Application Of Electronic Nose In Bulk-curing Barn

Making tobacco leaf yellowing and flavory is the fundamental objective influe-cured tobacco production, but at present the control of curing process is based onleaf color lacking in evaluation for flavor, this resulted in tobacco exterior quality not inaccordance with aroma quality. In recent years, traditional chemical methods of analysisfor the chemical components in tobacco leaf are capillary electrophoresis(CE) and gaschromatography-mass spectrometry (GC-MS). It’s not realistic to determine allcomponents in tobacco leaf and smoke which contain thousands of categories ofcompounds. Even if the long list of compounds is measured accurately, there isambiguous about the components determining aromatic quality and smoking quality.The above situation makes it difficult to search for the relation between chemicalcomponents and internal quality in tobacco leaf.An electronic nose is a system which comprises an array of electronic chemicalsensors with partial specificity and an appropriate pattern recognition system, capable ofrecognizing simple or complex odors. Applied researches about electronic nose aregradually increasing in tobacco industry, while electronic nose is widely used in theidentification of smoke, difference of different tobacco producing areas, cigarettequality control and so on. Few researches are reported yet on monitoring the smell oftobacco in curing process. Hence, in order to implement and promote the technique offlue-cured tobacco, it is great significance of establishing rapid, convenient, scientifictobacco analysis evaluation method based on electronic nose technology.Data acquisition was accomplished by sensor array in tobacco curing experiments.Primarily, the curve reflecting the entire trend of smell change was obtained through thestudy of data, and demonstration of the practical meaning was performed from the pointof curing theory. Then, due to the features of that smell curve, prediction for peak ofcurve was proposed. And the practical meaning of it is that, through predicting thecoming time of peak in advance, proper control method in the next stage of crucialtemperature can be obtained beforehand so that the complex series of chemical changeswithin the leaves are accomplished optimally. ARMA model and BP model were built insimulation and comparison. The results show that ARMA model is exceeded by BPmodel on aspect of long-term forecast like the time of peak coming, ARMA model isgood at short-term forecast such as optimum terminal-time of curing. Finally, on that basis, a preliminary approach for controlling temperature-humidity in flue-curedtobacco manufacturing was put forward, which is an exploration for technicalimplementation of tobacco curing. In this method, corresponding control mode isdetermined by prediction of peak time in tobacco curing process.