Crack Detection Using Combinations of Acoustic Emission and Guided Wave Signals from Bonded Piezoelectric Transducers

Piezoelectric transducers are broadband devices and can operate over a large frequency range. In structural health monitoring (SHM) applications, piezoelectric transducers are commonly operated at ultrasonic frequencies to excite and sense guided waves, detect acoustic emission (AE) events, and record electro-mechanical impedance (EMI). The transducers can also measure lower frequency information, such as strains resulting from operational loads or vibrations due to environmental factors. This paper presents a crack detection experiment using piezoelectric transducers to provide signals measuring guided waves, AE events, EMI, and strain. The experimental test articles are aluminum dogbone coupons. Each coupon has a small hole drilled in the gage section to serve as a crack initiation point. Each coupon is instrumented with two transducers, both on the same face of the coupon and on either side of the gage section. Each coupon undergoes cyclic tensile loading to initiate and grow fatigue cracks. At various intervals, the fatigue cycling is paused and the coupon is visually inspected for crack initiation and growth. While the cycling is paused, guided waves are generated and sensed by the pair of transducers; each being used in turn as a transmitter or receiver. In addition, EMI measurements at each transducer are made with the cycling paused. Ideally, AE and strain measurements would be made continuously during the cycling. However, for these studies the available hardware did not allow continuous collection of data. As a result, the AE and strain measurements also are taken over limited time segments while the primary fatigue cycling is paused. Collection of data for the various sensing modes continues as the crack grows to the point of coupon fracture. Baseline crack detection performance is established using statistical pattern recognition methods applied to the guided wave data. Additional crack detection algorithms are designed and evaluated to demonstrate the potential benefits of using decision fusion methods applied to combinations of guided wave and AE data. Artificial AE data has been utilized since the current testing hardware did not allow for the collection of useful AE data. Future studies will address the collection of AE data and consider data fusion incorporating EMI and/or strain data.