Development of automated real-time data acquisition system for robotic weld quality monitoring

Abstract The lack of reliable non-contact, non-destructive, online sensors with the ability to detect defects as they form and with the capacity to operate at high temperatures and in harsh environments is a big obstacle to fully automated robotic welding. This paper presents a non-contact automated data acquisition system for monitoring a robotic gas–metal arc welding process based on laser ultrasonic technology. While a robot welds between two 1040 steel strips, a Nd:YAG Q-switched pulse laser generates ultrasound on one side of the weld by ablation, and a non-contact electro-magnetic acoustic transducer (EMAT) placed on the opposite side of the weld detects ultrasound transmitted through the weld bead. Ablation is employed because high temperature specimens require strong signals to compensate for attenuation within the bulk of the material. The data is then analyzed to determine the time required for ultrasound to travel from the laser source to the EMAT, termed as the time of flight (ToF). When experimental ToF is compared to theoretical ToF, it is determined that surface waves are detected by this system. Therefore, this system can measure weld bead reinforcement distance. In most cases, weld bead geometry is an indication of the weld quality, and can be used as feedback to control a welding process.