Arc welding adaptive process control system

Claim 2

A vision-based robotic welding system comprising: an arc welding torch having an electrode to conduct welding current and create a molten weld puddle in a workpiece by the heat of an electric arc, and integrated optics to provide an image of the weld puddle and workpiece joint; and an improved adaptive feedback control system to control weld quality during the welding operation and assure full penetration weld beads, which is comprised of: means to analyze said image and measure puddle area and maximum width at discrete intervals; means to determine a puddle geometry error by comparing the area and width measurements with desired puddle area and width; means to correct the nominal welding current in dependence upon said puddle geometry error; and means for feeding corrected welding current to said welding torch to thereby change the heat input and weld puddle boundaries.

Claim 3

The welding system of claim 1 wherein said current correction means includes a dynamic compensator, to minimize the effects of process disturbances on puddle geometry and to stabilize the closed-loop system, to which said puddle error is input.

Claim 4

The welding system of claim 2 and a limiter to clamp said corrected welding current between upper and lower limits.

Claim 5

The welding system of claim 1 wherein said puddle geometry error determining means comprises deriving the difference between desired and measured puddle area and between desired and measured puddle width, multiplying each difference by a constant selected based on the control objective, and summing the results.

Claim 6

The welding system of claim 1 and arc length control means to maintain a constant arc length by varying arc voltage as welding current changes.

Claim 7

A vision-based robotic welding system comprising: a tungsten-inert gas welding torch which strikes an electric arc between its electrode and a workpiece and supplies heat to create a molten weld puddle, said torch having an integral optical system to image the weld puddle and workpiece joint, and a filler wire feed mechanism; a multiaxis robot supporting said torch; an improved adaptive feedback control system based on weld puddle area and width regulation to control the welding process in real time and assure full penetration welds, while compensating for process disturbances; said adaptive control system including means to analyze said image and measure weld puddle area and maximum width at discrete time intervals; means to determine a puddle geometry error from the difference between desired and measured puddle area and between desired and measured puddle maximum width, and preassigned weighting constants; means for inputting said puddle geometry error to a dynamic compensator and determining a correction to the preset nominal welding current to change the weld puddle boundaries; and arc length control means to maintain a constant arc length by modulating arc voltage to reflect changes in welding current applied to said welding torch.

Claim 8

A vision-based robotic welding system comprising: a tungsten-inert gas welding torch which strikes an electric arc between its electrode and a workpiece and supplies heat to create a molten weld puddle, said torch having an integral optical system to image the weld puddle and workpiece joint, and a filler wire feed mechanism; a multiaxis robot supporting said torch; an improved adaptive feedback control system based on weld puddle area and width regulation to control the welding process in real time and assure full penetration welds, while compensating for process disturbances; said adaptive control system including means to analyze said image and measure weld puddle area and maximum width at discrete time intervals; means to determine a puddle geometry error from the difference between desired and measured puddle area and between desired and measured puddle maximum width, and preassigned weighting constants; means for inputting said puddle geometry error to a dynamic compensator and determining a correction to the preset nominal welding current to change the weld puddle boundaries; and arc length control means to maintain a constant arc length by modulating arc voltage to reflect changes in welding current applied to said welding torch.

Claim 9

The welding system of claim 6 wherein said arc welding torch has a second integral optical system to project a structured light pattern onto said workpiece joint close to said weld puddle which is imaged and analyzed to determine joint location and width.

Claim 10

The welding system of claim 7 wherein said process disturbances compensated by said adaptive feedback control system include changes in filler wire feed rate, robot velocity changes, variations in joint width and tack welds, and heat sinks.

Claim 11

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 a workpiece, and a vision system integrated with said torch provides an image of said weld puddle and the workpiece joint to guide the torch and control the welding process, the improved method of closed-loop weld process control comprising the steps of: analyzing said image at discrete time intervals and measuring weld puddle area; determining a puddle geometry error from the difference between desired and measured puddle areas; inputting said puddle geometry error to a dynamic compensator whose output determines a correction to nominal welding current; and feeding said corrected welding current to said torch to change the weld puddle boundaries and regulate puddle area to assure full penetration weld beads.

Claim 12

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 a workpiece, and a vision system integrated with said torch provides an image of said weld puddle and the workpiece joint to guide the torch and control the welding process, the improved method of closed-loop weld process control comprising the steps of: analyzing said image at discrete time intervals and measuring weld puddle area; determining a puddle geometry error from the difference between desired and measured puddle areas; inputting said puddle geometry error to a dynamic compensator whose output determines a correction to nominal welding current; and feeding said corrected welding current to said torch to change the weld puddle boundaries and regulate puddle area to assure full penetration weld beads.

Claim 13

The method of claim 9 and the steps of measuring weld puddle maximum width at said discrete time intervals, determining said puddle geometry error as a predetermined proportion of the sum of the difference between desired and measured puddle area and the difference between desired and measured puddle width, and regulating a combination of puddle area and width to assure full penetration welds.

Claim 14

The method of claim 10 and maintaining a constant length arc by modulating arc voltage to reflect changes in said welding current.

Claim 15

The method of claim 10 wherein the closed-loop control automatically compensates for welding process disturbances such as variations in joint width, robot velocity changes, tack welds, and changes in filler wire feed rate.