Physical property changes and sensory evaluation modelling of osmotically dehydrated pumpkin using intelligent methods

The objectives of this study were to use intelligent methods to model sensory evaluation, as well as to predict physical properties changes of osmotically dehydrated pumpkin slices. The effects of process variables which are concentration of osmotic solution, immersion temperature and immersion time...

Full description

Saved in:

Bibliographic Details
Main Author:	Tang, Szu You
Format:	Thesis
Language:	English English
Published:	2018
Subjects:	TX341-641 Nutrition Foods and food supply
Online Access:	https://eprints.ums.edu.my/id/eprint/42204/1/24%20PAGES.pdf https://eprints.ums.edu.my/id/eprint/42204/2/FULLTEXT.pdf
Tags:	Add Tag No Tags, Be the first to tag this record!

id	my-ums-ep.42204
record_format	uketd_dc
spelling	my-ums-ep.422042024-12-16T04:07:51Z Physical property changes and sensory evaluation modelling of osmotically dehydrated pumpkin using intelligent methods 2018 Tang, Szu You TX341-641 Nutrition. Foods and food supply The objectives of this study were to use intelligent methods to model sensory evaluation, as well as to predict physical properties changes of osmotically dehydrated pumpkin slices. The effects of process variables which are concentration of osmotic solution, immersion temperature and immersion time were studied. For the first part of this work, the physical properties such as the net colour difference changes, ΔE, the texture which was determined in terms of hardness and shrinkage were determined. An Artificial Neural Network was developed by using the process variables as the input. A feed-forward backpropagation network with tangent sigmoidal function at hidden layer and output layer was trained and best network was chosen based on the highest correlation coefficients and lowest Root Mean Square Error (RMSE) between the experimental values versus the predicted values. As comparison, Response Surface Methodology (RSM) with three-level three-factor Box-Behnken design was employed. The performances were evaluated based on Model Predictive Error (MPE), correlation of determination (R2) and root mean square error (RMSE). In this study, the results showed that ANN has higher prediction capability as compared to RSM. Concentration showed the highest influence on colour change and shrinkage, followed by time and temperature. On the other hand, immersion time has the most significance effect on texture, followed by concentration and temperature. For the sensory evaluation modelling, 15 untrained panels evaluated the osmotic dehydrated pumpkin samples for various sensory attributes such as colour, taste, hardness and aroma. The sensory score for the sensory attributes was analysed using Fuzzy Logic techniques. The judges' preference on the significance of sensory attributes was taken as crisp numbers instead of linguistic forms by applying the 5-point fuzzy logic scale. Sensory quality of samples was then compared based on the estimated overall sensory scores, overall membership function, similarity values, and relative importance of sensory attributes. It was found that Sample S14 (Concentration=45 Brix, Temperature=50℃, Immersion Time=150 minutes) ranked the best in the category of “good” (Similarity value=0.7245), the samples that treated with low temperature is well accepted. The relative importance of quality attributes of OD pumpkin samples was ranked as Taste > Aroma > Hardness > Colour, indicating that taste is the most important quality attribute in this study. Besides, ANN was also employed to predict the overall acceptability of osmotically dehydrated pumpkin. Again, the process variables were used as input while independent overall acceptability was used as output of ANN. A feedforward backpropagation network with Training Levenberg–Marquardt (LM) as the training function was generated. The tangent sigmoid transfer function (tansig) at hidden layer and pure linear transfer function at output layer were used. It was found that ANN with one hidden layer comprising 9 neurons gives the best fitting with the experimental data, which can predict total acceptance with lowest RMSE (0.047) and highest correlation coefficients (0.9757). These results indicate that ANN is a powerful tool to estimate the overall acceptability of the pumpkin. Besides, solution concentration, solution temperature and immersion time has almost equal effect to the overall acceptability of OD pumpkin samples. 2018 Thesis https://eprints.ums.edu.my/id/eprint/42204/ https://eprints.ums.edu.my/id/eprint/42204/1/24%20PAGES.pdf text en public https://eprints.ums.edu.my/id/eprint/42204/2/FULLTEXT.pdf text en validuser masters Universiti Malaysia Sabah Faculty of Engineering
institution	Universiti Malaysia Sabah
collection	UMS Institutional Repository
language	English English
topic	TX341-641 Nutrition Foods and food supply
