Industry News

Research Status and Development Trend of Weld Seam Tracking System for Arc Welding

Since the invention of arc welding in 1881, with the mechanical, electrical, materials and other industries, welding technology has become one of the most important metal thermal processing technology. From the production of daily necessities to the manufacture of cutting-edge technology products are inseparable from the welding technology, modern production practice on the welding technology to further improve efficiency, optimize the quality and improve the working conditions and other requirements, welding automation is the welding workers to comply with this requirement Concrete manifestation. Industrial developed countries, such as the United States, Japan, Germany, welding automation, mechanization has reached 60% to 70%, while China is only 20% to 30%. Therefore, the development and application of new welding automation technology will play a huge role in China's national economy, and arc welding robot weld tracking system is an important aspect of welding automation research. In 1960, ConsoIidated ControI of the United States developed the world's first robot prototype according to DevoI's patented technology and set up Unimation, which produced Unimate (meaning "universal automatic") robot. Since then, the robot research, production and application has been a great development, by the end of 1997, the world's industrial robots have reached more than 80 million units, of which more than 40% of the robots for welding. In the introduction of foreign technology on the basis of China's early 70s began to study industrial robots. In 1985, Harbin Institute of Technology successfully developed China's first HY - 1 welding robot; in 1987, the Beijing Machine Tool Research Institute and South China University of Technology jointly developed for the Tianjin Bicycle Plant for the production of bike front tripod TJR - G1 type Arc welding robot; Shanghai Jiaotong University developed the "Shanghai 1", "Shanghai 2" and the Guangzhou Machine Tool Research Institute developed the teaching reproduction of the robot also have for arc welding and spot welding function, and some Automatic tracking, looking for starting points and maintaining posture and other functions [1].


Most of the current robots reprogram the robot for programmable teachings. This robot can accurately complete the teaching operation in its workspace. In the arc welding robot welding process, if the welding conditions are basically stable, the robot can guarantee the welding quality. However, due to various factors, the actual welding conditions often change. For example, due to strong arc radiation, high temperature, dust, splash, groove conditions, machining errors, fixture clamping accuracy, surface conditions and thermal deformation of the workpiece will cause the torch away from the weld, resulting in decreased or even failed welding quality. Therefore, this change in welding conditions requires the arc welding robot to detect the deviation of the weld in real time and adjust the welding path and welding parameters to ensure the reliability of the welding quality [2]. In order to make the robot in real time to detect the actual position of the weld in the welding process, people began to weld the welding robot tracking technology research.


1. The structure of the seam tracking system


Arc welding robot weld tracking system, although after several decades of development and the strong application of computer hardware and software, but the whole system structure is not much change. Arc welding robot weld tracking system structure generally includes sensors, PC processor, robot dedicated controller, robot body and welding equipment. The sensor sends the signal to the PC, after a series of data processing and image display [3], PC and robot dedicated controller for data communication, and then send the control signal to the robot body control the welding process of the correct operation. In the system structure, the torch and the robot body are fixed together, by the multi-degree of freedom of the robot to control the direction of the gun, displacement and speed. The position of the sensor is generally connected to the robot body via a mechanical interface, even with a steel connection and moving together. PC machine in the arc welding robot weld tracking system is the application of the past few years, in the process of tracking the system, the PC can receive the signal received by the sensor, through the image processing software in the visual interface to display the welding process Of the working state, PC and robot dedicated controller for data communication, according to the user's requirements in the case of offline programming on the welding process to track control. The system block diagram is shown in the figure.

In the whole closed-loop system, the sensor plays a very important role, it determines the tracking accuracy of the entire system on the weld. In the welding process, the sensor must accurately detect the weld (groove) position and shape information, and then sent to the controller for processing. With the rapid development of large scale integrated circuits, semiconductor technology, optical fiber and laser, the welding technology has been promoted to the automation and intelligent direction, and a variety of sensors have been developed for seam tracking. They are mainly for detecting electromagnetic and mechanical And other physical quantities of the sensor. We know that in the process of arc welding, there is a strong arc, smoke, electromagnetic interference and high temperature radiation, soot, splash, welding process with heat and mass transfer and physical and chemical metallurgical reaction, the workpiece will produce thermal deformation, so , For arc welding of the sensor must have a strong anti-jamming capability. With the rapid development of technology such as electronics industry, computer hardware and software, network communication and image processing algorithms, mechanical control, electrical and material technology are increasingly updated, and new systems such as sensors, controllers, control software and robots have been introduced. Arc welding robot weld tracking control technology has become more advanced and complex.


