Study On Drift Phenomenon And Correction Technology For Smartongue Sensor Array

With the wide application of automatization technique in food industry, there is a roaring requirement for an objective, efficient and non-destructive method to assess food quality. Electronic tongue, a quantitative and qualitative analytical equipment, bears the features of non-destruction, real-time and fastness which make it suitable for the control of quality and safety for the food industry. SmarTongue is a new type of electronic tongue which makes use of the inert metallic electrodes of platinum, gold, palladium, tungsten, titanium and silver to construct a stable sensor array. Voltammetrical electrochemical pulse technique is used to produce the combinatorial pulse relaxation spectrum with originality. Cross-response resolving method is applied to acquire the comprehensive information about the examined object. Assessing of general features of products and detecting of multi-components qualitatively and quantitatively are finally accomplished fast, real-timely and on-line. SmarTongue shows evident discrimination for such complex systems as beverages and alcohol drinks from different brands and areas and so on.Sensor array drift, a common problem for all the equipment with electrode as the detecting component, needs to be solved urgently because it can not only impair the precision and accuracy of measurement but also limit SmarTongue's long-term industrial application. In this research work, the drift rule of SmarTongue was firstly explored by detecting the basic taste substances, electrolytes, alcohol and so on. The drift correction method was then established, optimized and validated by actual application.(1) Research on the drift phenomenon of SmarTongue sensor arraySmarTongue sensor array demonstrated a drift generally and a lineardrift in most cases when the basic taste substances were detected for a long time and the kiwi fruit juice for a short time. Meanwhile, data processing method for drift correction was established.(2) Design and optimization of the electrochemical cleaning system for sensor arrayPulse electrochemical cleaning scheme for the sensor array was designed on the basis of SmarTongue pulse voltammetrical principle. The electrochemical cleaning time and voltage were optimized by deactivating the electrode surface with pure milk of Yili, ice black tea of Masterkong, yogurt of Mengniu etc, combined with the data processing method of Central Composite Face-centered (CCF) and Partial Least Squares (PLS). Results showed that the optimal electrochemical cleaning voltage and time were 1.35V and 4.40s for platinum electrode, 1.73 V and 4.80s for gold electrode, 0.90V and 4.30s for palladium electrode, and 0.57 V and 4.40s for tungsten electrode respectively. Validation and stability experiments proved the improvement in the stability of SmarTongue system and the reduction in the sensor array drift after the optimized cleaning scheme.(3) Establishment of SmarTongue sensor array drift correction arithmeticTwo drift correction algorithm, Component correction (PCA) andadditive correction were used to compare the drift correction effect for the sour and sweet substances for a long time and the tea beverage for a short time. Component correction, additive correction and principle component analysis were found to show a gradually decreasing correction effect for the sour substances with silver electrode in 100 Hz and 1 Hz, for the salty substance with the palladium electrode in 100Hz and 1Hz, for tea beverage with palladium electrode in 100Hz and 1Hz, tungsten electrode in 10Hz and 1 Hz, and silver electrode in 10Hz and 1 Hz. In short, the component correction was more suitable for the correction of SmarTongue sensor array linear drift. Types of electrodes and segments of frequency demonstrated influence on the correction effect.(4) Establishment of the method for the choice of liquor reference samples Simulated-liquor, 56% ethanol, 1mmol/L potassium ferrocyanide, mixed liquor, and Jianzhuang liquor were taken as the reference samples for Chinese liquor. Component correction and the original PCA were used to analysis the data from platinum electrode in 100Hz and 1Hz, gold electrode in 100Hz, and silver electrode in 1 Hz. Results showed that when mixed liquor was taken as the reference sample, drift correction and discriminate effect (d > 3) were best, and at the same time, the plots were consistent with statistic results.