| The expert fuzzy advisor (EFA) is an expert system for the statistical control of manufacturing processes. The EFA assists the operator during process capability studies and on-going process control. The EFA facilitates the solution of traditional problems found in the Statistical Process Control (SPC) functions of monitoring, analysis, diagnosis, and supervision.;In monitoring, the EFA aids in data collection, and plotting of sample statistics in Shewhart and EWMA control charts. During analysis and diagnosis, the EFA uses a fuzzy approach to model the information used in decision making.;In analysis, the EFA monitors process instability, and assists the operator in the interpretation of sample statistic patterns. The EFA detects process instability by classical statistical runs tests and the EWMA statistic. Each test is a fuzzy set with a grade of membership representing process instability. A fuzzy union operation provides a single composite instability index. Trend indicators formed by consecutive composite instability indices and the ability of the EWMA statistic to forecast, provide the operator an early start in the investigation of special causes of error. If process instability is detected, the operator determines the type of sample statistic pattern by starting an interactive analysis of pattern characteristics with the EFA. This interactive analysis is simplified by the EFA's interpretation of the statistical tests output.;In diagnosis, the EFA models relationships between problem symptoms and causes using fuzzy relational matrices. A diagnosis process uses fuzzy relational products between observed symptoms and the diagnosis knowledge-base to infer the possible causes of error.;In supervision, the EFA integrates operator, process, and SPC knowledge in interactive decision making. The operator makes decisions and continues to develop knowledge boosting the opportunities of success of the SPC implementation.;A system prototype has been developed to demonstrate the concept of the EFA. |
