Development and trend of welding robot tracking technology
Weld seam tracking as a comprehensive application of technology, with the characteristics of multi-disciplinary cross-integration, including electronic technology, computer, welding, structure, materials, fluids, optics, electromagnetic and other disciplines, many domestic and foreign researchers in this field Research, from teach-in robots to program-controlled welding systems, to mobile automatic seam tracking technology, dramatically improves productivity with every advance in welding automation. Welding technology automation, flexibility and intelligence is the future development of welding technology inevitable trend.
1 weld tracking sensor development
The sensor is a key part of the automatic weld tracking system. Its role is to accurately detect the weld position and shape information and converted into electrical signals. The control system can process the signal and control the automatic adjusting mechanism to adjust the welding torch position according to the test result so as to realize the automatic tracking of the welding seam.
Arc welding sensors generally can be divided into direct arc, contact and non-contact three categories. According to the working principle can be divided into mechanical, electromechanical, electromagnetic, capacitance, jet, ultrasound, infrared, optoelectronics, laser, vision, arc, spectroscopy and fiber optic and so on. Here are some common seam tracking sensors:
Contact sensors are the earliest used sensors, which are characterized by arc interference, reliable work, low cost, have been widely used in production, but due to the tracking accuracy is not high, wear large, easy to deform, not suitable for high-speed welding, Currently being replaced by other sensing methods.
Acoustic sensors, especially ultrasonic sensors, have the advantages of simple structure, high precision and low price. Ultrasonic sensors by the occurrence of ultrasound and receiving devices. Ultrasonic sensor measurement accuracy depends mainly on the ultrasonic frequency, the higher the frequency, the smaller the error, the general ultrasonic frequency 1.25-2.5 MHz. Ultrasonic sensors are not afraid of welding electromagnetic, light, smoke and dust interference, but susceptible to noise interference, more sensitive to noise, such as CO2 gas shielded welding and other welding methods have some restrictions.
Arc sensor works in the welding process, when the relative position between the welding torch and the workpiece will cause changes in the arc voltage and current, these changes can be extracted as a characteristic signal to achieve the torch height and left and right two Direction of tracking control.
The arc sensor uses the arc itself as a sensor, which has the advantages of simple structure, convenient and flexible, no interference from arc, magnetic field, splash, soot, etc., and has the characteristics of fast response, high precision and strong anti-interference. However, the swinging or rotating mechanism of the welding torch is complicated and the coupling between the arc parameters is very strong. The actual waveform obtained does not achieve the expected effect. Therefore, the obtained data needs to be filtered and the control amount determined based on a large amount of experience. In the case of an asymmetrical side-wall or an essentially no-side-wall joint, existing sensors do not recognize it.
The photoelectric sensor with high precision and good reproducibility can detect, weld and track the shape, width and cross section of bevel and provide the basis for adaptive control of welding parameters. Photoelectric sensors can be divided into discrete photoelectric elements based on the single-point photoelectric sensors and can obtain the groove image information of the visual sensor.
The most simple photoelectric conversion device adopted by visual sensor is a unit photosensitive device such as a photodiode, followed by a one-dimensional photosensitive cell linear array such as a linear CCD (Charge Coupled Device). The most widely used is a two-dimensional Sensing unit area array, such as the linear CCD is a two-dimensional image of the conventional photosensitive device, represents the current development of the latest stages of the sensor, which is increasingly widely used. Among the various kinds of welding robot vision sensors, CCD sensors have received universal attention due to their reliable performance, small size, low price and clear and vivid images. Depending on how the welding robot's vision welding system works, vision sensors for the welding robot's visual seam tracking system can be categorized into three types: structured light, laser scanning and direct shooting arc. Among them, structured light and laser scanning are active visual methods, direct shooting arc is a passive visual method.
2 intelligent control method in the weld tracking the development and application of the situation
Modern intelligent control mainly uses man's operation experience, knowledge and reasoning rules. At the same time, it uses some information provided by the control system to derive the corresponding control actions so as to achieve the expected control purpose. In the weld tracking system, the development and application of the following conditions:
2.1 fuzzy control method in the weld tracking the development and application of the situation
Fuzzy control absorbs the fuzziness of human thinking, and uses the tools of membership function, fuzzy relation, fuzzy reasoning and decision-making in fuzzy mathematics to get the control action. The most prominent advantage of fuzzy control is that there is no need to establish a mathematical model of the control system, and its control decision table and control rules are pre-summarized based on experience. According to the control rules, error and error transform rate of fuzzy subset to generate control decision table, through the direct query decision table can be obtained at each moment should be applied to the control system control action, so as to achieve the purpose of real-time control. In fuzzy control, we need to establish a fuzzy control rules table, usually by summing up the actual control experience and obtained by fuzzy reasoning.
As early as 1985, D.Lakov of Bulgaria proposed the fuzzy model to describe the uncertainty of the arc welding process. With the aid of the non-contact laser sensor, it can trace the welding seam according to the teaching contents. The experimental results show that using the The concept of fuzzy sets enables online assessment, prediction and control.
Japan's S.Murakami et al. Developed a welding seam tracking system based on fuzzy control. The control system determines the horizontal and vertical displacement of the welding spot according to the amplitude of the welding gun and the distance between the welding wire and the workpiece. According to the language rules, a fuzzy filter And fuzzy controller, the control effect is very good.
A new generation of laser weld sensor was used to measure the position of the weld in the system designed by CAO Lit-ting at Beijing Union University in China. Fuzzy-P dual-mode segmentation control was used to correct the weld, and good experimental results were obtained.
2.2 Artificial Neural Network Control Method in the weld tracking the development and application of the situation
Artificial neural network control is based on the study of human brain structure and function, through the simplification, abstraction and simulation, the establishment of neural network model, and then through the corresponding computer system to achieve the process control reflecting the structure and function of the human brain to deal with the problem. At present, the most widely used, the most intuitive idea is the error propagation neural network and BP network, BP network is characterized by inverse error propagation, which is based on the network's output and the actual output of the network error signal error, The middle layer corrects the connection weights and the output thresholds of each cell layer by layer, and then the BP algorithm solves the minimum of the error function. The samples are repeatedly trained and the weights are modified to reduce the deviation until the satisfactory accuracy is reached.
Japan's Y.Suga et al. Applied neural network to weld tracking. In this system, a visual sensor is adopted and the neural network is used for image processing to obtain the shape data of the weld. The experimental results show that the system has strong robustness Sex, can effectively weld tracking.
Contact: David Piao
Address：No.99 North Wusongjiang Road, Kunshan, Suzhou, Jiangsu, ChinaPR 215300