Out-of-control pattern recognition and analysis for quality control charts using LISP-based systems

Abstract Artificial intelligence has a wide variety of applications. The expert system area of artificial intelligence is expanding, as evidenced by the large number of new systems and the proliferation of expert system building tools or shells. This paper discusses the development and use of an expert system to detect and analyze various patterns of variation that can occur in manufacturing quality control charts. The expert system looks for the following six potential patterns of variation: trend, cycle, mixture, shift, stratification and systematic. Statistical significance tests as interpretive rules are used to determine the pattern of variation. Once the pattern is identified, the expert system supplies the user with possible causes for the out-of-control condition. The magnitude of the out-of-control condition and where it starts and stops are also provided.