2 sensing technology


A sensor is a device or device that is capable of sensing a specified measurement and can be converted into an available signal to implement information detection switching and transmission. With the rapid development of large scale integrated circuits, semiconductor technology and optical fiber and laser, modern welding technology is developing in the direction of automation and intelligence. Many of the world's leading welding equipment research and manufacturing institutions are working to develop this area. Welding process control system first to solve the problem is the seam tracking. After several decades of research and practice, weld tracking technology has made great progress, and the new type of seam tracking sensor is the emergence of this progressive signs. Arc welding sensors can be divided into direct arc, contact and non-contact 3 categories. According to the working principle can be divided into mechanical, electromechanical, electromagnetic, capacitor, jet, ultrasound, infrared, photoelectric, laser, vision, arc, spectrum and fiber type. (Such as temperature distribution, plasma particle density, pool behavior, etc.) for weld tracking, welding condition control (weld width, penetration, penetration, forming area, welding speed, cooling rate and dry elongation) The According to a recent survey conducted by the Japan Welding Technology Institute, 80% of the sensors used in the welding process in Japan, Europe and other developed countries are used for seam tracking. At present, China is more use of the arc, mechanical and photoelectric.


The arc sensor is directly from the welding arc itself to extract the weld position deviation signal, real-time good, do not need to attach any device on the torch, torch movement flexibility and accessibility of the best, especially in line with the low cost of automation requirements. The basic working principle of the arc sensor is: when the arc position changes, the arc of its own electrical parameters corresponding to change, reflecting the welding gun tip to the workpiece groove surface distance changes, and then according to the arc of the swing form and the welding gun and the workpiece relative Position relationship, derived from the welding gun and the relative position between the amount of deviation. Static parameters of the static changes and dynamic changes can be extracted as a characteristic signal to achieve the level of the two directions of the tracking control. At present, the widely adopted method is: the distance H = f (I, U, 1) between the workpiece and the welding wire is calculated by measuring the welding current I, the arc voltage U and the wire feed speed 1, and the fuzzy control technology is used Weld tracking. Arc sensing structure is simple, fast response, has been widely used. However, it is mainly applied to the symmetrical sidewall of the groove (such as V-groove), and for those asymmetric sidewalls or simply no side wall of the joint form, such as lap joints, do not open the groove of the butt joints Form, the existing arc sensor is not recognized [4].


A typical contact weld tracking sensor is to rely on the groove in the rolling or sliding fingers to the welding gun and the weld between the position of the deviation reflected in the detector, and use the detector built-in micro-switch to determine the polarity of the deviation, In addition to the micro-switch type, the detector to determine the polarity and size of the deviation there are potentiometer, electromagnetic and photoelectric. Contact sensors are suitable for type X, Y-groove, narrow gap weld and fillet weld. The system structure is simple, easy to operate, cheap and free from the arc soot and splash interference, but also the use of a wide range of welding tracking sensor. The problem is: the need for different different groove probe, the probe wear large, easy to deformation, point solid point obstacles difficult to overcome, not suitable for high-speed welding.

At present, there are many non-contact sensors for seam tracking, mainly electromagnetic sensors, photoelectric sensors, ultrasonic sensors, infrared sensors and CCD vision sensors. According to the Japan Institute of Welding Technology on the use of arc welding robot in Japan, the results show that leading the future of the weld tracking sensor is mainly optical sensors. Among them, the visual sensor is the most compelling, because the information obtained by the visual sensor is large, combined with the latest technological achievements of computer vision and image processing, the external adaptability of arc welding robot is greatly enhanced. The photoelectric conversion device used in the visual sensor is the simplest unit photosensitive device, such as photodiode, etc .; followed by one-dimensional photosensitive unit linear array, such as linear array CCD (charge-coupled device); application is the most complex structure of the two Photographic element array, such as the area array CCD, is a two-dimensional image of the conventional photosensitive device, which represents the current stage of the development of sensors, and thus increasingly widely used. In the welding robot a variety of visual sensors, CCD sensor because of its reliable performance, clear and intuitive images and the use of the effect has been generally attention. Especially since the 80s, CCD and high-performance microcomputer combined with the weld tracking system, so that the study of seam tracking on a new level.


