شناسایی و طبقه بندی ارقام مختلف انگور با استفاده از برگ ارقام بوسیله ماشین بویایی

Introduction
Grape is a creeping plant that has ivy in front of some of its leaves. France, Italy and Germany are among the most important grape producing countries in Europe, and Iran is one of the most important centers for grape production and cultivation in the world due to its favorable geographical and climatic conditions. Grape fruit is divided into two types, seeded and seedless, each of which is found in different colors of red, yellow, black and almost green. In areas where the maximum temperature is not more than 40 degrees Celsius and the minimum temperature is not less than 15 degrees Celsius below zero, grape fruit grows better. Grapes are made from raisins, jellies, raisins, jams, vinegar and juice, and various products are made from grape seeds. This product is a good source of potassium, fiber and a variety of vitamins and other minerals. Is. According to available reports, there are about 800 to 1000 grape cultivars in Iran, and some of these cultivars are of great economic importance, especially for fresh consumption and preparation of raisins. In Iran, edible grapes are of the genus Winifra, and in addition, there are two types of Labrosca grapes, which are scattered in the north of the country, and wild grapes of the subspecies Westeris in the northern forests and wetlands of the Zagros Mountains. Grapes are widely distributed in terms of climate and have recently been cultivated in temperate and tropical regions in all parts of the world. By recognizing grape cultivars before fruit growth, it is an effective step in determining the purpose and use of the harvest product, in the meantime, the type of grape cultivar can be identified using new post-harvest technologies. One of these methods is to use an electronic nose to identify volatile compounds in grape leaves and to identify its cultivar. Electronic nose has been used in extensive research to identify and classify food and agricultural products.

Methodology
First, 3 varieties of grape leaves were obtained from vineyards located in Bonab city of West Azerbaijan province. These 3 cultivars were: Jovini, Aq Shaliq and Qara Shaliq. 200 grams of each of these leaves were prepared. After preparing leaves from different grape cultivars, first the samples were placed in a closed container (sample container) for 1 day to saturate the container space with the aroma of grape leaves, then the sample containers were used for data collection with the case of the electronic nose.
In this research, an electronic nose made in the Department of Biosystem Engineering of Mohaghegh Ardabili University was used. This device uses 9 low-power metal oxide (MOS) semiconductor sensors.
The sample chamber was connected to the electronic nose and data collection was performed. The data collection was done by first passing clean air through the sensor chamber for 100 seconds to clear the sensors of odors and other gases. The sample odor was then sucked out of the sample chamber by the pump for 100 seconds and directed to the sensors, and finally fresh air was injected into the sensor chamber for 100 seconds to prepare the device for repetition and subsequent tests. 30 replicates were considered for each sample.
The study began with the chemometrics method with principal component analysis (PCA) to detect the output response of the sensors and reduce the data dimension. In the next step, linear detection analysis (LDA) and support vector machine (SVM) were used to classify 3 grape cultivars. Principal component analysis (PCA) is one of the simplest multivariate methods and is known as an unsupervised technique for clustering data by groups. It is usually used to reduce the size of the data and the best results are obtained when the data are positively or negatively correlated with each other.
Linear Detection Analysis (LDA) is the most common monitored technique for separating samples into predetermined categories. This technique selects independent data variables to differentiate the sample that is to follow the normal distribution. The LDA is based on linear classification functions in which intergroup variance is maximized and intragroup variance is minimized.

Conclusion
The scores diagram (Figure 2) shows the total variance of the data equal to PC-1 (82%) and PC-2 (11%), respectively, and the first two principal components constitute 93% of the total variance of the normalized data. When the total variance is above 90%, it means that the first two PCs are sufficient to explain the total variance of the data set. Grape cultivars are well differentiated by PCA method. Therefore, it can be concluded that e-Nose has a good response to the smell of grape leaves and grape cultivars can be distinguished from each other, which shows the high accuracy of the electronic nose in detecting the smell of different products.
The correlation loadings plot diagram can show the relationships between all variables. The loading diagram (Figure 3) shows the relative role of the sensors for each principal component. The inner ellipse shows 50% and the outer ellipse shows 100% of the total variance of the data. The higher the loading coefficient of a sensor, the greater the role of that sensor in identifying and classifying. Therefore, the sensors located on the outer circle have a greater role in data classification and it is clear that the three sensors TGS2620, TGS822 and TGS813 have played an important role in identifying grape cultivars from their leaf aroma.
LDA and SVM methods were used to identify and differentiate grape cultivars based on the output response of sensors. Unlike the PCA method, the LDA method can extract multi-sensor information to optimize resolution between classes. Therefore, this method was used to detect 3 grape cultivars based on the output response of sensors. The results of detection of cultivars were equal to 100% and also the accuracy of SVM method for detection of 3 grape cultivars was equal to 83.33% (Figures 4 and 5).
In this study, an electronic nose with 9 metal oxide sensors was used to identify and differentiate grape cultivars using their leaf aroma. Chemometrics methods including PCA, LDA and SVM were used for qualitative and quantitative analysis of complex data using electronic sensor array. The electronic nose has the ability to be used and exploited as a fast and non-destructive method to distinguish grape cultivars from leaf odor. Using this method in identifying grape cultivars will be very useful for consumers, especially processing units and food industries in order to select appropriate cultivars.