INFORM April 2026

ARTIFICIAL INTELLIGENCE INFORM 15

such as maximum yield or minimum cost. In PSO, each “particle” moves through the search space, updating its position based on its own best result and the best result found by the swarm. Over time, the swarm converges toward a high‑quality solution. Used alone or with machine‑learning models, PSO excels at solving nonlinear, multidimensional optimization problems. Other AI tools, such as digital twins and reinforcement learning (RL), can simulate factory behavior or use predictive models to adjust process variables without interrupting production. A digital twin provides a virtual replica of a facility, continuously synchronized with sensor data to update parameters and improve predictions. Within this environment, an RL agent can learn how to respond to changing conditions, such as shifts in feedstock, and test process adjustments before they are implemented in the actual plant. Although still in early stages, these AI tools could enable real‑time, adaptive process control. EDIBLE OILS AI tools are now being applied across every stage of edible oil production, from plant care and harvesting to processing, packaging, and quality assurance. “A few years ago, AI tools were used in the olive oil sector primarily for offline tasks, such as process optimization modeling,

AI tools have been used to optimize every stage of olive oil production, including cultivation, extraction, storage, quality assurance, and authentication. Source: Shutterstock

quality characterization, and authenticity verification,” says Juan Carlos Mejuto, professor of physical chemistry at the University of Vigo, in Spain. “Since then, the field has evolved from retrospective data analysis toward real time, integrated process control.” Jesus Simal-Gándara, also a professor at the University of Vigo, notes that the European Union’s OLEUM project and initiatives in Spain and Italy are already deploying AI-driven smart mills. As early as 2008, researchers combined an ANN with online near-infrared (NIR) spectroscopy to accurately predict moisture and fat content in olive pomace during cold extraction, allowing real‑time adjustment of centrifugation parameters to maximize yield. Since then, AI tools have expanded to

enable online monitoring of key quality indicators—such as acidity, peroxide value, moisture, and polyphenols— throughout the extraction process. AI tools are also reshaping quality assurance. ANN paired with UV/Vis spectrophotometry can estimate olive oil degradation during transportation and storage, enabling predictions of remaining shelf life. Other ANN models link chemical attributes (such as free acidity, peroxide value, specific absorbance, and phenol content) with sensory properties, demonstrating that AI can rapidly and objectively grade olive oil quality. Adulteration of extra virgin olive oil with cheaper seed oils remains a persistent challenge, driving interest in