Process Control Fully Integrated Process Control On-line quality inspection system for laser materials processing Christoph Franz and Michael Ungers Laser beam brazing and welding are well established joining techniques in the automotive industry; in particular for the body in white production. These processes are mainly used to join two-piece trunk lids as well as to connect the roof and sidewall with a visible seam or to weld doors in steel or aluminum. Hence, the requirements upon the optical appearance of the brazed seams are very high and quality monitoring is essential. In cooperation of Scansonic MI GmbH and Fraunhofer ILT, a first prototype of a fully integrated on-line process control system has been developed – SCeye. The system is an add-on for laser welding and brazing processes in the automotive production of bodies in white. It consists of an uncompared innovative and powerful illumination module combined with a high-speed camera; all fully integrated into the ALO3 brazing and welding head, as can be seen in Fig. 1. The illumination and camera module are aligned coaxially, which is a key technology for integrated process monitoring and which has been invented by Fraunhofer ILT many years ago. Without affecting the accessibility of the brazing optics, the system is capable to acquire high resolution images during mass production, which can be used for process and quality control purposes. It allows a full documentation of the process behaviour and the product quality. This article describes features of the SCeye product, which has been released at the EALA congress in January of 2015 and shows furthermore the newest scientific achievements in controlled laser brazing and pore detection. The investigated control algorithm is based on Fig. 1 Technology transfer – left: Scientific setup of ILT; right: SCeye product development of Scansonic. Both setups consist of a high dynamic camera and an innovative illumination module. velocity measurement and on direct control of the feeding wire velocity in real-time. The quality inspection algorithm is demonstrated as an on-line pore detection solution. The SCeye system SCeye is an add-on for laser welding and brazing processes and will first be available for the Scansonic ALO3 head, of which actually more than 1000 heads are in use worldwide. While other inspection systems often require additional control cabinets, SCeye is fully integrated into the Scansonic processing head. The system consists of a CMOS imaging camera with high dynamic range, an illumination module for bright laser illumination on the work piece and an © 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim advanced image and data processing unit attached to the camera directly at the processing head. The data processing unit processes the acquired images in real-time, compresses the data, sends the images via broadcast to the webbased interface (Fig. 2) and additionally stores the raw data on its built-in storage. SCeye for ALO3 assists the use of ALO3 heads in three scenarios: Szenario 1: Teach-in process In teach mode the process field is illuminated by eye safe LEDs. The camera system broadcasts a live video to any network connected client. While programming the robot trajectory, the user may use the live feed to position the processing head relatively to the work Laser Technik Journal 2/2015 23 www.laser-journal.de area homogeneously and ensure crisp images even in bright processes. Whilst still broadcasting compressed live video to the network, SCeye records raw image data by fieldbus command. Each video will be saved on the implemented file system (“first in first out” continuously) while part numbers provided via the fieldbus interface may be used to assign videos to the actually produced part. The system logs the signals of the fieldbus interface and the additional analogue values of the ALO3 head synchronized to the taken process video. Fig. 2 Browser based user interface of SCeye. Szenario 3: Inspection process piece very accurately. At the same time, the system provides continuously actual values of the ALO3 system such as the swivel axis position and the measured forces applied to the feeding wire. Furthermore, the system logs all fieldbus signals coming from the robot and thereby facilitates the programming process at the robot. Szenario 2: Welding/Brazing process During the welding or brazing process, the SCeye system may switch its illumination module from LED illumination to laser VCSEL illumination. Up to 40 W of optical power can then be provided to illuminate the process Company Scansonic MI GmbH Berlin, Germany Scansonic develops and manufactures systems for automated joining in modern production processes. With new methods and products Scansonic expands the limits of what is technically feasible. For its customers, Scansonic opens up the advantages of laser and arc technology – creating higher efficiency, precision and quality in production. In the field of seam tracked brazing and welding Scansonic has reliable and effective solutions. Scansonic also offers the capabilities of a laser application center, where a vast variety of robots, laser sources and different welding and brazing heads can easily be tested and new applications are developed. Since its founding in 2000, Scansonic has gained an established position with its sales and service partners in the international automotive industry. Many parts of the body in white of most OEMs are joined with equipment from Scansonic. While still producing and recording, users are empowered to review the