PJB-2022-483
Analogizing Of Artificial Neural Network (ANN) and Response Surface Methodology (RSM) In the Prediction and Optimization of Quality Attributes By PCA in Heat Pump-Assisted Dehumidified Air-Dri
G. Pandidurai
Abstract
Moringa oleifera (M. Oleifera) is a storehouse of essential nutrients like protein, fibre, vitamins, minerals and phytochemicals. The consumption of moringa leaf in cooked form or supplemented as a fine powder in processed food products. The traditional drying method takes more time and energy and that will affect the organoleptic property, product quality and safety. Hence, the study aims to apply advanced techniques like heat pump-assisted dehumidified air drying (HPD), running effectively and efficiently to achieve higher retention of nutritional properties and create a forecasting model for the process optimization. The use of ANN as a tool to create a predictive model was evaluated against the existing RSM modelling method. For this, the output variables of crude protein (%), crude fibre (%), and colour values (L*, a* and b*) as an input of drying temperature (45 to 65 °C) and drying time (45 to 75 minutes). Physicochemical and features of a moringa leaf during drying were highly found at 55⁰C with special reference to maximum powder recovery, excellent flowability and better retention of nutrients like crude protein (29.64 %) and crude fibre (16.37 %). The potential and sensitivity analysis of the two models were compared using the regression coefficient (R2) and statistical functional error values RMSE, MAE and MAPE. The R2 values for ANN models were higher and the error values lower than that of RSM for all the response variables.
To Cite this article:
Download
