Development of an intelligent system for real-time diagnosis of manufacturing systems

Abstract

Automated and efficient system diagnosis is important for a manufacturing system to achieve high yield and good product quality. After reviewing and analyzing various techniques used for manufacturing diagnosis, this work presents a hybrid diagnostic approach based on fuzzy set and graphical theory. In this approach, triangular fuzzy numbers (membership functions)were incorporated into a directed graph, thus forming a fuzzy directed graph with its nodes representing the system components. A #45;first search strategy for the identification of possible fault propagation paths was established. Using the membership functions attached, the real-time condition of a node can be easily assessed. A prototype FDG-based diagnostic system which utilizes the bespoke approach was developed. This approach also provides an avenue for the worst-first search strategy developed in this work to interface with system modeling tools. Further work was carried out to integrate the prototype system with DESIGN IDEF™, a commercial software for systems design.