Welding is a method of connecting metals or other thermoplastic materials together using heating or pressurization. It has been around for hundreds of years. But, before 20th century semi-automatic welding patterns emerged such as buried arc welding, slag welding, etc., workers on the front lines of construction had been struggling. With the rapid development of automation technology, fully automated robots have emerged, and since the introduction of the world’s first Unimate robot in the United States in 1959, the number of industrial robots has been increasing steadily worldwide. Currently, more than half of robots are welding ones. Thanks to this, humans have been able to emerge from the harsh welding environment, using robots instead of manual labor to achieve high-quality welding.
The welding robot market
Welding robots account for up to 50% of industrial robots worldwide since 2010. With the development of automation technology and the continuous innovation of industrial robot technology, the global demand for industrial robots has accelerated significantly. In 2012-2017, global robot sales grew at a compound annual rate of about 19%, and sales of industrial robots are at an all time high. In 2005-2008, the average annual sales of robots worldwide were about 115,000 units, but in 2009 they fell sharply as a result of the financial crisis.
In 2010, restricted investments in 2009 were the main drivers of the increase in robot sales. In 2011-2017, global average annual sales increased to 236,000 units, about twice the average annual sales volume for 2005-2008. In 2015-2017, global average annual sales of robots increased to about 310,000 units, a sign of a significant increase in global demand for industrial robots. According to the latest ifR (International Federation of Robotics) report, global sales of industrial robots continued to grow at a high rate in 2018, reaching 384,000 units, up 1% year-on-year.
In terms of market size, the global industrial robot market was on a steady upward trend between 2013-2018, reaching 16.82 billion US$ in 2018.
Welding robots are an important part of industrial robots. Because welding as an industrial “tailor”, is a very important processing means in industrial production. But also because of the existence of welding soot, arc light, metal splash, that make the welding working environment conditions a decisive factor in the final welding quality. Welding is also one of the most important applications of industrial robots in industrial robot market. Welding robots have been widely used in automobiles, motorcycles, construction machinery and other fields. Welding robots play an important role in improving production efficiency, improving workers’ labor intensity and environment, and improving welding quality.
The importance of welding robot applications
- Stabilize and improve welding quality, to ensure its homogeneity. Welding parameters such as welding current, voltage, welding speed and dry elongation have a decisive effect on welding results. When using robot welding, the welding parameters of each weld are constant, and the quality of the weld is less affected by human factors, which reduces the requirements of the worker’s operating technology, so the welding quality is stable.
- Improved working conditions. With robotic welding, workers only need to load and unload workpieces, away from welding arcs, smoke and splashes. For spot welding, workers do not have to carry bulky hand-welded pliers, freeing workers from high-intensity manual labor.
- Improve labor productivity. Robots can be active continuously for 24h. With the application of high-speed and efficient welding technology, the efficiency of robot welding is improved more obviously.
- Product cycle is clear, easy to control product output. The production beat of the robot is fixed, so the production plan is very clear.
- Shorten the cycle of product modification and replacement, reduce the corresponding equipment investment. Welding automation of small batch products can be achieved. The biggest difference between a robot and a private jet is that it can modify the program to suit the production of different workpieces.
Welding robot technology research status quo
From the current research on welding robot technology, welding robot technology research mainly focuses on weld tracking technology, coordination and control technology of multiple welding robots and peripheral equipment, robot-specific arc welding power supply technology, welding robot system simulation technology and welding process methods for robots 5 aspects.
(1) Research on weld tracking technology
In the process of arc welding robot welding, if the welding conditions are basically stable, or the working conditions of arc welding robot are more suitable, then the robot can generally guarantee welding quality. However, due to various factors such as welding environment, the actual welding conditions often change. For example: due to strong arc radiation, high temperature, soot, splashing, bevel condition, processing errors, fixture clamp accuracy, surface state and thermal deformation of workpieces, etc. are common in these welding situations, often make the torch deviate from the weld, resulting in a decline in welding quality or even failure. 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 welding quality. In order to enable the robot to detect the actual position of the weld in real time during the welding process, the research of arc welding robot weld tracking technology was started. The research of arc welding robot weld tracking technology is mainly based on sensor technology and control theory.
