How do OCT and AI enable real-time laser welding monitoring and closed-loop control? Explore sensor technologies for keyhole depth, layer thickness, and defect detection. Learn how AI classifies pores and spatter in laser brazing—and how acoustic sensors detect weld failures via sound.
Evolving sensor technology is providing more and better data from laser material processing operations to better monitor welds and identify defects, and to automatically optimize welding processes.
Welding is an ancient technology. Five thousand years ago, Egyptians used it to join copper water pipes. Over the centuries, it evolved into a fundamental technique for fusing various types of metals as experienced welders learned to optimize the process by observing and reacting to the colors in the material and the flame. Welding technology has made more progress in the last 50 years than in previous millennia. A key part of this development relies on optical sensing of process parameters far beyond what our eyes can perceive. Monitoring these parameters has become a technological task in which photodetectors and cameras record radiation during the welding process. Spectroscopic measurements provide precise temperatures with good temporal or geometrical resolution. x-ray and ultrasonic methods allow post-processing evaluation of the characteristics inside the bulk material. Meanwhile, laser beams have driven faster and farther advances in welding technology in recent decades. Not only does the beam transmit large amounts of energy to the joining partners with great precision, but it can also illuminate and observe the welding process. Some modern optical techniques even make it possible to look inside the molten pool.
Welding process development has always been a lengthy experimental endeavor involving much post-processing analysis of weld samples. In recent years, advances in optical technology have helped to accelerate these developments by evaluating weld quality in a much shorter period of time. This capability raises the question of when and where this data can close the loop, making welding an automated process in a self-controlled production environment.
“In fact,” says Peter Abels of the Fraunhofer Institute for Laser Technology ILT in Aachen, Germany. In May, Abels chaired the Process Control and Digitization Forum at the AKL’22 International Laser Technology Conference, which brought together experts from industry and academia to discuss advances in online process control and digitalization. advances in welding.
“Today, we have a wide variety of sensors to monitor the joining process before, during and after the actual laser welding,” he said “They provide a lot of very detailed data.
The purpose of welding is to join two parts. The quality of the weld is usually tested by post-processing analysis such as X-rays or computed tomography. In addition to this NDT method, parts can be cut and etched to assess process performance. It is clear that only fast NDT techniques can contribute to closed-loop process control.
A few decades ago, looking inside the melt pool seemed unthinkable.
Manufacturers of process monitoring equipment have been working for years to develop a way to measure the depth of keyholes in the welding process using Optical Coherence Tomography (OCT) Precitec’s technology sends a broadband laser signal along the path of the welding laser. A portion of this signal is reflected at the end of the molten pool and compared to a reference signal inside the laser head. The interference of the two signals provides the depth of the locked hole. The technique cannot see through the metal, but the OCT signal can pass through the vapor inside the molten pool (Figure 1).
OCT was originally developed to map the inside of the human eye. Applied to the welding process, the imaging technology measures the depth of keyholes with micron-level accuracy and improves throughput by automating and reducing post-processing testing.
One of Precitec’s customers applied OCT process control to the manufacture of its automotive seat adjusters. “The main reason for introducing the sensors was to reduce post-processing non-destructive testing, which resulted in the system paying for itself in a very short period of time.” In this production system, standard deviations in weld depth have been reduced by 40 to 50 percent as a result of active control of the welding laser via OCT data.
For another customer, Precitec uses OCT to measure layer thickness in a directed energy deposition additive manufacturing process. In this process, metal powder is blown onto a surface and melted with a laser beam. Based on sensor data, engineers determined that the speed at which the powder is fed has the greatest impact on layer thickness. (Figure 2) Thanks to the closed loop enabled by the OCT sensor, the collected data can be fed into the Siemens control system.
Combining proven camera-based sensor technology with artificial intelligence (AI) to automate quality control in laser material processing. Initial results have been achieved in laser brazing.
“We have a camera that observes the laser process and records images of the area before, during and after the actual laser process.” “We trained a neural network to recognize problems such as pores or splatter in these images.”
Since 2000, Scansonic has been producing optics for industrial brazing and soldering, including its ALO4 laser head (Figure 3). The device is equipped with a process monitoring system that captures images of the brazing process and feeds them into a neural network capable of classifying pores and spatter.
