Method to determine weld puddle area and width from vision measurements

• US 4,611,111 A
• Filed: 01/22/1985
• Issued: 09/09/1986
• Est. Priority Date: 01/22/1985
• Status: Expired due to Term

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Corby, Nelson R. Jr., Baheti, Radhakisan S.
• Primary Examiner(s): Shaw, Clifford C.
• Application Number:
  US06/693,545
• Time in Patent Office: 595 Days
• Field of Search: 219/124.34, 219/130.21, 901/42, 901/47, 364/477, 358/101
• US Class Current: 219/124.34
• CPC Class Codes: B23K 9/0956 using sensing means, e.g. optical

Abstract

A computationally efficient technique to determine weld puddle area and maximum width in real time from noisy measurements of the puddle trailing edge. Using the torch electrode as the origin, image intensities are sampled in radial directions, and potential pool edge points are extracted. The boundary data is prefiltered to remove extraneous points, and a least-squares algorithm is used to estimate the area and width of the elliptical puddle. The values are sent to a vision-based arc welding robot control system which regulates these pool parameters to assure full penetration welds. Angular orientation of the weld puddle is estimated in an analogous manner.

Claims

1. In a robotic welding operation wherein welding current conducted by the nonconsumable electrode of a welding torch generates an electric arc whose heat forms a molten weld puddle in the workpiece, and a vision system integrated with said torch provides images of said weld puddle to control the welding process, the improved method of measuring puddle geometry parameters comprising:

sampling image intensities along N radial rays intersecting only the trailing edge of said weld puddle, using the torch electrode as the origin;

determining pudlde boundary point data by processing each radial image intensity waveform to identify patterns that indicate potential edge points;

prefiltering the puddle boundary data to remove extraneous points;

estimating in real time by means of a least-squares algorithm the area and maximum width of the elliptical weld puddle; and

passing these values to a closed-loop weld process control system.

2. The method of claim 1 wherein the boundary data is prefiltered based on a parabola fit to the puddle trailing edge.

3. The method of claim 1 and additionally estimating puddle orientation relative to the direction of torch travel.

4. The method of claim 1 wherein said N radial rays along which each image is sampled cover a total angle of about 90°

to 120°

.

5. The method of claim 4 wherein the puddle boundary data is prefiltered by selecting points inside a parabolic zone approximately fitted to the puddle trailing edge.

6. A method of determining weld puddle area and maximum width in real-time from weld region images provided by a vision system on an arc welding torch which has a nonconsumable electrode, comprising:

acquiring images of the weld puddle;

sampling each image along N radial rays centered on said electrode and intersecting only the trailing edge of said weld puddle;

determining all potential puddle boundary points from intensity variations along each ray;

prefiltering the foregoing boundary data to remove clearly extraneous points;

estimating by means of a least-squares algorithm the unknown parameters in an equation of an ellipse; and

employing said estimated parameters to explicitly calculate puddle area and maximum width, and passing these values to a closed-loop weld process control system.

7. The method of claim 6 wherein said N radial rays along which each image is sampled are equally spaced and cover a total angle of about 90°

to 120°

.

8. The method of claim 7 wherein said ellipse has two degrees of freedom and two parameters are estimated.

9. The method of claim 7 wherein said ellipse has three degrees of freedom and said estimated parameters are used to additionally calculate weld puddle orientation relative to the direction of torch travel.

10. The method of claim 6 wherein the boundary data is prefiltered by selecting points inside a parabolic zone roughly fitted to the puddle trailing edge.