasound imaging. This is achieved through advances in signature analyses and feature extractions. Novel fusion modalities based on multiple and distributed sensors are examined for further system performance improvements.
This partnership recognizes the growing acoustic and ultrasound local and national job markets, spanning the healthcare, automotive, and aerospace sectors with applications ranging from medical diagnostic and therapeutic needs to predictive maintenance of machinery and products, and extending to acoustic source training and localizations. The proposed activities will serve to improve acoustic and ultrasound data analyses and imaging. The enhancements in medical ultrasound imaging have clear societal impact, as they will increase diagnostic capabilities and quality of life while reducing medical costs. In the predictive maintenance field, if more information can be acquired describing the state and properties of existing machinery, better decisions can be made with regard to failure prediction and safety issues.
Partners at the inception of the project are Academic Institutions: Villanova University (lead institution), Bucknell University, and Gwynedd-Mercy College; Private Sector Organizations: Siemens and The Boeing Company; State and Local Government Organizations: Ben Franklin Technology Partners; and Federal Government Laboratory: Naval Sea Systems Command (NAVSEA).
The NSF PFI was led by Villanova University and included the two academic partners, Bucknell University and Gwynedd Mercy College. The industrial partners were Siemens Corp., Technical Vision Inc., and Vibrational Specialty Corp. It also included the economic development partner, the Ben Franklin Technology Partners (BFTP) of Southeastern Pennsylvania. The extensive expertise housed at the Center for Advanced Communications, Villanova University in the area of urban radar and through-the-wall radar imaging has been utilized to solve outstanding problems in acoustics and ultrasound imaging such as flaw detection, localization and classification under complex scattering, reverberations, and multipath scenarios. In particular, multipath modeling and multiple-input multiple-output (MIMO) imaging principles, prominently used for radar imaging of targets in indoor environments, have been successfully applied and employed for ultrasound testing of materials with known geometry. The NSF PFI project has improved the efficiency of structural health monitoring (SHM) techniques via advanced modeling and processing of Lamb waves that are used to inspect critical plate structures such as airplane panels, wind-turbines, and helicopter blades. We have shown that the interaction of Lamb waves with defects in plates is quite complex, as it depends on the angle of incidence, the incident mode, plate properties, and operating frequencies. We have successfully modeled such complex interaction and supported our theoretical model development by experimental verification. We implemented separate modeling of the independent symmetric and anti-symmetric modes to provide additional capability for enhanced detection and classification of defects. Such models prove critical in designing acoustic sensing systems for SHM. We cast the defect imaging problem in Lamb wave SHM in a sparse reconstruction framework by recognizing that the number of defects is typically small. The emerging compressive sensing techniques are used to achieve high-resolution defect images. New methods are developed to effectively utilize the group sparsity provided by the MIMO structure, different propagation modes, and the spatial extent of defects. Sparse Bayesian learning techniques have been employed to provide robustness for high-resolution signal reconstruction. These techniques are considered attractive in the underlying application due to their effective exploitation of signal structure and prior information relative to flaw contiguity and flow position likelihood. We have applied signal processing techniques to improve discrimination against abnormalities of condition monitoring systems via advanced statistical modeling and analysis of vibration signals from industrial running machines. Faulty rolling element bearings have been detected using techniques based on high-resolution subspace methods, maximum likelihood estimation, harmonic analysis, time-varying filtering, and time-frequency analysis. The NSF PFI grant was instrumental in the establishment of the Acoustics and Ultrasound Lab (AUL) which has become the center of acoustic and ultrasound research and education in the College of Engineering, Villanova University. The AUL conducts experimentations aiding in proof of concept in ultrasound imaging, acoustic sensing and structural health monitoring. The lab promotes collaborations between Mechanical and Electrical Engineering disciplines in the area of non-destructive evaluation and provides the necessary equipment for training undergraduate and graduate students in the broad area of acoustic and ultrasound signal processing.
