A new study has successfully developed and evaluated a compact milling unit for date seeds, employing advanced machine learning techniques to predict optimal operational conditions. This research not only enhances the utility of date seeds—a significant agricultural by-product—but also addresses the pressing issue of agricultural waste.
Date seeds, representing 5.6–14.2% of the fruit weight, have often gone underutilized, contributing to waste during date production. These seeds hold potential nutritional value, being rich in phytochemicals, dietary fiber, and beneficial fatty acids. Traditionally, the grinding of date seeds has posed challenges due to their hardness and the lack of specialized milling equipment, which has limited their use and the broader goal of sustainable resource utilization.
To overcome these challenges, researchers from Taif University and Al-Azhar University have developed and tested a compact milling unit integrating both crushing and hammer milling mechanisms. The machine's performance was rigorously evaluated under various operational parameters, and feedforward neural networks (FNN) were utilized to predict the best conditions for grinding.
The milling unit was first subjected to extensive testing of the cylindrical section, analyzing factors like cylinder speed, feed gate opening size, and clearance between cylinders. The results from this phase indicated optimal conditions with a cylinder speed of 150 rpm, 45 cm² feed gate opening, and 2 mm clearance. These parameters were then applied as baseline conditions for the hammer mill section.
Subsequently, the hammer mill was tested with variations of hammer speed and screen hole diameters. The FNN model excelled at predicting the most suitable operational parameters, achieving near-perfect accuracy with an R² value of 0.99974 and providing optimal conditions of 1750 rpm hammer speed and 6 mm screen hole diameter.
When operating under these ideal conditions, the milling unit was able to produce 30 kg/h of output with specific energy consumption of just 49 kW h/ton and yielding mean particle size of 2.14 mm. The authors noted, "Under optimal conditions, the machine achieved throughput of 30 kg/h, suitable for small to medium-scale operations."
These findings signify not just improved efficiency but also open new avenues for sustainable agricultural practices by turning waste materials, such as date seeds, to functional products. The study establishes the FNN's potential within agricultural machine optimization, marking the first implementation of such algorithms for date seed milling.
Moving forward, future research could explore real-time monitoring capabilities, adaptive control systems, and applications of this technology to other agricultural materials, potentially revolutionizing waste valorization processes.Drawing from extensive data, this research confirms the pivotal role of machine learning innovations like FNNs for enhanced productivity, lower operational costs, and reduced environmental impact, making it instrumental for future agricultural advancements.