CAPACITANCE-BASED TOMOGRAPHY FLOW PATTERN CLASSIFICATION USING INTELLIGENT CLASSIFIERS WITH VOTING TECHNIQUE

This paper presents a method for Electrical Capacitance Tomography (ECT) flow classification using voting technique, employing Multilayer Perceptrons (MLPs) as the intelligent pattern classifiers. MLP classifiers were trained with a set of simulated ECT data associated to various flow patterns and was tested with untrained data to verify their performances. MLP classifiers which gave high percentage of correct classification were integrated into a voting system and tested over a distinct set of ECT data. The performances of the individually selected classifiers were compared with the voting system. The results showed superiority of the voting system over individual classifiers.