An Expert System With External Optimization Module for Quality Control Decisions

Although statistical quality control is already largely automated for the gathering and processing of data, improperly implemented quality control can yield erroneous results, which may go unnoticed indefinitely in a manufacturing environment. Quality control, which combines decisions and interpretations based on analytical data, on heuristics, and on user judgment, is suited for rule based expert systems. Many commercial software packages store data and perform calculations for quality control; this comprehensive expert system, named Quincy, is envisioned as a complement to those. Consisting of 118 rules, 246 facts, 16 optional explanations, 34 textual expansions, 36 displays and an optional optimization module, Quincy focuses on three main areas of quality control decision making: A. Selection of applicable control charts for various manufacturing processes. B. Interpretation and generic diagnoses of control chart results. C. Selection of appropriate acceptance sampling plans. Quincy is designed to be applicable to a wide number of manufacturing organizations, even those with limited computing experience or availability. The expert system shell has a user friendly interface, an undemanding PC hardware platform, and is quite portable. The C optimization module operates transparently to the user, and is also portable. Furthermore, Quincy is built modularly so that future additions and modifications can be made relatively simply. 1.0 INTRODUCTION Artificial intelligence includes several fields of scientific research including natural language processing, artificial neural networks and expert systems. All are concerned with replicating and sometimes improving upon human abilities. For expert systems, the human traits of knowledge