Wilson, Duncan John;
(1998)
Classification of defects using uncertainty techniques in industrial web inspection.
Doctoral thesis (Ph.D.), University College London.
Text
Classification_of_defects_usin.pdf Download (16MB) |
Abstract
This research investigated how Artificial Intelligence techniques could be applied to industrial web inspection to improve the process of classifying defects. It focuses on applying uncertainty techniques to aid the construction of a classification scheme. In many industrial process control situations, the need to identify and classify defects is key to enabling process improvements. Inspection is used for this task. Constructing a classification scheme to correctly identify defects in a product is difficult. Many people assume that a description of what is to be identified exists. That is not necessarily true. The domain knowledge is vague and nearly always incomplete, and the constraints on inspection equipment means the data output often produces ambiguous results. A method was required which would allow such uncertainties to be incorporated into an automated visual inspection system. Fuzzy Set Theory and Dempster Shafer were used to aid the development of the classification scheme. Fuzzy Set Theory provided an intuitive numerical method for capturing knowledge about defects (for example, concepts such as 'defect has small area') and Dempster Shafer provided a framework for processing such heuristic knowledge so that a classification could be made. The result was a set of rules where each rule had a fuzzy antecedent and a consequent describing a set of possible defect types. Dempster Shafer was used to combine the consequants of the rules to calculate the most likely outcome. For comparison with the output generated using Dempster Shafer, decision criteria proposed by Smets and Wesley were implemented. The accuracy of the classification scheme and an analysis against five desirable criteria for an inspection system are the focus of the discussion. An experimental rig was constructed using a linescan CCD camera to simulate a production line. Real samples of defects in plastic film were used to test the proposed classification scheme.
Type: | Thesis (Doctoral) |
---|---|
Qualification: | Ph.D. |
Title: | Classification of defects using uncertainty techniques in industrial web inspection. |
Open access status: | An open access version is available from UCL Discovery |
Language: | English |
Additional information: | Thesis digitised by ProQuest. |
URI: | https://discovery.ucl.ac.uk/id/eprint/10104034 |
Archive Staff Only
View Item |