A fuzzy logic approach for sensor validation in real time expert systems

Abstract

Data validation of the information coming from sensors is a fundamental problem when developing on-line real-time expert systems. In this paper an approach using fuzzy logic methods is presented. This approach has been originated during the development of the MIP project, a real-time expert system for assistance to petrochemical processes that is described in [1]. The MIP expert system is deployed and in production in a petrochemical plant of INH-REPSOL in Tarragona, Spain. The method presented in this paper complements the MIP expert system by preprocessing the incoming data and assigning confidence values to each measurement from the sensors. These confidence values can be used subsequently to enrich the conclusions obtained by the MIP system. The method is currently in field test in the plant.