3 PC machine in the arc welding robot weld tracking system application [5]

In recent years, with the development of microelectronics technology to PC as the representative of the computer hardware and software has been an unprecedented development, the manufacturing industry to establish FM as the goal, to the "open factory automation" direction. In order to conform to this trend, the arc welding robot controller in the weld tracking control system must be switched by a dedicated controller to a PC-based open universal controller. The application of PC in Weld Tracking Control System not only solves the problem of heavy load and poor real-time performance of special controller, but also can be used in other fields such as image processing, voice recognition, optimal control, artificial intelligence and other advanced research results Applied to the system in real-time operation. In addition, PC has a good openness, security and networking, the standard real-time multi-tasking operating system, the standard bus structure, the standard interface and other characteristics, breaking the robot controller structure closed the situation, to develop structural openness , Functional modularized arc welding robot weld tracking control system. At present, the system application of PC is generally in the original robot dedicated controller configuration and PC communication bus and interface board, with the corresponding communication software and the development of the corresponding application software and interface board, the user can The PC reads the robot to track the current position of the weld and real-time control of the movement of the robot to monitor the working status of the robot's seam tracking system.

4. Application of artificial intelligence technology in the system

With the development of welding automation, the automatic tracking of welds has been paid more and more attention. The previous seam tracking system is often based on the classical control theory, using the corresponding controller to control. But the welding is a very complicated process, due to the weld groove processing, welding fixture, welding car walking deviation and welding thermal deformation and other reasons, will cause the welding torch position deviation, and these factors are uncertain , Non-linear, therefore, in the weld automatic tracking system using classical control theory and methods to control, in the actual welding production often can not get satisfactory results. In order to improve the flexibility and precision of the seam tracking system, the use of advanced sensor technology and the introduction of artificial intelligence technology in the arc welding tracking system, improve the control algorithm is an inevitable trend.

Artificial intelligence is an emerging technical discipline that studies how to imitate, extend and extend human intelligence, and achieve some "machine thinking" or mental labor automation. Therefore, from the perspective of automation, artificial intelligence is the technical basis of mental labor automation; from the engineering point of view, it is a multi-disciplinary knowledge and the formation of a high-tech [7]. At present, there are three kinds of artificial intelligence methods in the seam tracking system, namely, expert system, fuzzy control and artificial neural network technology. The so-called expert system is specifically designed to imitate a field of experts knowledge and reasoning methods, and can make their knowledge used by others computer software system. The system generally includes knowledge systems, development environments and operating environments. Knowledge systems are systems that contain all expert knowledge and are flexible to apply; the development environment includes the necessary software tools needed to develop expert systems; the operating environment is software and hardware that communicate with the outside. The knowledge system is the core of the expert system, which consists of three main parts: knowledge base, reasoning machine and user interface. The development of experts in the field of welding began in the mid-80s, Britain, the United States, Japan, Germany and other countries have carried out this research and development work. China from the late 80s began the development of welding expert system [8]. At present, it seems that the installation of welding expert system is not mature enough, most of the expert system reasoning process is simple, mostly for fruit-based reasoning, and rarely related to attribution reasoning. All welding expert systems only cover the knowledge and intelligent reasoning of artificial intelligence, and generally lack the ability to acquire new knowledge. The future development direction is: integration and network communication will be further strengthened; expert system development tools will be updated to Windows as the development platform, combined with ANN, 00, FUZZY and other artificial intelligence development tools will continue to emerge and become mainstream; The ability to automatically acquire knowledge will continue to improve; the development of meta-knowledge processing capacity of the welding expert system; multimedia technology will be applied to the system.

In recent years, with the development of fuzzy control theory and technology, especially it has a more complete mathematical theory, fuzzy control in the seam tracking has been widely used, so that the seam tracking system to adjust the speed and tracking accuracy Has been greatly improved. Japan and other countries a large number of fuzzy control technology for seam tracking and welding dynamic process control, and achieved satisfactory results. Tsinghua University, Harbin Institute of Technology, South China University of Technology and other units in the arc welding process of fuzzy tracking and control of a large number of fruitful research work [9]. Artificial neural network (ANN) is beginning to emerge with the development of neuroscience and brain function research and has developed rapidly. Compared with the traditional expert system (ES), the artificial neural network has the characteristics of self-learning, self-organization, fault tolerance and self-repair, pattern recognition and retrieval [10]. The characteristics of artificial neural network provide an important way to develop the welding control system of arc welding intelligent control system, especially high intelligent arc welding robot, and apply the technology to the weld tracking system of arc welding robot at home and abroad. Many successful examples. Neural network parallel processing ability, adaptive ability and strong fault tolerance and mapping force is the development of visual information with parallel processing capabilities and intelligent seam tracking control system required characteristics.