Porosity and spatter are two key indicators of the quality of the brazing process. Their size and number are used to classify OK and NOK (non-conforming) parts. From the process images and classification data, the system learned to classify seam defects. After initial training, the AI’s neural network was applied to additional samples and the results validated.
In operation, the AI recognizes images with porosity or splatter and then marks the part as OK or NOK. the system works in near real time, allowing for quality assessment during the normal production cycle, a key requirement for in-line quality monitoring.
Currently, the system does not analyze process parameters, but quality data can also be used for process control.
“By combining process and image data, AI will be a powerful tool not only for checking quality, but also for learning more about which parameters need to be adjusted in order to end up with high and consistent production quality.”
Another, less rigorous visual approach to process monitoring uses photodetectors to significantly extend another human sense: hearing.
Experienced welders can listen for signs of problems in the welding process. But human hearing is limited to about 10 to 15 kHz. balthasar Fischer, a physicist and sound engineer, has invented an optical microphone that detects sounds in the range of up to a few megahertz and thus detects changes in the welding process, just like a sophisticated optical sensor.
The sound of interest comes from the molten pool. During the welding process, the temperature of the material changes dramatically: the material evaporates and erupts; cracks form. The welding process becomes a source of broadband ultrasound traveling through the air. A small portion of this sound can be heard when a person listens to the welding process with his ears alone.
Fischer’s microphones can record sound signals from 10 Hz to as high as 2 MHz. The technology is not based on the piezoelectric effect, but rather on the measurement of small changes in air density produced by sound waves in the cavity of a miniature Fabry-Perot interferometer. This interferometer consists of two semi-reflecting mirrors between which a light wave propagates back and forth. As the optical length of the wave varies with the sound-induced change in air density, so does its interference signal – a principle that is independent of frequency. The interference signal is measured and digitized at 2 MHz with 24-bit resolution. The short-time Fourier transform provides up to about 32,000 acoustic spectra per second. The small size of the probe makes it easy to integrate near the process area.
XARION Laser Acoustics has utilized the concept of this small, non-contact microphone and tested it for a wide range of applications, including the automation of non-destructive inspection of spot welds.
At AKL’22’s Process Control and Digitization Forum, XARION’s Wolfgang Rohringer reported on a test using the technology to monitor hundreds of individual weld seams on a sheet metal plate measuring a few tens of square centimeters.
In the acoustic spectrogram he presented (Fig. 5), the acoustic characteristics of the welds clearly indicated the success (OK) or failure (NOK) of each weld. In fact, they identified types of failures with different acoustic characteristics, such as laser misalignment or surface contamination. At the time of the demonstration, XARION was performing an AI-based spectrogram evaluation. Although the raw acoustic data stream is quite small, this type of evaluation can face challenges similar to optical methods, which need to deal with huge data rates (~32 MB/s).
New sensors and data processing methods improve the monitoring of welding operations and open the door to new process control strategies. For closed-loop process control, two challenges exist.
First, the system developer must determine which process parameters (inputs) determine the outcome of the process (outputs). If the input and output parameters can be measured online, the control loop can be closed.
Second, monitoring, feedback, and process control must occur in real time. Increasingly powerful computers are able to process sensor data fast enough to make the necessary changes to the input parameters to keep the output parameters within the desired range. the AI helps to correctly adjust the appropriate input parameters.
In a simple example, OCT records the results of a process, such as layer thickness in additive manufacturing. This data can then be used to control the powder feed rate input parameter. This closes the control loop, and the process can be modified during operation to keep the actual output parameter (the geometry of a particular part) within defined tolerances.
“Different processes require different sets of control parameters, but the path to closed-loop welding process control is clear, “and in addition to the benefits of better quality, we will see far-reaching effects in the digital world.”
Digitalization has already had a big impact on sheet metal forming. For example, when a sheet metal press at the BMW factory in Germany needs more pressure to bend a part, it can transmit information to a press in Brazil that processes the same batch of sheet metal. In a digital factory, process data from one machine can be used to optimize processes in completely different locations. The data can also be used to predict when maintenance is recommended.
In conclusion, the future of laser welding lies in multi-sensor fusion and AI-driven analytics. OCT for depth control, AI for pore/spatter detection, and optical microphones for acoustic monitoring—combined, they enable intelligent, closed-loop welding processes. As digital twins and edge computing advance, laser welding will evolve from manual tuning to autonomous optimization.
