A MULTISENSOR SYSTEM FOR BEER FLAVOR MONITORING USING AN ARRAY OF CONDUCTING POLYMERS AND PREDICTIVE CLASSIFIERS

This paper describes the results of a three year project, jointly fund ed by the UK government and industry, to develop a multisensor system capable of discriminating between the aromas of different beers. The s ystem consists of an array of up to 24 conducting polymer sensors (thi n films electrodeposited onto a microelectrode structure). The conduct ing polymers provide the active layer in these conductometric odour se nsors and respond differentially to the headspaces of beers and lagers . The interface circuitry and signal conditioning have been designed a nd realized in custom PCBs housed in a Eurorack-based multisensor syst em. A comprehensive suite of software modules has been developed to au tomate the sampling system and process the sensor array data. The outp ut from the polymer array is pre-processed using a variety of algorith ms (e.g., fractional change in conductance, normalized relative respon se) and then classified using a statistical (chemometric fingerprintin g technique) or neural predictive classifier (multi-layer perceptron u sing back-propagation learning). The odour-sensing system can distingu ish subtle taints, e.g., 0.5 ppm of diacetyl in an ethanol solution wi th only nine different varieties of conducting polymers.