Data-Driven Hybrid CNN-LSTM Neural Networks for Predicting Drying Kinetics of Green Tomatoes (Physalis ixocarpa) Integrating Mathematical Models

This study investigates the drying kinetics of green tomatoes under controlled conditions, comparing 14 mathematical models including Newton, Page, Modified Page, Henderson and Pabis, Logarithmic, Two-Term, Wang and Singh, Thomson, Approximate Diffusion, Verma, Modified Henderson and Pabis, Two-Term Exponential, Simplified Fick’s Diffusion, and Midilli et al. The Verma, Two-Term Exponential, and Modified Page models demonstrated superior accuracy (R2 > 0.998, RMSE 0.01268 - 0.01345). A novel approach integrating a Convolutional Neural Network (CNN) with Long Short-Term Memory (LSTM) optimized drying process prediction, significantly enhancing moisture content prediction over time. This hybrid CNN-LSTM model, coupled with the accurate mathematical models (especially the Two-Term Exponential), offers a robust framework for predicting and optimizing food dehydration.