نوع مقاله : مقاله پژوهشی
نویسندگان
1 دانشجوی دکتری تخصصی گروه مهندسی بیوسیستم، دانشکده کشاورزی و منابع طبیعی، دانشگاه محقق اردبیلی، اردبیل، ایران.
2 دانشیار گروه مهندسی بیوسیستم، دانشکده کشاورزی و منابع طبیعی، دانشگاه محقق اردبیلی، اردبیل، ایران.
3 دانشیار گروه مهندسی بیوسیستم، دانشکده علوم کشاورزی، دانشگاه گیلان، رشت، ایران.
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
Rice is the staple food of about 2.5 billion people in the world. The quality of this product is influenced by various factors. As a result, quality control and fraud detection are important issues in the rice industry. The purpose of this research was to investigate and detect mixing in white rice cultivars using Vis/NIR spectroscopy, chemometrics and machine learning. In order to conduct experiments, 13 classes were prepared based on the amount of mixing of the main variety of rice (Hashmi variety) with non-main varieties of Fajr, Gilaneh, Khazar and Shiroudi, so that the gross classes resulting from mixing 90, 80 and 70% of Hashmi variety rice respectively with 10, 20 and 30% of other figures mentioned. Multilayer perceptron (MLP), decision tree (DT) and support vector machines (SVM) algorithms were used to classify rice samples. To evaluate the performance of the studied classifiers, statistical indices including sensitivity (Se), specificity (Sp), accuracy (Ac) and root mean square error (RMSE) were used in the form of developing asymmetry matrices of the classifiers. The results of the study showed that J48 DT, MLP and SVM with RBF kernel function with accuracy and RMSE values of 100% and zero, 96.92% and 0.0951, 92.31%, and 0.2483% respectively, were able to classify different samples of white rice.
کلیدواژهها [English]