1. Introduction
Sensors based on electrochemical techniques are used to determine the concentration of specific chemical compounds, or theaccurate measurement of physiochemical parameters. But generally they have an important drawback; namely that of susceptibilityto interference from other species that mask the species of interest.However this drawback can be converted to advantage if, instead of looking for that type of accurate measurement, another kind 7 measurement of a rather more qualitative nature is employed, such as the discrimination or classification of samples of complex chemical nature. Under this concept, electronic tongue systems that employ different sets of non-specific electrodes were developed some years ago. Each of the electrodes provides a signal that is proportional to the set of species in the system under analysis. As electronic tongue systems tend to produce a qualitative result, multivariate analysis techniques are generally required in order to process the data obtained from the measurements.
2. System description
2.1. Samples
A total of five Spanish natural mineral waters of different brands(Bezoya, Bronchales, Cortes, Lanjarón and Solán), one sparkling water (Primavera) and tap water from Valencia City have been selected as representative samples and they have been studied by using the array of electrodes described below.
2.2. Electrodes
A wide range of electrodes with different surfaces were selected in order to explore their differential response in potentiometric measurements. Following this approach various electrodes fabricated using thick-film technology were prepared. To this purpose,several inks with different active element were used; the pasteswere supplied by HERAEUS and they are RuO2 of 10 /sq (modelR8911) and 1 M/sq (model R8961), Cu (model C7257), Ag (model C8829), and Pt (model C1076D). The AgCl was manufactured by mixing Ag and AgCl powder in a ratio of 1:1 and using low temperature EG2020 glass (supplied by Ferro). The protective upper layer paste was model D2020823D2 supplied by GWENT.
2.3. Electronic system
Two boards with 18 electrodes on each were used. Therefore 36 channels could be measured simultaneously. The external reference electrode employed was an Ag/AgCl device (supplied by CRISON).
3. Data analysis
The procedure for working with artificial neural networks consists of two stages, a first stage of training of the network and a second stage for its verification. The training stage is performed with some of the available measures. At this stage the network categories are set out (in our case the seven different types of water).The data form six electrodes for each measurement are applied as an input vector. With these data the coefficients of the algorithm that configures the network are calculated. In the verification stage,the data from new measures are applied to the inputs, checking whether the output of the active network is correct or not.
3.1. Training the Fuzzy ARTMAP
3.2. Training the multi-layer feed-forward neural network
3.3. Training the linear discrimination analysis
4. Conclusion
A microcontroller-based electronic tongue system, capable of discriminating between drinking water samples has been successfully developed. An 82.5% recognition rate has been achieved for the samples tested. This intelligent system may find application in the area of water quality monitoring.Pattern recognition algorithms have been applied to the classification. The main memory requirement for the algorithms canbe minimized sufficiently to fit in the limited memory space of a microcontroller. MLFF networks need many more training cycles than fuzzy ARTMAP and LDA. The algorithm which used the most memory of the microcontroller was the Fuzzy ARTMAP.MLFF and LDA used similar amounts of RAM memory but MLFF needs more program memory. Thus, the best pattern recognition algorithm to be implemented on a microcontroller is LDA. At present we are working with three research lines based on this work; honey classification, meal classification and chemical classification in a waste water depuration plant. Moreover, we are developing new electrodes as well as improvements in electrode stability as major topics for future work.
References
[1] F. Winquist, C. Krantz-Rülker, I. Lunström, Electronic tongues and combinations of artificial senses, Sensors Update 11 (2002) 279–306.
[2] A. Legin, A. Rudnitskaya, Y. Vlasov, C. Di Natale, F. Davide, A. D’Amico, Tasting of beverages using an electronic tongue, Sensors and Actuators B 44 (1997)291–296.
[3] A. Arrieta, C. Apetrei, M.L. Rodríguez-Méndez, J.A. de Saja, Voltammetric sensor array based on conducting polymer-modified electrodes for the discrimination of liquids, Electrochimica Acta 49 (2004) 4543–4551.
[4] G. Pioggia, F. Di Francesco, A. Marchetti, M. Ferro, A. Ahluwalia, A composite sensor array impedentiometric electronic tongue: part I. Characterization,Biosensors and Bioelectronics 22 (2007) 2618–2623.
[5]H. Sakai, S. Liyama, K. Toko, Evaluation of water quality and pollution using multichannel sensors, Sensors and Actuators B 66 (2000) 251–255.
[6]J. Gallardo, S. Alegret, M. del Valle, Application of a potentiometric electronic tongue as a classification tool in food analysis, Talanta 66 (2005) 1303–1309.
[7]G. Verrelli, L. Francioso, R. Paolesse, P. Siciliano, C. Di Natale, A. D’Amico,A. Logrieco, Development of silicon-based potentiometric sensors: towards aminiaturized electronic tongue, Sensors and Actuators B 123 (2007) 191–197.
[8] L. Lvova, E. Martinelli, E. Mazzone, A. Pede, R. Paolesse, C. Di Natale, A. D’Amico,Electronic tongue based on an array of metallic potentiometric sensors, Talanta70 (2006) 833–839.
[9]J. Soto, R.H. Labrador, M.D. Marcos, R. Martínez-Mᘠnez, C. Coll, E. García-Breijo,L. Gil, A model for the assessment of interfering processes in Faradic electrodes,Sensors and Actuators 142 (2008) 56–60.
[10] R. Martínez-Mᘠnez, J. Soto, E. García-Breijo, L. Gil, J. Iba˜ nez, E. Gadea, A multi-sensor in thick-film technology for water quality control, Sensors and ActuatorsA 120 (2005) 589–595.
[11] B.A. Botre, D.C. Gharpure, A.D. Shaligram, Embedded electronics nose and supporting software tool for its parameter optimization, Sensors and Actuators B146 (2010) 453–459.
[12] A.S. Abdul Rahman, M.M. Sim Yap, A.Y.Md. Shakaff, M.N. Ahman, Z. Dahari, Z.Ismail, M.S. Hitam, A microcontroller-based taste sensing system for the verification of Eurycomalongifolia, Sensors and Actuators B 101 (2004) 191–198.