Abstract

In this study, the correlation between welding quality and features of acoustic emission (AE) signals collected during laser microwelding of stainless-steel sheets was analyzed. The performance of selected AE features for detecting low joint bonding strength was tested using a developed monitoring system. To obtain the AE signal for analysis and develop the monitoring system, lap welding experiments were conducted on a laser microwelding platform with an attached AE sensor. A gap between the two layers of stainless-steel sheets was simulated using clamp force, a pressing bar, and a thin piece of paper. After the collection of raw signals from the AE sensor, the correlations of welding quality with the time and frequency domain features of the AE signals were analyzed by segmenting the signals into ten 1 ms intervals. After selection of appropriate AE signal features based on a scatter index, a hidden Markov model (HMM) classifier was employed to evaluate the performance of the selected features. Three AE signal features, namely the root mean square (RMS) of the AE signal, gradient of the first 1 ms of AE signals, and 300 kHz frequency feature, were closely related to the quality variation caused by the gap between the two layers of stainless-steel sheets. Classification accuracy of 100% was obtained using the HMM classifier with the gradient of the signal from the first 1 ms interval and with the combination of the 300 kHz frequency domain signal and the RMS of the signal from the first 1 ms interval.

Highlights

  • Laser microwelding has received considerable attention in past decades for the manufacturing of high-precision products in the electronics industry

  • The correlation between acoustic emission (AE) signals and joint bonding strength resulting from laser lap microwelding was analyzed and verified by applying the traditional feature extraction method and the developed hidden Markov model (HMM) classifier

  • The results show that the root mean square (RMS) and gradient of signals obtained during the first millisecond of welding and the 300 kHz frequency signal feature are promising features for identifying low-strength joints based on the conditions in this study

Read more

Summary

Introduction

Laser microwelding has received considerable attention in past decades for the manufacturing of high-precision products in the electronics industry. Some studies have investigated the correlation between AE signal features and defects resulting from keyhole-mode welding [10], and AE-based quality monitoring has been studied. Fang et al [16] observed that the frequency-domain AE signal correlated closely to the cold crack is at around 200 kHz. Shao et al [6] reviewed studies involving the AE signal for the development of the monitoring issues in welding. The capability of the AE signal to monitor defects in welding is confirmed in most reports, the installation of AE sensors on a workpiece makes it not easy to be implemented in the production line. In 2014, close relationships between laser spot welding and AE signals were reported by Lee et al [19], and the selected features were adopted as the input features to a back-propagation artificial neural network to predict the weldability of stainless-steel sheets. For keyhole-mode monitoring, hybrid monitoring systems have been developed that integrate AE signals with other sound or optical signals [10,20]

Methods
Results
Conclusion
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call