Implementation of an olfactory machine system for the classification of different types of black pepper based on geographical origin and detection of cheating in Indian black pepper

Spices are the most valuable medicinal plants used in food and medical science industries and due to quality and price diversity between various species, distinction, classification and separation of them based on purity and quality degree have great importance. Spices are produced in different countries, including India, Pakistan, China, and East and South Asian countries. The difference in the percentage of aromatic compounds in various types of spices from different regions has led to a distinction between spices. Also, profitable individuals for economic purposes and more profit without regard to the general health of the community will lead to the creation of adulteration in different types of spices. The most important of these adulterations is the addition of volatile ingredients such as cubeb pepper and palm kernel powder in black pepper.

In this study, an olfactory machine system based on eight metal oxide semiconductor sensors in combination with pattern recognition methods were used to classify and separate of black pepper samples based on geographic origin and also to detect cubeb pepper adulteration and palm kernel powder. The adulterated black pepper samples were tested with different adulteration levels (10, 20 and 30%).The fractional method was used to improve and optimize the electronic nose output signals before entering diagnostic methods. In order to analyze the extracted data from the sensor response signal, the principal component analysis method (PCA) was used. Based on the results, PCA with two main components of 96% for black pepper and 95% of cubeb pepper and palm kernel adulteration can be described from the variance of data. Also, three methods of linear separation analysis (LDA), Support vector machine (SVM) and decision tree (DT) were used to classify the samples. The use of the LDA method for black pepper showed a classification precision of 100%, and for adulterations, accuracy was 97.14%.

The results showed that SVM with Gaussian function has the highest accuracy in classifying black pepper samples, cubeb pepper, and palm kernel adulteration Also, the success rate of the DT method in separating and categorizing black pepper, cubeb pepper, and palm kernel was 96.66% and 88.5%  respectively.According to the results obtained, the machine olfaction system in combination with pattern recognition methods has the ability to detect and classify different black pepper samples from different geographical origin and the lowest level of adulteration.