A New Fracture Detection Algorithm of Low Amplitude Acoustic Emission Signal Based on Kalman Filter-Ripple Voltage

Abstract

In this study, an acoustic emission (AE) sensor was utilized to predict fractures that occur in a product during the sheet metal forming process. An AE activity was analyzed, presuming that AE occurs when plastic deformation and fracturing of metallic materials occur. For the analysis, a threshold voltage is set to distinguish the AE signal from the ripple voltage signal and noise. If the amplitude of the AE signal is small, it is difficult to distinguish the AE signal from the ripple voltage signal and the noise signal. Hence, there is a limitation in predicting fractures using the AE sensor. To overcome this limitation, the Kalman filter was used in this study to remove the ripple voltage signal and noise signal and then analyze the activity. However, it was difficult to filter out the ripple voltage signal using a conventional low-pass filter or Kalman filter because the ripple voltage signal is a high-frequency component governed by the switch-mode of the power supply. Therefore, a Kalman filter that has a low Kalman gain was designed to extract only the ripple voltage signal. Based on the KF-RV algorithm, the measured ripple voltage and noise signal were reduced by 97.3% on average. Subsequently, the AE signal was extracted appropriately using the difference between the measured value and the extracted ripple voltage signal. The activity of the extracted AE signal was analyzed using the ring-down count among various AE parameters to determine if there was a fracture in the test specimen.