Industrial solution for arc welding control system.

Bucur, Gabriela ; Moise, Adrian

1. INTRODUCTION

The final purpose of a welding operation is to achieve a welding seam that satisfies many imposed conditions. These conditions results from technological general system analysis, system witch is composed by power source -- welding source - arc welding system, welding seam, manipulation system and control system.

This paper present an industrial solution to process the welding-arc voltage with neural network. That result can be applied in automated robotic welding based on the correction of the trajectory welding torch through the welding seam.

2. INDUSTRIAL CONTROL SYSTEM

Using the anterior researche, we have proposed a WIG welding control system with neural network controller (Haykin, 1999). For welding head position control we use an acquisition system for arc voltage values from welding process. So, the arc voltage is a measure of welding head position. This signal will be the input for a neural network structure especially made by the authors for modifies the position of welding head.

This equipment may achive real time control for algorithms of welding seam sensor, alghorithm based on artificial neural networks -- two interconnected adaptive linear filters: one filter work on interferences compensation principle and that filter extract the useful signal from welding process signal and the second one eliminate the error between reference signal and first filter output signal (Bucur et al.,2008).

Industrial welding head control system, presented in figure 1, contain:

* Welding source.

* Filter for processing of acquisitioned signal who realize low-pass filter of welding electric arc voltage.

* Process-computer interface AX 5411.

* Signal amplifier A.

* DC engine (MCC) used for oscillation movement of welding torch and welding head position correction.

* Reference signal generator.

* Sycronization block (BS) for syncronize the reference signal with the process signal obtaining the moment for processing data start.

* Personal computer PC, who realise the acquisition of voltage values, thru AX 5411 interface, processing of them and generate the command signal for DC engine, thru the same interface.

[FIGURE 1 OMITTED]

The welding source is also controlled by the PC, thru serial port RS 232.

The software is composed by A/N conversion for acquisitioned information, numerical commands generation and N/A conversion programs and also a graphical interface with AX 5411 and welding source (Bucur et al.,2002).

This paper considere only position correction and osccilation welding torch thru the welding seam, in the median plan of seam, without welding torch advance speed control.

3. PROBLEMS DURING WELDING PROCESS

During processing data with neural network, it's possible to appear some desynchronisations between reference signal and process signal who determine a low quality of welding joint ("slapping" phenomena)(Miclosi et al.,1984). So, it's necessary to start processing data in the moment of coincidence of two signals.

Refering to that, we propose a sinchronisation signals structure shown in figure 2. This sinchronisation block (BS) is a comparing element (CSC) between the arc voltage values and reference signal, analytical compute. After comparison, in the moment of coincidence, that generates a signal for start processing data (Dumitrescu & Bucur, 2000).

After this synchronisation, we can demonstrate the neural network controller realize the welding torch position correction, in horizontal plan, irrespective of these frequencys and whatever is the moment of these coincidence.

[FIGURE 2 OMITTED]

In figure 3 are shown two aleators signals with different frequency.

Neural network simulation was realized with the MATLAB program (Kharab & Guenther, 2002). The program is:

time=0:0.05:10;

x= sin([pi]t);

plot(time,x) ; (figure 3 -- blue courve)

p= sin(0.5[pi]t);

plot(time,p); (figure 3 -- green courve)

t=x+p';

[w,b]=initlin(p',t);

[a,e]=adaptwh(w,b,p',t,0.00001);

[w,b]=initlin(e,x);

[a,e]=adaptwh(w,b,e,x,0.01);

plot(time,a,time,x) ;

plot(time,e) ; (figure 4)

In this program x is the reference signal, p is the process signal, w and b is adjusted weights, a is the output signal of neural controller and e is the error signal. Simulation time is 10 seconds.

We can observe that signals are synchronised and our neural network can realize the frequency correction.

The error signal obtained during synchronisation process is presented in figure 4. This error is only an amplitude error, not a frequency error. So, we can say, our neural controller generates an output signal wich is correction signal for welding head position and also for osscilation movement of this tool.

These results are available when the signals are different frequency and the coincidence between them is on ascendant steepness of courves. We made the others simulations for signals with different frequency and the coincidence on opposite steepness of courves.

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

The result was the same: position and frequency correction, with similar error values.

4. CONCLUSION

In this paper was presented a technical solution for WIG welding control system.

This industrial system considere only position correction and osccilation welding torch thru the welding seam, in the median plan of seam, without welding torch advance speed control.

During the WIG welding process, if the oscillation frequency is modified over the reference frequency applied to the neural network, we just demonstrate that the designed network can make also the correction of the oscillation frequency of welding torch.

All this considerations was made for [delta]=0, where 5 is welding torch declination from medium plane of seam. Will be very interesting to considere [delta][not equal to]0, then to determine the function between [delta] and length of arc welding and study the monotony of this function. We can verify if our neural network can realize the designed corrections.

In the future, the solution of neural network control of trajectory welding torch to the welding seam is very interesting, especially for MIG-MAG welding process, because will appear the burned drop transfer phenomena thru the welding arc.

5. REFERENCES

Bucur, G.; Popescu, C. & Popescu, C. (2008). Neural Network Control for Wig Welding Processes, Proceedings of the 19th International DAAAM Symposium, Katalinic, B. (Ed.), pp. 84-85, ISBN 978-3-901509-68-1, Austria, October 2008, Published by DAAAM International, Vienna

Bucur, G., Dumitrescu, St. & Miclosi, V. (2002). Neural Network Control in Robotic Welding Processes. Journal of Symposium "35 de ani de activitate a Universitatii Petrol-Gaze la Ploiesti", Vol.LIV, No.2/2002, pp. 38-43, ISSN 1221/9371

Dumitrescu, St. & Bucur (Chiriac), G. (2000). Measurement Technique, Petroleum-Gas University Publisher House, ISBN 973-99506-5-5, 973-99506-7-1, Ploiesti, Romania

Haykin, S. (1999). Neural Networks, Prentice Hall, ISBN 0-13- 273350-1, U.S.A.

Kharab, A. & Guenther, R. (2002). An Introduction to Numerical Methods in MATLAB, Chapman & Hall/CRC, ISBN 1-58488-281-6, U.S.A.

Miclosi, V., Scorobetiu, L., Jora, M. & Milos, L. (1984). Welding Processes Fundaments, Didactica si Pedagogica Publishing House, Bucharest, Romania