Automated optical inspection system for professional double face printed circuit boards

In the present article, we show the results of the research on knowledge engineering for the development of an automated optical inspection system for printed circuit boards. An innovative methodology for the modelling of knowledge and construction of knowledge based systems was used as the basis for the development: KADS. This methodology relies in the stratification of knowledge in well defined levels of abstraction and in the construction of models for each knowledge level. The specification and a project of a knowledge based system for diagnoses of double face professional printed circuit boards based on the representation of physical structures by segment graphs and characteristic points map is presented. The faults in the boards are identified through discrepancies between the characteristic points map generated from the printed circuit project specification and the map generated from images generated by video cameras. The knowledge based system is used to identify if the discrepancies are due to differences introduced by noise in the image acquisition process, or otherwise due to faults in the board from which the image was generated. If the system identifies the discrepancy as a fault, them the nature of the fault is derived by the knowledge based system from a set of possible faults. A diagnostic is passed to the operator of the system and a set of statistics is generated. The system accepts additional knowledge supplied by the operator and, thus, is able to increase dynamically its knowledge base