Fuzzy Subtractive Clustering for Polymer Data Mining for SAW Sensor Array Based Electronic Nose

Fuzzy subtractive clustering (FSC) has been applied as data mining tool for making selection of a small set of polymers from a large set of prospective polymers having potential for being chemical interfaces for electronic nose sensor array. The basic idea behind applying FSC selection is to cluster the prospective polymers according to some measure of similarity among them in relation to their interaction with the chemicals targeted for sensing. The polymers defining the cluster centers are taken to make the selection set. The basis for defining similarity among different polymers is the partition coefficients associated with sorption of chemical analytes from vapor phase to polymer phase in thermodynamic equilibrium. The goal for selection is to identify a minimal set of polymers that provide the most diversely interaction possibilities with the target vapors. The proposed selection method has been validated by simulating responses of a polymer-coated surface acoustic wave (SAW) sensor array for detection of freshness and spoilage of milk and fish food products. The end use of the proposed selection method is suggested for developing low-cost high-performance sensor array based electronic noses for commercial and consumer applications.