Predicting physicochemical properties of melon (Cucumis melo L.) using ultrasonic technology and artificial neural network
As one of the favorite fruits widely produced and consumed in Indonesia, quality testing for melon (Cucumis melo L.) fruit is mostly done using destructive testing. To overcome this problem, this study aims to predict physicochemical quality properties of melon fruit non-destructively using ultrason...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
EDP Sciences
2023-01-01
|
Series: | BIO Web of Conferences |
Online Access: | https://www.bio-conferences.org/articles/bioconf/pdf/2023/25/bioconf_icosia2023_06008.pdf |
_version_ | 1797353078160097280 |
---|---|
author | Yazid Izdihar Fahar Khuriyati Nafis Wagiman |
author_facet | Yazid Izdihar Fahar Khuriyati Nafis Wagiman |
author_sort | Yazid Izdihar Fahar |
collection | DOAJ |
description | As one of the favorite fruits widely produced and consumed in Indonesia, quality testing for melon (Cucumis melo L.) fruit is mostly done using destructive testing. To overcome this problem, this study aims to predict physicochemical quality properties of melon fruit non-destructively using ultrasonic and artificial neural network (ANN). Fifty-nine Hami melons were tested to measure the attenuation value of ultrasonic wave emission as a non-destructive variable, along with density and age. Then, destructive testing was conducted to measure physicochemical properties consisting of fruit flesh firmness, total soluble solids (TSS), titratable acidity (TA), pH, and vitamin C. The test data obtained were processed using ANN to acquire prediction models, with non-destructive data as input variables and each destructive data as output variable. The results showed that the prediction values were still not accurate. Reliability analysis conducted on the test data set based on R2 values (R2 ) and Normalized Root Mean Squared Error (NRMSE) values showed that the predicted values were still unreliable with R2 values that were still very low, ranging from -0.776 to 0.485, although the NRMSE value was relatively good. |
first_indexed | 2024-03-08T13:26:00Z |
format | Article |
id | doaj.art-e4a1f980203a4471b944a954ab9ea44f |
institution | Directory Open Access Journal |
issn | 2117-4458 |
language | English |
last_indexed | 2024-03-08T13:26:00Z |
publishDate | 2023-01-01 |
publisher | EDP Sciences |
record_format | Article |
series | BIO Web of Conferences |
spelling | doaj.art-e4a1f980203a4471b944a954ab9ea44f2024-01-17T14:58:04ZengEDP SciencesBIO Web of Conferences2117-44582023-01-01800600810.1051/bioconf/20238006008bioconf_icosia2023_06008Predicting physicochemical properties of melon (Cucumis melo L.) using ultrasonic technology and artificial neural networkYazid Izdihar Fahar0Khuriyati Nafis1Wagiman2Department of Agroindustrial Technology, Faculty of Agricultural Technology, Universitas Gadjah MadaDepartment of Agroindustrial Technology, Faculty of Agricultural Technology, Universitas Gadjah MadaDepartment of Agroindustrial Technology, Faculty of Agricultural Technology, Universitas Gadjah MadaAs one of the favorite fruits widely produced and consumed in Indonesia, quality testing for melon (Cucumis melo L.) fruit is mostly done using destructive testing. To overcome this problem, this study aims to predict physicochemical quality properties of melon fruit non-destructively using ultrasonic and artificial neural network (ANN). Fifty-nine Hami melons were tested to measure the attenuation value of ultrasonic wave emission as a non-destructive variable, along with density and age. Then, destructive testing was conducted to measure physicochemical properties consisting of fruit flesh firmness, total soluble solids (TSS), titratable acidity (TA), pH, and vitamin C. The test data obtained were processed using ANN to acquire prediction models, with non-destructive data as input variables and each destructive data as output variable. The results showed that the prediction values were still not accurate. Reliability analysis conducted on the test data set based on R2 values (R2 ) and Normalized Root Mean Squared Error (NRMSE) values showed that the predicted values were still unreliable with R2 values that were still very low, ranging from -0.776 to 0.485, although the NRMSE value was relatively good.https://www.bio-conferences.org/articles/bioconf/pdf/2023/25/bioconf_icosia2023_06008.pdf |
spellingShingle | Yazid Izdihar Fahar Khuriyati Nafis Wagiman Predicting physicochemical properties of melon (Cucumis melo L.) using ultrasonic technology and artificial neural network BIO Web of Conferences |
title | Predicting physicochemical properties of melon (Cucumis melo L.) using ultrasonic technology and artificial neural network |
title_full | Predicting physicochemical properties of melon (Cucumis melo L.) using ultrasonic technology and artificial neural network |
title_fullStr | Predicting physicochemical properties of melon (Cucumis melo L.) using ultrasonic technology and artificial neural network |
title_full_unstemmed | Predicting physicochemical properties of melon (Cucumis melo L.) using ultrasonic technology and artificial neural network |
title_short | Predicting physicochemical properties of melon (Cucumis melo L.) using ultrasonic technology and artificial neural network |
title_sort | predicting physicochemical properties of melon cucumis melo l using ultrasonic technology and artificial neural network |
url | https://www.bio-conferences.org/articles/bioconf/pdf/2023/25/bioconf_icosia2023_06008.pdf |
work_keys_str_mv | AT yazidizdiharfahar predictingphysicochemicalpropertiesofmeloncucumismelolusingultrasonictechnologyandartificialneuralnetwork AT khuriyatinafis predictingphysicochemicalpropertiesofmeloncucumismelolusingultrasonictechnologyandartificialneuralnetwork AT wagiman predictingphysicochemicalpropertiesofmeloncucumismelolusingultrasonictechnologyandartificialneuralnetwork |