Application of experimental, numerical, and machine learning techniques to improve drying performance and decrease energy consumption infrared continuous dryer

Abstract

Machine learning algorithms offer innovative and reliable solutions for addressing food spoilage and optimizing the drying processes. Stable food supply chains, lower post-harvest agricultural losses, and less perishable fruit and vegetable deterioration can be achieved using efficient drying techniques. This study explored and evaluated the energy dynamics of an infrared continuous drying system for garlic slices. Machine learning models (ML), including self-organizing maps (SOM) and principal component analysis (PCA), were employed to model and predict the relationships between process input parameters, such as infrared power, airflow rate, and air temperature, and response parameters, including thermal efficiency, effective moisture diffusivity, total energy consumption, drying duration, and specific energy consumption. The results showed that higher intensities of infrared radiation and air temperature significantly shortened the drying duration, whereas higher airflow rates extended the drying duration. Moreover, elevated air temperatures, increased infrared intensity, and reduced airflow rates considerably improved energy efficiency metrics. This research offers valuable insights into optimizing garlic slice drying while promoting energy conservation. The ANN model proved to be a robust tool for predicting and optimizing drying parameters, including drying duration, energy consumption, and thermal efficiency. Notably, SOM visualization demonstrated that elevated air temperatures and infrared radiation intensity were associated with reduced energy use, specific energy consumption, and dehydration.