spellingShingle	TX341-641 Nutrition Foods and food supply Tang, Szu You Physical property changes and sensory evaluation modelling of osmotically dehydrated pumpkin using intelligent methods
description	The objectives of this study were to use intelligent methods to model sensory evaluation, as well as to predict physical properties changes of osmotically dehydrated pumpkin slices. The effects of process variables which are concentration of osmotic solution, immersion temperature and immersion time were studied. For the first part of this work, the physical properties such as the net colour difference changes, ΔE, the texture which was determined in terms of hardness and shrinkage were determined. An Artificial Neural Network was developed by using the process variables as the input. A feed-forward backpropagation network with tangent sigmoidal function at hidden layer and output layer was trained and best network was chosen based on the highest correlation coefficients and lowest Root Mean Square Error (RMSE) between the experimental values versus the predicted values. As comparison, Response Surface Methodology (RSM) with three-level three-factor Box-Behnken design was employed. The performances were evaluated based on Model Predictive Error (MPE), correlation of determination (R2) and root mean square error (RMSE). In this study, the results showed that ANN has higher prediction capability as compared to RSM. Concentration showed the highest influence on colour change and shrinkage, followed by time and temperature. On the other hand, immersion time has the most significance effect on texture, followed by concentration and temperature. For the sensory evaluation modelling, 15 untrained panels evaluated the osmotic dehydrated pumpkin samples for various sensory attributes such as colour, taste, hardness and aroma. The sensory score for the sensory attributes was analysed using Fuzzy Logic techniques. The judges' preference on the significance of sensory attributes was taken as crisp numbers instead of linguistic forms by applying the 5-point fuzzy logic scale. Sensory quality of samples was then compared based on the estimated overall sensory scores, overall membership function, similarity values, and relative importance of sensory attributes. It was found that Sample S14 (Concentration=45 Brix, Temperature=50℃, Immersion Time=150 minutes) ranked the best in the category of “good” (Similarity value=0.7245), the samples that treated with low temperature is well accepted. The relative importance of quality attributes of OD pumpkin samples was ranked as Taste > Aroma > Hardness > Colour, indicating that taste is the most important quality attribute in this study. Besides, ANN was also employed to predict the overall acceptability of osmotically dehydrated pumpkin. Again, the process variables were used as input while independent overall acceptability was used as output of ANN. A feedforward backpropagation network with Training Levenberg–Marquardt (LM) as the training function was generated. The tangent sigmoid transfer function (tansig) at hidden layer and pure linear transfer function at output layer were used. It was found that ANN with one hidden layer comprising 9 neurons gives the best fitting with the experimental data, which can predict total acceptance with lowest RMSE (0.047) and highest correlation coefficients (0.9757). These results indicate that ANN is a powerful tool to estimate the overall acceptability of the pumpkin. Besides, solution concentration, solution temperature and immersion time has almost equal effect to the overall acceptability of OD pumpkin samples.
format	Thesis
qualification_level	Master's degree
author	Tang, Szu You
author_facet	Tang, Szu You
author_sort	Tang, Szu You
title	Physical property changes and sensory evaluation modelling of osmotically dehydrated pumpkin using intelligent methods
title_short	Physical property changes and sensory evaluation modelling of osmotically dehydrated pumpkin using intelligent methods
title_full	Physical property changes and sensory evaluation modelling of osmotically dehydrated pumpkin using intelligent methods
title_fullStr	Physical property changes and sensory evaluation modelling of osmotically dehydrated pumpkin using intelligent methods
title_full_unstemmed	Physical property changes and sensory evaluation modelling of osmotically dehydrated pumpkin using intelligent methods
title_sort	physical property changes and sensory evaluation modelling of osmotically dehydrated pumpkin using intelligent methods
granting_institution	Universiti Malaysia Sabah
granting_department	Faculty of Engineering
publishDate	2018
url	https://eprints.ums.edu.my/id/eprint/42204/1/24%20PAGES.pdf https://eprints.ums.edu.my/id/eprint/42204/2/FULLTEXT.pdf
_version_	1818611450798669824

Physical property changes and sensory evaluation modelling of osmotically dehydrated pumpkin using intelligent methods

Similar Items