Comparison of Particle Swarm Optimization and Genetic Algorithm for Molten Pool Detection in Fixed Aluminum Pipe Welding

This paper proposes a study on the comparison of particle swarm optimization with genetic algorithm for molten pool detection in fixed aluminum pipe welding. The research was conducted for welding of aluminum alloy Al6063S-T6 with a controlled welding speed and a Charge-couple Device (CCD) camera as vision sensor. Omnivision-based monitoring using a hyperboloidal mirror was used to detect the molten pool. In this paper, we propose an optimized brightness range for detecting the molten pool edge using particle swarm optimization and compare the results to genetic algorithm. The values of the brightness range were applied to the real time control system using fuzzy inference system. Both optimization methods showed good results on the edge detection of the molten pool. The results of experiments with control show the effectiveness of the image processing algorithm and control process. 

Fixed aluminum pipe welding; Fuzzy inference system; Genetic algorithm; Molten pool detection; particle swarm optimization

The conclusions of this paper are summarized as follows:

? This research proposes molten pool detection of fixed aluminum pipe welding using PSO as compared to using GA. The brightness range for edge detection was constructed using the percentage of outer brightness (pout) and inner brightness (pin).

? The experimental results show that the proposed method can detect the edge of the molten pool with minimum error, although PSO can reach minimum error faster than GA. The method can perform the optimization of brightness range, reducing the computational cost and time consumption.

? PSO optimized image processing algorithm was applied into the real time process using omnidirectional vision-based monitoring of the molten pool. From the experimental results using fuzzy inference system, image processing algorithm and the described control system are effective. 

Ahn, C.W., 2006. Advances in Evolutionary Algorithms: Theory, Design and Practice, Springer-Verlag Heidelberg, New York, USA.

Baskoro A. S., Kabutomori M., Suga Y., 2008. Automatic Welding System of Aluminum Pipe by Monitoring Backside Image of Molten Pool Using Vision Sensor, Journal of Solid Mechanics and Materials Engineering, JSME, Volume 2, Number 5, pp. 582-592.

Baskoro A. S., Masuda R., Kabutomori M., Suga Y., 2009. Welding Penetration Control for Aluminum Pipe Welding Using Omnidirectional Vision-based Monitoring of Molten Pool, Quarterly Journal of the Japan Welding Society, Volume 27, Number 2, pp.17-21

Baskoro et al. 83 Baskoro, A.S., Masuda, R., Kabutomori, M., Suga, Y., 2009. An Application of Genetic Algorithm for Edge Detection of Molten Pool in Fixed Pipe Welding, International Journal of Advanced Manufacturing Technology, Volume 45, pp. 1104-1112.

Haupt R. L. & Haupt S.E., 1998. Practical Genetic Algorithms, 2nd Edition, A John Wiley & Sons Inc. publication, New Jersey, USA. Kennedy J. & Eberhart R., 1995. Particle Swarm Optimization, Proceedings of IEEE International Conference on Neural Networks, p. 1942-1948.

Kim J., Muramatsu M., Murata Y., Suga Y., 2007. Omnidirectional Vision-Based Ego-Pose Estimation for an Autonomous In-pipe Mobile Robot, Advanced Robotics, Volume 21, Number 3-4, pp.441-460.