Anomaly detection in laser metal deposition with photodiode-based melt pool monitoring system

Abstract

A lack of quality assurance is a common concern in laser metal deposition (LMD) additive manufacturing and mainly stems from undetected equipment and/or material exceptions. In-situ process monitoring based on detecting melt pool emissions via photodiodes has significant potential for addressing this issue, but the inherent correlations between the optical emissions and LMD process parameters are lacking. In this work, the characteristic changes in melt pool emissions with variations in the laser power, scanning velocity, and powder feeding rate were carefully examined using a statistical method, and a criterion was proposed for estimating parameter variations in terms of emission signals. Based on this criterion, several common exceptions in LMD processes can be detected via low-cost photodiode measurements. The detected emission characteristics (e.g., average intensity and variable coefficient) are closely associated with the melt pool temperature and stability and are affected by powder sheltering phenomena. These characteristic quantities can be used as synthetic parameters for process monitoring in industrialized applications, i.e., instead of monitoring the laser power, laser movements, and powder feeding conditions separately.