This research studies the application of e-nose for a high quality perfume classification differs by using sensor array. The main motivation for the development of e-nose The device offers a low-cost, and can be analyzed to identify is reliable is a perfume based on purpose and could be measured repeatedly. In the trial, for example, the perfume from perfume and fake perfumes 3 type using 7 gauge sensors of various types of gas response from each perfume scent. Do the experiment by multilayer perceptron method comparison, RBF, SVM, J48, Bayesnet, and experiment with the source data. Data dimension reduction by extraction of information (Feature Extraction) with PCA and the data dimension reduction using attribute selection (Feature Selection) with the cfsSubsetEval to teach and test 10 fold cross validation outcomes demonstrate that SVM methods are recognized and be able to recognize the different types of perfume and when used in conjunction with SVM feature selection (Feature Selection) with the cfsSubsetEval method gives you 100% accuracy from experimental results demonstrate that e-nose has the ability to recognize when a perfume is forged and can be applied to e-nose in the perfume. Perfume review more effectively.
การแปล กรุณารอสักครู่..
