Laser welding automation.

Popa, Marcel ; Precup, Mircea ; Contiu, Glad 等

1. INTRODUCTION

The goal of the team was the automation of the HL54P laser device for the overlap welding, in the case of non-CNC (Computer Numerical Control) machine or in the case of missing CAD (Computer Aided Design) file and CAM (Computer Aided Manufacturing) software for CNC programming. In our case the laser device had the laser beam generator, the processing optics and optical cable. The optical cable guides the laser light from the laser device to the processing optics which focuses the laser light onto the surface of the work piece.

The HL54P was obtained after an international collaboration program with University of Stuttgart. It is a solid-state laser with an Nd: YAG rod (Neodymium-doped Yttrium Aluminium Granat).

The advantages of laser beam machining compared to other methods are: greater precision, increasing work speed, allowing the point-like processing of the piece - this is a process that does not imply touching the piece with an instrument (beam) (Hugel, 1992).

The use of this procedure reduces the heating of the working area because the energy is introduced faster and on a small surface. This way, large power densities can be reached. Other advantages are flexibility and accessibility (Popa, 2005).

2. GENERAL ASPECTS

Nd:YAG lasers emit light in the near infrared range, at a wavelength of 1.06[micro]m. This means that the light emitted by Nd:YAG lasers is almost in the visible range (Koechner, 2006).

[FIGURE 1 OMITTED]

The laser light of an Nd: YAG laser can be routed through glass optics and optical fibbers.

The positioning system has the following configuration: two axes for moving the working table in xOy plane and one axis for moving the processing optics Oz axis in order to focus the laser beam on the work piece.

The precision of the positioning system depends on the accuracy of the axis. In our case we use DGE -25-200-SP for the Oz axis and two DMES-25-200 for the positioning in the xOy plane.

The repeatability of the focusing axis, Oz, is 0.02 mm and for the other axes, Ox and Oy is 0.05 mm. The stepping motor, MTR-ST-57-48S, has the full step angle of 1.8[degrees] with a 5% maximum error. It is possible to supply two coils simultaneously with different current share. The result is a 1/2, 1/4, 1/5, 1/8, 1/10, 1/32 of a step, depending on the current share. These allow considerable refinement in the maximum resolution of positions to which a stepping motor can run.

The smallest incremental path (resolution) on a positioning axis is determined by the motor's step angle (number of steps per revolution) and the feed constant of the positioning axis (determined by the diameter of the input pinion or the slope of the spindle).

This can be calculated as follow:

[n.sub.x] = 360[degrees]/[[alpha].sub.s] (1)

[d.sub.r] = [d.sub.f] x i/[n.sub.s] (2)

In the equations above [n.sub.s] represents number of steps per revolution, cts motor angle, [d.sub.r] resolution, [d.sub.f] feed constant of the axis, i gear multiplication.

For the axis of the work table, in case of the full step, the resolution is 0.012 mm per step and for the focusing axis is a 0.05 mm resolution per step. From this point the position of the work piece can be calculated. Some error can occur if the stepping motors lose steps. This happens only when the load exceeds the maxim load on the axis.

3. RESEARCH COURSE

The connection between positioning system and laser device was made by programming an Atmel microcontroller (Atmel ATMEGA8535). For machining was necessary to control the laser device at the same time with the positioning system. With the help of the PC interface made in Delphi software was possible to synchronize the laser device and the positioning system for machining. The laser pulses were generated when the work piece was in the processing position.

For path identification we used the monitoring system of laser optics and an adapted web camera. The web camera had a manual focus which allows calibrating the focus distance to 182mm. After this calibration the working distance can be adjusted by moving the optics on the Oz axis.

[FIGURE 2 OMITTED]

For material processing the laser beam has to be focused to the necessary power density. A convex lens is used for this.

A collimation unit is used to focus the laser beam onto the work piece. The collimation lens converts this beam into a parallel beam. A mirror is inserted into the beam guidance to monitor and adjust the work piece. The mirror allows the transmission light having the wavelength of the laser beam. It reflects visible light which comes from the work piece through the lens into the beam guidance.

Most of the existing path following solutions performs complex analysis on the images of the path using curve matching or Kalman filters (Ma et al., 1999). These methods produce accurate information about the curvature of the path and allow a positioning system, typically moving at slow speed, to accurately drive the path. However these methods require a relatively large amount of processor time and often require considerable modification if the position of the camera is altered.

The colour image proved too difficult to work with. Fortunately with the PC interface provides an efficient routine to convert the 24-bit colour image to a 8-bit grey scale. A threshold, chosen to select the maximum amount of track and a minimum amount of the surroundings, is then applied to the grey scale image to produce a black and white image that is further processed. We extract two pieces of information from the image: the position of the path, offset from the centre of the image, and the gradient of the path.

Rather than processing the entire region, which would require too much computing time, only the outer edges of the processing region are checked to locate the path.

These edges are offset from the edges of the image due to the presence of a one pixel black border and the presence of noise caused by synchronisation problems if we use the 30 frames per second camera driver.

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

The point P1 is tested at first, across the lower edge, then up the left edge and finally up the right. If P1 does not exist the track will be assumed to be absent. Point P2 is located by checking the top edge, then the left and right edges. If P1 is found on one of the side edges, that edge will not be checked for P2. The positions of P1 and P2 are computed by taking the average position of the white pixels on the edge. One result of this is that if two parallel lines are present P1 and P2 will be between the lines, which would allow the positioning system to perform simple autonomous road driving if the field of view from the camera is wide enough. The line between P1 and P2 gives the trend of the path in the region of interest.

The offset of the path is computed by averaging the differences between the horizontal positions of P1 and P2 and normalising the result so that it is positioned between -1, for the left-hand side of the image, and 1, for the right hand side.

The gradient is computed and then scaled so that a gradient of zero indicates a vertical line (in the image), a negative gradient indicating that the line slopes to the left and a positive gradient indicating a slope to the right. A horizontal line is represented by a gradient of 1 (the maximum positive gradient) as there is no way of knowing which direction it is leaning in.

With the processing path obtained we can start the welding process by setting the laser parameters. The laser parameters are specific for different types of material, in our case metal sheets.

4. CONCLUSION

With this method the team proposes the optimisation of the welding process by decreasing the process time. This can be obtained by automatically identifying the working zone, with a minimal intervention from the operator. Certain issues need further research: false edge, discontinued edge, the overlap of the paths for the proximate edges, start and end points. [paragraph]

5. REFERENCES [paragraph]

Hugel, H. (1992). Strahlwerkzeug Laser (Laser Beam Tool) Teubner Verlag, ISBN 3-519-06134-1, Stuttgart

Koechner, W. (2006), Solid-State Laser Engineering, Springer Science and Business Media, Inc., e-ISBN: 0-387-29338-8

Ma, Y., Kosecka, J. and Sastry, S. (1999). Vision Guided Navigation for a Nonholonomic Mobile Robot, IEEE Transactions on Robotics and Automation, vol. 15, no. 3, (june 1999) page numbers (521-536), ISSN: 1042-296X

Popa, M. S. (2005). Tehnologii si masini neconventionale, pentru mecanicd find si mecatronicd (Unconventional Technologies and machines for Fine Mechanics and Mechatronics), U.T. Pres, ISBN 973-662-148-0, Cluj-Napoca, Romania

Trumpf Laser (2005). LCB Laser Devices. Basic Training, Trumpf Laser GmbH+Co.KG, Germany