Fuzzy rules for automated sensor self-validation and confidence measure

Abstract

In this research we present a methodology for the development of a generic, automated self-validation technique that can be used to improve the operation of a controller based system. The reliability of a controller-based system depends on the validity of the data provided for the development and operation of the controller. The self-validation algorithm described in this paper is based on fuzzy logic rules described by membership functions. The membership functions are created from data set parameters (e.g., the standard deviation and the range of the data set). Raw data is median filtered and then passed through these membership functions to obtain a measure of confidence. The methodology is illustrated using temperature data from an iron-melting cupola furnace. The confidence measured is used in two subsequent companion papers to (1) replace low-confidence data and (2) fuse similar sensor data.