Umudike Journal of Engineering and Technology

Michael Okpara University of Agriculture, Umudike


PREDICTION OF PHYSICAL PROPERTIES OF DRIED TURMERIC (CURCUMA LONGA) RHIZOME USING ARTIFICIAL NEURAL NETWORK

Ude, C. J.
Food Process System Engineering Research Unit, Chemical Engineering Department, Michael Okpara University of Agriculture, Umudike, Abia State State, Nigeria

Nwankwo, H. F.
Food Process System Engineering Research Unit, Chemical Engineering Department, Michael Okpara University of Agriculture, Umudike, Abia State State, Nigeria

Ibrahim, U. M.
Food Process System Engineering Research Unit, Chemical Engineering Department, Michael Okpara University of Agriculture, Umudike, Abia State State, Nigeria

Ayanyemi, J. O.
Food Process System Engineering Research Unit, Chemical Engineering Department, Michael Okpara University of Agriculture, Umudike, Abia State State, Nigeria

Dirioha, C.
Agricultural and Bio-resources Engineering Department, Michael Okpara University of Agriculture, Umudike, Abia State, Nigeria

Oke, E. O.
Food Process System Engineering Research Unit, Chemical Engineering Department, Michael Okpara University of Agriculture, Umudike, Abia State State, Nigeria



ABSTRACT

This study was based on optimal prediction of the physical properties of turmeric rhizome drying using Artificial Neural Network (ANN) developed in Matlab R2014b. The drying experiments were conducted at Drying Temperature: 40-650C, Drying Time: 0-240 minutes, thickness: 2mm – 5mm and air velocity: 1.5-3.0 to determine the angle of repose of the dried turmeric using wood, mild steel and glass surface, bulk density and surface area. Levenberg–Marquardt backpropagation (LMA) showed the best prediction among the ten backpropagation algorithms in all the physical properties studied. The optimization of the ANN model using LMA algorithm gave the best MSE value for angle of repose using wood at neuron 8 (0.000041), angle of repose using mild steel at neuron number 13 (0.000021), angle of repose using glass at neuron 7 (0.031223), bulk density at neuron 9 (0.000017) and surface area at neuron 14 (0.000256). It was concluded that increase in the number of neurons yield better prediction for the model which will be essential in process control and modelling of drying systems for food crops.



Keywords: Turmeric, Artificial Neural Network, Drying, Physical properties, Backpropagation


View: 178 | Download: 6

Conference Code
IECON2019

Conference Title
ENGINEERING FOR SUSTAINABLE ECONOMIC DIVERSIFICATION, FOOD AND NATIONAL SECURITY

ISBN
978-978-53175-8-9

Date Published
Friday, September 20, 2019

Conference Date
2nd - 4th September, 2019

The contents of the articles are the sole opinion of the author(s) and not of UJET.


Open Access
Umudike Journal of Engineering and Technology makes abstracts and full texts of all articles published freely available to everyone immediately after publication thereby enabling the accessibility of research articles by the global community without hindrance through the internet.

Indexing and Abstracting
We are index in Google Scholar