Trend-weighted rule-based expert system for process diagnosis

Abstract

This paper presents and innovative technique-referred to as trend-weighted rule-based expert system (TWRBES) - grounded in the integration of two existing tools of the artificial intelligence field, expert systems (ES) and qualitative trend analysis (QTA). The main goal of this approach is to benefit of the main advantages associated with each of the techniques used, such as the ability to represent knowledge through rules and the ability to extract the behavior and the trends of a continuous signal. Such integration allows a direct purpose in industrial environment applications, especially in the intelligent automation field. This paper introduces this technique and preliminary results obtained from applying it to industrial process diagnosis.