This website uses cookies primarily for visitor analytics. Certain pages will ask you to fill in contact details to receive additional information. On these pages you have the option of having the site log your details for future visits. Indicating you want the site to remember your details will place a cookie on your device. To view our full cookie policy, please click here. You can also view it at any time by going to our Contact Us page.

The welding system of the future learns from its mistakes

20 March 2015

Researchers have developed a welding system that is able to solve quality and productivity issues by predicting what problems may arise ahead of the weld gun.

LUT's neural-network guided self-adjusting welding system (photo: Teemu Leinonen, Lappeenranta University of Technology)

The system, developed by researchers at Lappeenranta University of Technology (LUT) in Finland, is self-adjusting, flexible and adaptable, such that it can be integrated as part of different robotic systems and different manufacturers' power supplies.

It's self-adjusting properties are based on a new kind of sensor system which is controlled by a neural network program. Most often in welding a monitoring sensor is used which tracks the bevel angle, an essential part of the welding process. In the system being developed by LUT, there are also monitoring sensors for the thermal profile (the weld pool's heat values) and the weld form. The monitoring data is transferred from the sensors to the neural network, which is able to deduce and react to simultaneous changes in multiple variables.

"When a mistake is detected, the system is able both to correct it during the welding process and also calculate what other faults may arise," explains project manager, Markku Pirinen. "Thus the final product is flawless."

In the gas-shielded arc welding process, factors affecting outcome quality include the welding current, the arc voltage, the wire feeding and transporting speeds, and the position of the welding gun. With the help of the neural network, a regulating window can be set for these system variables, and they can then be controlled so that they remain within certain limits, which ensures that final product is as required.

"In practice this means that when the welding values approach the boundary values set in the parameter window, the system corrects the process so that the welding values move back towards the centre of the value range and the possible defect is prevented," adds Pirinen.

The new system works well with high-strength steel welding, as the parameter windows for the high-strength materials are significantly narrower than those for construction steel, and the harder the steel is, the more difficult it is to weld. High-strength steel is used, for example, in arctic steel construction work, where the materials used must be light, robust, and strong.

In Pirinen's opinion, the welding industry has been waiting for a control system such as this ever since automated welding came onto the market. "This system will bring significant savings to the welding industry, as resources will no longer be required for post-welding checks and repairs. However, the system can only by used for mass-welding operations, so hand-operated welding will continue to be used for the kinds of work which the robotic welders cannot do."

Print this page | E-mail this page