Comparison of predictions of conventional mathematical and the artificial neural network models for the thermo physical properties of soya meal
Professorial Chair Lecture
Philippine National Oil Commission (PNOC) Professorial Chair Lecture
Place
University of the Philippines Los Banos, College, Laguna
Abstract
The study was performed to determine the thermophysical properties of soyameal such as moisture sorption isotherms (MSI), density and specific heat. MSI were determined using relative humidities, controlled by saturated salt solutions, ranging from 11% to 90% and exposed to constant temperatures of 15oC, 25oC, and 35oC. The data were fitted to several models available in the literature and the modified Chung-Pfost and the Guggenheim, Anderson and DeBoer (GAB) models satisfactorily described the sorption behavior of the product. The said models were chosen based on the low means square errors (MSE) as well as high coefficient of determination (r2). The bulk density and specific heat data were found to be linearly related to moisture content. Simple artificial neural network (ANN) models with 3 hidden layers were also fitted to the said thermophysical properties and predictions were closed to the predictions of conventional mathematical models.
Location
UPLB Main Library Special Collections Section (USCS)
College
College of Engineering and Agro-Industrial Technology (CEAT)
Language
English
Recommended citation
Madamba, Ponciano S., "Comparison of predictions of conventional mathematical and the artificial neural network models for the thermo physical properties of soya meal" (2024). Professorial Chair Lecture. 759.
https://www.ukdr.uplb.edu.ph/professorial_lectures/759