Application And Study Of Electronic Tongue Based On Virtual Instrument

Electronic tongue is a multi-sensors system which can be used to analyse liquids . It is composed of sensors array, signal process and pattern recognition system. After the responses and output signals of sensors array processed by computer, we can get the results that represent the taste characters of samples. The applications of electronic tongue are very abroad, such as the diagnoses and repair of human organ in the field of biomedicine, intellectual robot, food, environment, detection of atmosphere pollution, military safety, chemical and biologic weapons, antiterrorism.In this paper, we developed a electronic tongue system on the basis of comprehensively analyzing the research instance, development trend and the sensing mechanism of sensors. The configurations and functions of each part as follow:1.Sensors array: Sensors array is made up of ion selective electrodes(pH, p-Na, p-K,p-Ca) and reference electrode. The main function of electrodes is response to liquid samples.2 Signal process module: It can transform the high impedance and micro-current signals from sensors array to data acquisition card and match the iuput parameters.3 Data acquisition card: A data acquisition card (USB-6008 ) from NI(National instrument)company is used to acquire data transformed by signal process module and send them into computer.4 Interface of electronic tongue system : We developed the instrument interface of electronic tongue system by the software LABVIWE 7.1 offered by NI company .it can achieve man-machine conversation and instrument control.5 .Matlab pattern recognition arithmetic: Two typical electronic tongue pattern recognition arithmetics , namely Principal component regression(PCA),Back propagation- Artificial neural network(BP-ANN) , and their Matlab programs are described in this paper. They analyse and identify the data processed and return the results to be displayed by Lab VIEW.Experiment is implemented to distinguish from five orange juices and six mineral waters. Both PCA and BP-ANN are used as the pattern recognition arithmetics, and, the results are compared. It is shown that the stabilization and reiteration of electronic tongue system or choice of different pattern recognition arithmetics or sensors will affect the precision of electronic tongue greatly. The effect of identifying mineral waters is better than orange juices by electronic tongue. The results are satisfied and the total precision is 94.4% and 90% respectively.We evaluate the capability of the system and present improvement methods in the end of this paper.