The basic sensor is only a signal transformation element, in recent years, with the emergence of intelligent technology, there has been the emergence of internal sensors with specific processing of the signal, that is, intelligent sensors. Arc sensors and optical sensors play an outstanding role in the study of arc welding robot sensing technology. Arc sensor is from the welding arc itself directly extract weld position deviation signal, real-time good, do not need to attach any device on the welding gun, welding gun movement flexibility and accessibility is the best, especially in line with the welding process low-cost automation requirements. The basic principle of arc sensing is to detect the height and left and right deviation of the torch by using the change of welding parameters caused by the change of the distance between the torch and the workpiece. Arc sensors are generally divided into three categories: side-by-side dual-wire arc sensors, swing arc sensors, rotating scanning arc sensors. Among them, the rotary arc sensor has high sensitivity to deviation detection than the previous two, improves the control performance, and has been widely concerned by welding scholars and related technical personnel at home and abroad. There are many types of optical sensors, including infrared, photoelectric, laser, visual, spectral and fiber optics. In the study of a variety of optical sensors with the most eye-catching visual sensors, due to the large amount of information obtained by visual sensors, combined with computer vision and image processing of the latest technology, can enhance the external adaptability of arc welding robots.
Weld tracking control theory and method
Due to the emergence of fuzzy mathematics and neural networks, and applied to welding this complex nonlinear system, so that weld tracking into a new era, – intelligent weld tracking era. Fuzzy control is an earlier form of intelligent control, which draws on the fuzzy characteristics of human thinking, and uses tools such as membership function, fuzzy relationship, fuzzy reasoning and decision-making in fuzzy mathematics to get control actions. Because the welding robot system has nonlinear and time-changing characteristics, it is difficult to describe with accurate mathematical models, so it is difficult to achieve optimal control by traditional control methods.
Fuzzy control has the characteristics of adaptive and robust, which provides an excellent control method for robot welding control. Combining the fuzzy control theory with the actual welding process, it has developed into a special welding controller, and further developed into a general-purpose welding fuzzy controller. Fuzzy control has better control rules, but the ability of fuzzy control to comprehensive quantitative knowledge is poor.
In addition, the selection of quantitative factors and scale factors also affects the quality of the whole system, and when the dynamic characteristics of objects change, or random interference will affect the effect of fuzzy control, the above problems will lead to fuzzy controller in welding applications encounter more difficult problems to solve. Because of the existence of the above problems, in the field of fuzzy control theory, people have improved the conventional fuzzy control, designed some high-performance fuzzy controller, summarized in 5 forms: the control rules can adjust the fuzzy controller; the fuzzy controller with integral effect; the self-adjusting fuzzy controller of parameters; the composite fuzzy controller; the self-learning fuzzy controller.
Neural network control is the study and use of certain structures and mechanisms of the human brain and human knowledge and experience to control the system. Because the welding process is a time-change, multi-factor, non-linear complex system, strong arc light, power grid voltage fluctuations, welding process specifications and high temperature deformation of the workpiece and other factors interference, making it difficult to use the traditional control system to achieve accurate tracking of welds. Artificial neural network through software or hardware simulation of human brain function, so that the machine can achieve learning, memory, association and other functions, because of the artificial neural network using parallel processing, distributed information storage, with large amount of information storage, strong fault tolerance and so on, from an automated point of view is very suitable for welding tracking in the visual pattern recognition and tracking intelligent control. Compared with traditional expert systems, welding parameters handle neural network systems differently.
- The system can learn directly from examples or pilot data without consulting domain experts.
- The source data may be incomplete, inaccurate or fuzzy.
- When the relationship between input and output data is indirect, no ready-made model or algorithm can be used. Now the welding robot control system is mostly using feed-forward multi-layer neural network, multi-layer neural network learning algorithm is more like Rumelhart proposed in 1986 reverse propagation algorithm, commonly known as BP algorithm.
(2) Simulation technology
In the process of development, design and testing, robots often need to analyze their kinematics, dynamics and track planning and design, while robots are multi-freedom, multi-link space institutions, their kinematics and dynamics problems are very complex, computational difficulty and computer are very large. If the robot as a simulation object, the use of computer graphics technology, CAD technology and robotics theory in the computer to form geometry, and animation display, and then the robot’s institutional design, kinematic positive and negative analysis, operating arm control and the actual working environment obstacle avoidance and collision interference and many other problems to simulate, so that we can solve the problems in the process of developing robots.
