12 A High-Speed Dual-Stage Ultrasonic Guided Wave System for Localization and Characterization of Defects 127 automatically determine the location of each sensor in the array based on time-of-flight information that can be extracted from each sensor-pair waveform. Using three or four reference nodes at known positions, the computer can compare the time-of-flight information from each sensor to each other sensor in order to determine the maximum likelihood location of each other transducer in the array. 12.3.4 Stage 1 – Detection and Rough Localization Using Rayleigh Maximum Likelihood Estimation Once all of the aforementioned tasks have been performed, DSSHM can be implemented. The following material assumes that damage has only occurred at a single location in the inspection area since the baseline readings were recorded. 12.3.4.1 “Ping” Mode In stage one, the initial procedure is identical to the baseline acquisition phase. Each transducer will take a turn to act as actuator, delivering a narrowband high-frequency Gaussian pulse to the structure. Every other transducer acts as a sensor to record the structural response at that location. The result is another set of Mtime-histories of the voltage produced at the sensor for each actuator-sensor pair. The test waveforms produced during what we refer to as the “Ping” mode will appear similar to the baseline waveforms. The differences between corresponding baseline and test waveforms are due to the scattering of waves from the damage location. Once the test waveforms have been produced, the next step is to extract the variations caused by the damage. This is accomplished by baseline subtraction. 12.3.4.2 Optimal Baseline Subtraction and Damage Detection In this step, the baseline waveform for each sensor pair is subtracted from the corresponding test waveform to reveal the changes caused by scattering from the damage location. The idea is that the baseline time-history and the test time-history for any given actuator-sensor pair will be identical until the arrival of the first scattered wave from the damage location. The time sample corresponding to the arrival of this first reflection will be the first non-noise feature of the difference between baseline and test data (Fig. 12.3). Transducers are shown in blue and the defect is shown in dark gray In order for this subtraction to be effective, however, the correct baseline must be selected. Recall that each baseline in the database corresponds to a different structure temperature. A quick way of choosing which baseline to use for the baseline subtraction would be to simply select the baseline taken at the temperature closest to the temperature at the time of testing. However, it is possible (indeed quite likely) that the temperature throughout the structure is not uniform. Thus, the baseline with the corresponding temperature may not really be the best one to use in the subtraction process. Fig. 12.3 Representation of wave propagation through plate with damage present
RkJQdWJsaXNoZXIy MTMzNzEzMQ==