Development of an Expert System for the Electronics Manufacturing Sector

The knowledge repository proposed in this paper uses a neural network to identify patterns of words from a vocabulary of key words used in a specific manufacturing environment to describe rejects commonly encountered during the assembly process. The problem of indexing through the most appropriate solutions of a case-based reasoning system can be addressed in a more efficient way through the use of an artificial neural network (ANN) front end to analyze and identify patterns of words which lead to a certain goal or solution. Using this method, keyword matches are unnecessary as the link between problems and solutions are in reality, trained patterns of words which do not necessarily contain the terms that describe the problem explicitly. The method is used in conjunction with a self organizing map (SOM) in the back end of the system to organize data into groups of solutions which are related to one another through the problems that they address.