Show simple item record

Multivariate analysis and artificial neural network approaches of near infrared spectroscopic data for non-destructive quality attributes prediction of Mango (Mangifera indica L.)

dc.contributor.advisorLücke, Wolfgang Prof. Dr.
dc.contributor.authorMunawar, Agus Arip
dc.titleMultivariate analysis and artificial neural network approaches of near infrared spectroscopic data for non-destructive quality attributes prediction of Mango (Mangifera indica L.)de
dc.contributor.refereePawelzik, Elke Prof. Dr.
dc.description.abstractengMango is one of the most important and popular tropical fruits for people around the world due to its taste, appearance and excellent overall nutritional source from which lead to a heavy demand in world fruit market. With the increasing demand and consumption of mango, quality control becomes more and more important nowadays. Many relevant authorities are setting such criteria for quality factors to ensure good chain supply of mangoes. Therefore, to ensure the chain supply of good quality fruit, it is important to sort and grade mango based on its quality. To determine quality parameters in mango, several methods were already widely used in which most of them are based on solvent extraction followed by other laboratory procedures. However, these methods often require laborious and complicated processing for samples. Also, they are time consuming and destructive. Hence, a rapid and non-destructive method is required as an alternative method in determining quality parameters of mangos. Near infrared spectroscopy (NIRS) has become one of the most promising and used non-destructive methods of analysis in many field areas including in agriculture due to its advantage; simple sample preparation, rapid, and environmental friendly since no chemical materials are used. More importantly, it has the potential ability to determine multiple quality parameters simultaneously. Since NIRS itself cannot reveal chemical information in the spectra, chemometrics is required to extract the information about quality attributes buried on NIR spectra through a process called multivariate calibration from which a mathematical relationship between NIR spectra and the measured quality parameter will be revealed to determine desired quality attributes. Thus, the main objective of this study is to investigate the use of NIRS as non-destructive method combined by chemometrics for quality attributes in term of soluble solids content (SSC), titratable acidity (TA) and ascorbic acid (AA) predictions of intact mango. A total of 99 mangos were used as samples in the study from which NIR spectra were acquired at wavelength range of 1000-2500 nm. A reference measurement for desired quality attributes were obtained by standard laboratory procedures: solvent extraction, refractive index by refractomter (for SSC), and titration method (for TA and AA). Chemometrics, which are include principal component analysis (PCA), outlier detections, spectra pre-processing (mean centering (MC), mean normalization (MN), de-trending (DT), multiplicative scatter correction (MSC), standard normal variate (SNV) and orthogonal signal correction (OSC)), linear calibration models by principal component regression (PCR) and partial least square regression (PLSR), and non-linear regression by supporting vector machine regression (SVMR) and artificial neural networks (ANN) were applied to reveal chemical information buried in the NIR spectra by creating calibration models followed by validation or prediction for models evaluation. The results show that for linear regression method, PLSR seems to be more accurate and robust than PCR. From the spectra pre-processing point of view, the use of MSC, SNV and OSC prior to PLSR models development, significantly has an impact to the model accuracy and robustness. This can be seen from the reduction of latent variables used in PLSR (3 LVs after MSC and SNV, and 2 LVs after OSC), and increasing coefficient of determination (R2) and residual predictive deviation (RPD) index in calibration and validation. Based on accuracy (R2, RMSEC and RMSEP) and robustness index (RPD), the non-linear regression method (SVMR or ANN) was found to be better than linear regression (PLSR). The most optimal models for mango quality attributes prediction were achieved when ANN is used in combination with PCA as input. Thus, it may conclude that NIRS coupled with proper spectra pre-processing and regression method may be used as non-destructive technique for quality attributes measurement of intact
dc.contributor.coRefereeMörlein, Daniel Dr.
dc.subject.engNear infraredde
dc.subject.engnon-destructive methodde
dc.subject.engmultivariate analysisde
dc.subject.engartificial neural networkde
dc.affiliation.instituteFakultät für Agrarwissenschaftende
dc.subject.gokfullLand- und Forstwirtschaft (PPN621302791)de

Files in this item


This item appears in the following Collection(s)

Show simple item record