Probabilistic inference in a knowledge-based quality information system for consumer electronics production

Abstract

A quality information system (QIS) installed in a West German plant for consumer electronics production that features a number of conventional data processing capabilities and an expert system for both data interpretation and fault diagnosis is discussed. The diagnostic portion of the QIS expert system is discussed, and a detailed description of the inference algorithm, which mainly uses the approaches of semantic nets and probabilistic inference is given. The experience gained during knowledge acquisition, system realization, and system application in the industrial environment is discussed.<>