recorded videos and fieldbus data of past processes. SCeye logs up to eight hours of video and data. Thereby, an inspection of the process state can be performed after a faulty process has been detected in a further processing step. Users may also decide to copy and document all acquired data for example after each part or after each shift to their own network attached storage or server in their network. Scientific setup for system development As SCeye is not yet capable of process control, its real time engine, combined with a powerful FPGA, allows the application of advanced algorithms for further monitoring and control in the future as scientific research of ILT shows. To develop, test and evaluate image processing algorithms for monitoring and control purposes, ILT has built a scientific setup, see Fig. 1 left side. It uses the same camera chip and the same illumination principle as realised in the SCeye system. As illumination source the VCSEL technology has been shown to be ideally suitable to ensure homogeneous and directional independent illuminated images [1]. The resulting images are shown in Fig. 3. The entire process zone is visualised; the incoming brazing wire, the liquid melt pool and the solidified seam are visible. Unlike in the SCeye system, the raw images are transferred to an industrial PC unit via CameraLink standard in the scientific setup. They are acquired and processed with FPGA technology making the algorithms suitable for real-time applications such as control purposes and quality inspections during the manufacturing process. The full brazing process is documented in an on-line configuration; machine parameters are monitored as well as the product quality by applying dedicated image processing algorithms to the captured images. In the following two examples of such image analysis are demonstrated. ■ Monitoring of seam imperfections like pores and documentation of the product quality ■ Measurement of the actual movement of the handling system at the tool-centre-point Documentation of product quality Laser brazed seams are frequently used as stylistic elements in the body in white production. After painting they Fig. 4 Documentation of the product quality – pores are marked in a panorama image of the entire brazed seam. www.scansonic.de Fig. 3 Visualization of the process zone. 24 Laser Technik Journal 2/2015 © 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Process Control part where seam is already solidified. This opens up the possibility of saving an image of each joint during mass production and will massively reduce the amount of data that needs to be saved for quality assurance purposes over the long-term view. Application for process control – controlled laser brazing Fig. 5 Controlled laser brazing – laser power PL as well as wire feed rate vW are controlled based on the actual velocity measurement. are direct visible to the end user, making pores in the surface of the seam an unacceptable seam defect. Hence, it is important to detect those seam defects. As pores are mostly open to the surface they remain as dark spots in the bright illuminated seam. It seems to be obvious to detect such dark spots via blob detection for example. But this approach is non-satisfying as it depends on thresholds and the actual illumination situation. One approach that is more promising is to use a classification-based detection of pores, as demonstrated in [2]. Dedicated image features are used to classify the solidified seam into defective or non-defective parts. Mean value and standard deviation of the intensity and the gradient image are used for classification. Trained values for defective and non-defective parts are compared on FPGA technology making it possible to judge the joint quality in a real-time configuration. Pores in a range of diameter from a few hundred microns up to a few millimetres can be detected with this approach. The results are visualised in panorama images as shown in Fig. 4; pores are marked in red whereas non-defective parts of the seam are marked in green. This classification algorithm benefits from the power-full and homogeneous VCSEL illumination, developed for the SCeye system. Moreover, the high image quality makes it possible to put together the entire seam out of small image patches taken form the image © 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Despite form quality inspections imaging based process control is also demonstrated successfully to measure machine parameters. Beside joint tracking and detection of the laser spot position relative to the joint the measurement of the process, velocity is one main parameter that can be measured on basis of the acquired images [3]. Due to the homogeneous and directional independent illumination image processing algorithms can be applied to the acquired images in order to track a specific pattern in two sequenced images. A displacement vector can be determined and in combination with the acquisition rate the resulting velocity is calculated. In this case a block-matching algorithm [4] is used as it can be easily implemented on FPGA technology. This approach has not only real-time capabilities but also suits as input single for velocity based controlling strategies, e. g. to ensure a constant energy per unit length. In the case of laser brazing laser power PL as well as the velocity of the filler wire vW needs to be adjusted pro- Laser Technik Journal 2/2015 25 www.laser-journal.de portionally to the measured velocity [5]. The results of controlled laser brazing are shown in Fig. 5. The