(3) Coordinated control technology for multiple welding robots and peripherals
Strictly speaking, welding robot is a welding robot system or workstation, usually including robot body, robot control cabinet, welding machine system and wire feed unit, transposer, tool fixture and other components. In production applications, a single robot often cannot give full play to its role, which requires welding robots and converters, arc welding power supply and other peripheral equipment to achieve flexible integration. During welding, the flexible coordination and control of welding robot and peripheral equipment can help reduce auxiliary time and is one of the keys to improving production efficiency.
For most workpieces, the welds always have flat, horizontal, vertical, back welding and other welding positions. And this has a great impact on welding quality and weld forming, if only by adjusting the robot position to ensure satisfactory joints is quite difficult, but also to the operator brought great inconvenience. If the control converter can be coordinated at this time, so that the workpiece is always in the level of welding position, will greatly improve the welding quality. That is, the converter in the welding process is not static, but to do the corresponding coordination movement. Arc welding power supply and tooling fixtures should also be under the unified control of the robot to do the corresponding coordination movement, in order to ensure that the whole system high efficiency, high-quality work
(4) Dedicated arc welding power supply
In welding robot system, the special arc welding power supply with good electrical performance directly affects the performance of welding robot. At present, arc welding robots generally use melted pole gas protection welding (MIG welding, MAG welding, CO 2 welding) or non-melting polar gas protection welding (TIG, plasma arc welding), melting polar gas protection welding power supply mainly uses thyristor power supply and inverter power supply. In recent years, the technology of arc welding inverter has become mature, robot-specific arc welding inverter power supply is mostly transistor-controlled transistor arc welding inverter, and with fine waveform control and fuzzy control technology, operating frequency 20-50kHz, up to 200kHz, welding system dynamic characteristics are excellent, suitable for robot automation and intelligent welding. There are also special functions of the power supply, such as suitable for aluminum and its aluminum alloy TIG welding square wave AC power supply, welding power supply with expert system. At present, there is a welding power supply using fuzzy control method, which can better ensure that the melting width and depth of the weld are basically the same, not only the surface of the weld is beautiful, but also reduce the welding defects. Arc welding power supply continues to develop in the direction of digitalization, characterized by: stable welding parameters, by network voltage fluctuations, temperature rise, component aging and other factors, with a high degree of repeatability, welding quality is stable, good forming. In addition, the use of DSP rapid response, through the main control system instructions to precisely control the output of inverter power supply, so that it has the output of a variety of current waveforms and arc pressure high-speed stable adjustment function, to adapt to the requirements of a variety of welding methods on the power supply.
(5) Robotic welding process
At present, arc welding robots generally use gas protection welding method, mainly melting pole gas protection welding, that is, rich argon mixture gas protection welding and melting pole (pulse) arc welding; Arc welding robots in advanced countries have been more commonly used high-speed, high-efficiency gas protection welding process, such as double-wire gas protection welding, hot wire TIG welding, hot wire plasma welding and other advanced technology methods, these process methods not only effectively ensure excellent welding joints, but also make welding speed and melting efficiency increased several times to dozens of times.
Welding robots play an important role in high-quality, high-efficiency welding production. The research, development and application of industrial robot technology have strongly promoted the progress of industrial technology in the world. In recent years, the research and application of welding robot technology has made many outstanding achievements in weld tracking, information sensing, offline programming and path planning, intelligent control, power supply technology, simulation technology and software, welding process method, remote control welding technology, etc. With the continuous development of computer technology, network technology, intelligent control technology, artificial intelligence theory and industrial production system, there are still many problems to be solved in the field of welding robot technology, especially the visual control technology, fuzzy control technology, intelligent control technology, embedded control technology, virtual reality technology, network control technology and so on will be the main direction of future research. At present, the application of welding robots has ushered in a rare opportunity for development.
About the authors
David de Brouwer
Founder / Managing Director Engibex
Doctor of Philosophy (PhD), Electrical Mechanics Engineering
Expertise in numerical modelling & experimental data mining. Great enthusiasm in innovative product design & R&D.