There are many problems in welding metallurgy for which it is difficult to develop a first principles scientific model due to their complexity. A significant problem faced by today's welding engineers, is the need to relate welding parameters to the quality of the finished weld. This is usually done by experience, and the need for many experimental trials, eventually leading to optimal welding parameters. Important characteristics in the evaluation of line-pipe seam weld quality are the weld bead shape and size, which can have a significant effect on the microstructure and mechanical properties of the weldment through heat flow effects. Properties of the final weld may therefore be difficult to predict, especially quantities such as weld metal toughness, which are known to be dependent on many factors. One approach to such complex problems is to use neural networks. A neural network is an artificial simulation of the brain which models data through a learning process and stores the information as a set of rules akin to knowledge. This research is concerned with the application of neural network techniques to the prediction of the mechanical and physical properties, including the shape of the weld bead, of submerged arc line-pipe steel welds. A limited experimental investigation has been carried out using optical and transmission electron microscopy to establish an understanding of the complex microstructures that result from the welding processes used in the production of line-pipe.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:420738 |
| Date | January 2002 |
| Creators | Ridings, Gareth E. |
| Publisher | Loughborough University |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | https://dspace.lboro.ac.uk/2134/34693 |
Page generated in 0.0113 seconds