Design of a Dynamic Weighing System and AI-Based Sorting Process for Egg Sorting Machines

Abstract

Eggs are one of the world's most significant food sources since they include numerous critical nutrients such as protein, vitamins, minerals, and omega-3 fatty acids. Egg production and consumption have expanded dramatically in the previous two decades because of population growth and industrialization. To fulfill rising demand, automating of egg production facilities has become necessary. To distribute eggs to consumers and maintain quality requirements, eggs must be divided into weight categories. Due to production capacity, this process must be carried out using machines. High production volumes necessitate a rapid weighing process; hence eggs are weighed dynamically in machines. The weighing signal obtained from the load cell is filtered to determine the stable weight, which is then used to calculate the egg's weight class. In this paper, instead of performing all of these processes using classical approaches, a Stacked Autoencoder (SAE) based classification system is developed that will predict the egg's class using only raw weight data. To assess the effectiveness of the suggested method, classification performance was compared using support vector machines (SVM), k-nearest neighbors (kNN), and decision trees (DT). The suggested approach determines the weight class of the egg in roughly 0.084 sec with 100% accuracy. Given the increasing egg demand, the proposed technology allows for a considerably faster egg categorization procedure, boosting production speed and lowering production costs.