Printer Friendly

Seeing-eye seam tracker.

Seeing-eye seam tracker

The challenge of fully automating the arc-welding process is to guarantee that an adaptive robotic system can sense (or see) a seam and weld puddle, and then adjust program path and welding parameters on the fly. This was exactly the need of General Electric's Aircraft Engine Business Group.

At AEBG, TIG welding of engine structural components is a major work element. Although there were strong economic and quality motivations for automating this task, virtually none of the welds could be performed reliably by existing robots. And, despite all the activity in the robot welding arena, no university or equipment manufacturer had been exploring a full-featured closed-loop welding system. For this reason, GE's Research & Development Center (Schenectady, NY) initiated a project to develop such a system.

"We decided the system required seam tracking for closed-loop control of robot motion and closed-loop control of weld-process parameters, including weld penetration, wire feed rate, puddle position, and electrode stand-off distance,' reports Dr Norman R Kuchar, manager of the R&D Center's Process Technology Branch.

Welding engineers at AEBG noted that humans follow weld seams and control the welding process mostly by watching the seam and weld puddle. Even characteristics such as penetration, which a weldor can't see directly, are inferred from information obtained visually from the puddle.

This idea initially was explored by Professor Dick Richardson at the Ohio State University Center for Welding Research (Tooling & Production, June 1983, pg 49). Encouraged and partially funded by GE, he demonstrated a technique for optically imaging a weld puddle along the axis of the welding electrode. Based on this concept, GE began developing a vision sensor for weld-process control.

"We brought electricity, gas, and cooling water in from the side, freeing up the torch barrel,' says Dr Kuchar. "Then in the barrel we placed a lens that focused on the welding area. The image was transferred by a fiber-optic bundle to a TN2500 solid-state video camera.

"The camera provides digitized output of light intensity that subsequently is stored in a frame buffer. Further, we developed an algorithm for reliably identifying the weld puddle's edge based on light intensity variations. The system analyzes about 5 puddle images/sec--fast enough for TIG process control.'

The vision sensor, integrated in the welding torch, also provides a means for in-process, noncontact seam tracking. GE engineers decided to use structured light, where two laser-light stripes are projected about 0.25 in front of the electrode and onto the workpieces. By projecting at an angle, 3-D information about the seam is revealed. The image is captured by the same optics and camera that monitor the weld puddle.

"A break in the stripes marks the location of the seam,' points out Dr Kuchar. "A microcomputer analyzes the light intensity pattern to identify seam boundaries, determine local seam centerline orientation, and calculate seam width. All this happens at about 10 times/sec; minimum detectable seam width is about 0.01.'

Visual parameters, however, don't perform direct control functions. Variables for this purpose include arc current, arc voltage, welding speed, and filler wire feed rate. New algorithms were needed to link these bona fide process controllers to the specified weld characteristics and visually generated data.

According to Dr Kuchar, several algorithms were developed with this in mind. One, based on the measured geometry of the weld puddle, is used for controlling weld penetration. If the puddle differs from a known pattern (which depends on workpiece characteristics), arc current is changed automatically.

A second algorithm compares the measured puddle location with the measured location of the seam, automatically adjusting the electrode position transverse to the seam if the puddle and seam centerlines don't coincide. This corrects errors caused by heat sinking, a problem when welding parts of different thicknesses.

A third algorithm uses measurement of local seam width to automatically adjust filler-wire feed rate. Finally, standoff distance is automatically controlled by monitoring are voltage and adjusting the robot's vertical position so it is kept at the proper value.

A prototype adaptive-control system, complete with GE P50 robot and two-axis Aronson Rab30 part positioner (the motion of which is fully coordinated with the robot), has been shipped to GE's Evendale, OH, aircraft engine plant. Prior to being placed on the factory floor, it will undergo several months of shakedown tests. In the future, this system will be the focal point of a fully automated arc-welding cell.
COPYRIGHT 1984 Nelson Publishing
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1984 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Publication:Tooling & Production
Date:Sep 1, 1984
Words:732
Previous Article:Robotic welding: big payoff for a small shop.
Next Article:Controlled atmospheres increase furnace-brazing productivity.
Topics:


Related Articles
High-deposition welding with vision.
Police dog Sam gets the bird.
JAILBREAK IRISHMAN ON THE RUN FOR 42 YEARS; Missing John sets a record.
RIDDLE OF LISA AND SCRUFF AT CALL BOX.
Mail On Money; Trust me - the future is golden.
Clothing Series Introduced.
Stock up on cheap shares; SLOW AND STEADY BEATS THE MARKET.
Shari Low : They have to be making this study up.
No-sew sleep sack.
Quanta provides vision for welding.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters