Application of Genetic Algorithm in Sorting Dry Fruits and Agricultural Products: Optimization of Sorting Devices

Abstract

Sorting is a term that means grading and categorizing agricultural products and is an introduction to agricultural product packaging. In addition, in the fruit and vegetable markets of modern societies, almost all fruits and vegetables are offered in sorted and labeled form, and this will make it easier for the customer to recognize the quality of the product and will lead to a more regular distribution and supply. Sorting and grading of products is a continuous system that includes preservation, transportation, distribution, sale, and final consumption. However, all sorting and grading devices should be optimized and evaluated so that in case of any technical violation or update, they do not encounter problems in providing services and their work is not disrupted. Therefore, in this study, evaluation, and optimization of bulk raisin sorting machine, which is one of the main challenges of raisin producers and buyers in the world, was done. In this research, grape samples were randomly selected and prepared from the seedless white variety. For classification, digital image processing techniques were used to extract features from an image from the image processing toolbox in MATLAB. Other algorithms were used to evaluate and validate the accuracy of the results. These algorithms included honey bee inter colony harmony, differential evolution, and PSO. According to the results of the bee colony algorithm, it had higher accuracy, but the speed of convergence was lower and therefore needed to spend more time on calculations. But the genetic algorithm had almost the same accuracy as the honey bee algorithm, and on the other hand, the convergence speed was higher. Differential evolution and harmony search algorithms required many iterations of image processing and were not cost-effective in terms of computation time. On the other hand, raisin clustering requires high speed and accuracy in an industrial unit to avoid product wastage. Therefore, according to the results, genetic algorithms and PSO have the best performance in terms of speed, accuracy, and the need for fewer calculations than other algorithms, and these algorithms can be used for other sorting devices as well.