In the novel Double Electrode Gas Metal Arc Welding (DE-GMAW), the transfer of the liquid metal from the wire to the work-piece determines the weld quality and for applications where the precision is critical, the metal transfer process needs to be monitored and controlled to control the diameter, trajectory, and transfer rate of the droplet of liquid metal. In this doctoral research work, the traditional methods of tracking, Correlation, Least Square Matching (LSM) and Kalman Filtering (KF), are tried first. All of them failed due to the poor quality of the metal transfer image and the variety of the droplet. Then several novel image processing algorithms, Brightness Based Separation Algorithm (BBSA), Brightness and Subtraction Based Separation Algorithm (BSBSA) and Brightness Based Selection and Edge Detection Based Enhancement Separation Algorithm (BBSEDBESA), are proposed to compute the size and locate the position of the droplet. Experimental results verified that the proposed algorithms can automatically locate the droplets and compute the droplet size with an adequate accuracy. Since the final objective is to automatically process the metal transfer in real time, a real time processing system is implemented and the details are described. In traditional Gas Metal Arc Welding (GMAW), the famous laser back-lighting technique has been widely used to image the metal transfer process. Due to laser imaging systems complexity, it is too inconvenient for practical applications. In this doctoral research work, a simplified laser imaging system is proposed and two effective image algorithms, Probability Based Double Thresholds Separation Algorithm and Edge Based Separation Algorithm, are proposed to process the corresponding captured metal transfer images. Experimental results verified that the proposed simplified laser back-light imaging system and image processing algorithms can be used for real time processing of metal transfer images.
| Identifer | oai:union.ndltd.org:uky.edu/oai:uknowledge.uky.edu:gradschool_diss-1533 |
| Date | 01 January 2007 |
| Creators | Wang, Zhenzhou |
| Publisher | UKnowledge |
| Source Sets | University of Kentucky |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | University of Kentucky Doctoral Dissertations |
Page generated in 0.0021 seconds