system has been tested under process conditions for brazing a flanged joint configuration. In the experiments the velocity was varied in a range from 3 down to 0.72 m/min. Despite this wide variation the controlled brazing process remains stable and fluctuations in the velocity of the guiding robot system are compensated. Furthermore, a smooth and nearly homogeneous optical appearance of the seam surface is achieved. A relevant application one can think of is the joining of the boot lid where the laser head needs to be realigned during the brazing process due to the geometry of the work piece and the brazing velocity drops down to low values while the optics is turned around the edge of the numberplate’s salient. Conclusion and Outlook An on-line quality inspection as well as closed loop controlled laser brazing has been demonstrated successfully. Online inspection of the seam quality will open up the possibility to disclaim post process inspections and reduce effort for quality assurance. Whereas controlled laser processes will enhance the process stability and benefits the product quality especially when process windows are small. In particular teach-in procedures will benefit from the measurement of machine parameters. Future work will not only concentrate on further enhancement of algorithms for velocity measurement but also think of new monitoring and controlling tasks, e. g. the detection of brazing wire tip or position of the laser spot relative to the joint geometry. Since the actual SCeye system is not yet capable of automated process inspection, some of the features shown with the scientific setup of Fraunhofer ILT may find their use in future production enhancements. In the long-term view, the goal is to integrate full process control, i. e. control and monitoring algorithms and connect more data sources and provide automated parameterisation, to reduce the complexity in the use of welding and brazing heads for their users. As algorithms are suitable for the implementa- tion on FPGA technology, no additional control cabinet or any external processing unit will be needed to monitor and control the laser joining process in networks of sensors and actors. DOI:10.1002/latj.201500013 [1] U. Thombansen, M. Ungers: Illumination for Process Observation in Laser Material Processing, Physics Procedia 56 (2014) 1286; DOI: 10.1016/j.phpro.2014.08.053 [2] M. Ungers et al.: FPGA-Programmed Detection of Seam Defects for the Application of Laser Brazing. In: Proceedings of LAMP2013. The 6th International Congress on Laser Advanced Materials Processing, (2013) 23–26.07.2013. Toki Messe, Niigata, Japan. [3] S. Kaierle et al.: Understanding the Laser Process, Laser Technik Journal 7 (2010) 49; DOI: 10.1002/latj.201090027 [4] D. Liu, W. Sun: Block-Based Fast Motion Estimation Algorithms in Video Compression, Department of Electrical Engineering and Computer Engineering, Stevens Institute of Technology, Hoboken, NJ 07030, (1998) [5] M. Ungers et al.: Hardware based Analysis and Process Control for Laser Brazing Applications, Physics Procedia 41 (2013) 517; DOI: 10.1016/j.phpro.2013.03.111. Authors Christoph Franz Institute Fraunhofer-Institute for Laser Technology ILT Aachen, Germany With more than 250 employees and 10,000 m² of usable floor space the Fraunhofer-Institute for Laser Technology ILT is world-wide one of the most important development and contract research institutes of its specific field. The activities cover a wide range of areas such as the development of new laser beam sources and components, the use of modern laser measurement and testing technology and laser-supported manufacturing. This includes for example laser cutting, caving, drilling, welding, soldering and brazing as well as surface treatment, micro-processing and rapid prototyping. Furthermore, the Fraunhofer Institute for Laser Technology is engaged in laser plant technology and process control as well as the entire system technology. 26 studied mechanical engineering in Aachen and received his Diploma in production technologies at the University of Applied Sciences Aachen. In a total of seven years of experience in process monitoring and control of laser processes at Fraunhofer ILT, Aachen and at Coopération Laser Franco-Allemande in Evry, France, he worked as project manager for scientific and industrial projects and solutions for process recording, monitoring and control of a vast variety of laser processes. Since 2012 Christoph Franz works as product manager at Scansonic MI GmbH for process monitoring products as well as for laser edge welding heads.. Michael Ungers studied physics at RWTH-University Aachen and at HeriotWatt University of Edinburgh, Scotland, and received the diploma in physics as well as the Master of Physics MPhys. Since six years he is engaged as scientific staff in the Process Control and System Technology Group at Fraunhofer ILT. As project manager for scientific and industrial projects he is involved in research and development of process observation systems, process monitoring and process control. During the last years he works especially in the field of process monitoring, on-line quality control and controlled laser processes for the application of laser brazing. www.ilt.fraunhofer.de C. Franz, Scansonic, Rudolf-Baschant-Straße 2, 13086 Berlin, E-mail: christoph.franz@scansonic.de, M. Ung­ ers, Fraunhofer-Institut für Lasertechnik ILT, Steinbachstraße 15, 52074 Aachen, E-mail: michael.ungers@ ilt.fraunhofer.de 2/2015 © 2015 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Laser Technik Journal