Reference-free damage detection by means of wavelet transform and empirical mode decomposition applied to Lamb waves

Abstract

Guided ultrasonic waves are increasingly used in all those structural health monitoring applications that benefit from built-in transduction, moderately large inspection ranges, and high sensitivity to small flaws. This article describes a monitoring system based on the generation and detection of the guided ultrasonic waves from an array of sparse transducers. In a round-robin manner, ultrasonic waves are generated and measured from all possible different pairs of excitation and sensing transducers. The ultrasonic signals are then processed using continuous wavelet transform and empirical mode decomposition to extract few damage-sensitive features that enable the detection and localization of damage. With respect to most of the existing guided ultrasonic wave–based methods, the proposed approach does not require to record data from a pristine structure (baseline data), and damage is inferred by examining the selected features obtained from all the possible combinations of actuator–sensor pairs of the array. In this study, the method is validated using commercial finite element software to model the presence of 10 ultrasonic transducers bonded onto an aluminum plate. The results are promising and ongoing studies are focusing on the experimental validation and the application to other waveguides.

Optimal sensors layout design based on reference‐free damage localization with lamb wave propagation

“…Also, r i , j ( x , y ) is the relative distance of node ( x, y ) to the actuator and to the sensor associated with the j th sensing path at the i th path type, that is, where d i , j is the length of the actuator–sensor line of sight, is the node–actuator distance, and is the node–sensor distance. The relative distance is zero at any grid node located along the line of sight of the actuator–sensor pair, and it linearly increases at locations away from that path . For illustrative purposes, the relative distance r of an arbitrary sensing path, AB, is presented in Figure .…”

“…The frequency of the excitation signal is an important parameter that must be correctly selected to achieve more accurate damage identification results. 21,35,36 Therefore, a four-cycle sinusoidal tone burst electrical potential function having a 200-kHz center frequency and 10-V amplitude, peak to peak, was chosen for the actuator signal. Figure 3a,b exhibits a Hanning windowed excitation signal of 200 kHz in the time and frequency domains, respectively.…”

“…Therefore, through‐thickness damages cannot generate mode conversion but only transmit and scatter the incoming wave packets. Another baseline‐free damage detection method was based on instantaneous baseline measurements . According to this method, comparison of the receiving data is done for only equivalent sensing paths, that is, arrays of spars transducers with equal distances between sensors and actuators and also similarity in positioning of the transducers from the edge of the test plate.…”

“…Another baseline-free damage detection method was based on instantaneous baseline measurements. [18][19][20][21] According to this method, comparison of the receiving data is done for only equivalent sensing paths, that is, arrays of spars transducers with equal distances between sensors and actuators and also similarity in positioning of the transducers from the edge of the test plate. Obviously, the layout of transducers is obligatory for detecting the damage in this manner.…”

Plate components ultrasonic guide wave detection based on the transducer array

“…Due to the diffusion characteristic of the wave beam inspired by the transducer wafer with a fixed radius b , by supposing the half-angle of diffusion D and the circle transducer array's radius r , the effective detecting area's radius could be expressed as follow: 2 sin d a r D (14) In the detecting practice, the guide wave's effective radiating beam length L must be equal or greater 3 times than the transducer array's radius( 3 L r t ), the received time-domain signal's amplitude m would be greater than the product of detecting voltage and threshold, then the flaw's reflecting signal could be detected. From the analysis above, the ellipse algorithm discrimination for the flaw's direction and shape will be improved with the increasing of the number contained in the array, and the effective detecting area could also be calculated easily.…”