| Rapid and exact nondestructive detection came to catch scientist and researcher's attentions with rapid development of society and improvement of productivity recently. Detection with electronic nose is one of novel nondestructive detection. With rapid development of science and technology, the study on electronic nose has a quickly evolvement. Electronic nose has been broadly applied in industrial fields, and which showed a greatly potential application in evaluation of farm produce quality.The aim in this research is mainly developing the implicational application of electronic nose on wheat freshness assessment A portable electronic nose (PEN2) with ten Metal Oxide Semiconductor (MOS) array made by Airsense Analysis GmbH Corporation of Germany has been used. The research object was the wheat yielded from 1999 to 2003 and their intermixed sample in diverse proportion. Their odor quality, physical and chemical qualities were monitored by electronic nose and standard method of P. R. C. respectively.Based on some responsibility detection of electronic nose on various dated wheat, some feature parameters were confirmed: detection time and flush time is 65s respectively, sample mass used is SOg, sealed time is 90min, and the sampling method is static headspace with simple manual sampling device.In this study, statistic pattern recognition principle component analysis (PCA) and linear discriminant analysis (LDA) were used to distinguish the different dated wheat. The result showed that PCA can well discern the various samples which were the five different-year wheat and their interblend samples in the proportion of 23%, 37.5% and 50%, and the proportion of 12.5% might be a lower limit for PCA. Similarly, LDA can also discern the various samples which were the five different-year wheat and their interblend samples in the proportion of 12.5%, 25%, 37.5% and 50%. And that, the return discriminant ratio for the training muster acquired with electronic nose was 100%. It indicated mat the results accepted from PCA and LDA had good feasibility and veracity.The signal data gathered with ten sensors was compared; it was found that different sensor's responsibility to signals was diverse. The second sensor showed the strongest signal responsibility, then the ninth sensor, then the sixth, seventh and eighth sensor, last the first, third, fourth and fifth sensor. According to different sensors have different response to examine targets, we can optimize and choose the sensor array which make up of the electronic nose system. Optimizing the sensors array, can improve the concentricity of all kinds of wheat analyzed by the method of PCA, and the total contribution rates of the principal component rise too; but on the other hand, it has no apparent influence to LDA, the concentricity of distinguishing area does not change much, some become better slightly, but some worsen slightly, and the total contribution rates rise and drop a litter too.In order to evaluate the prediction potential of electronic nose technique to different dated wheat, the calculation and analysis were carried out using a comprehensive data analysis tool The Unscrambler for Exploratory Statistics, Multivariate Analysis, Classification, Prediction, and Design of Experiments. Partial least square (PLS) was used to build the predict models, and predicted correlation coefficient for new sample is 0.8613.The same samples' physical and chemical quality was detected, and the result showed it has a evident positive linear correlation (R=0.9815) between fatty acid value of wheat and storage years, a lightly negative linear correlation between protein content and storage time. Longer storage time, wet gluten content was lightly... |
PDF Full Text Request