13511

12304

كارشناسي ارشد

مكانيك بيوسيستم

اصفهان: دانشگاه صنعتي اصفهان، دانشكده كشاورزي

۱۳۹۶

سيزده، ۸۷ص.: مصور، جدول، نمودار

احمد ميره‌اي

جهانگير خواجه‌علي، مجيد ناظري

مرتضي صادقي، نفيسه پورجواد

1397/02/18

كتابنامه

كشاورزي

مهندسي كشاورزي

ID12304

88 Non destructive Detection of Quality and Insect Infestation in Quince Fruit Using Vis NIR Spectroscopy Elham Jahangiri elhamjahangiry@yahoo com March 14 2018 Department of Biosystems Engeeniring Isfahan University of Technology Isfahan 84156 83111 IranDegree M Sc Language FarsiSupervisor S A Mireei samireei@cc iut ac ir Abstract Today the quality evaluation of agricultural products is one of the most crucial post harvest activities Quince Cydonia Oblonga is a native fruit of Iran and due to its high nutritional value it is considered byother countries Quince pest Euzophera bigella Zeller and apple pest Cydia pomonella L are twoimportant pests that reduce the quality and quantity of the product Detecting the intact and infested fruitsis economically important however the infestation has usually no visual symptoms Therefore developing a nondestructive technique that can evaluate the quality indices of quince and detect theinfested ones in a fast and reliable manner seems to be necessary In this study Vis NIR spectroscopy intwo measurement modes of half transmittance and interactance was used to estimate the maturity indicesof quince including soluble solids content SSC firmness and color parameters of L a and b Moreover the ability of this method was assessed to nondestructive detection of insect infestation ofquince Spectral acquisition system consisted of the major components including adjustable light sources a PDA spectrometer equipped with high resolution CCD detector in the range of 1200 200 nm fiber opticprobe and flexible set up After collecting the spectra the samples were cut to ensure the degree and thetype of infestation Then the intact samples were used to evaluate the quality indices by using the standardmethods Different preprocessing methods were applied to the spectra to eliminate the unwantedinformation including the noise Partial least squares PLS regression was used for quantitative analysisand soft independent modeling of class analogy SIMCA and partial least squares discriminant analysis PLS DA was used for qualitative analysis Among different quality indices the best result was obtainedfor estimating the L in the interactance measurement mode with a coefficient of determination inprediction R2p of 0 866 root mean squares error of prediction RMSEP of 2 27 and residual predictivedeviation RPD of 2 63 The best result for estimating the firmness was obtained in the half transmittancemeasurement mode with R2p of 0 638 RMSEP of 8 98 N and the RPD of 1 57 The best result for the SSCprediction was achieved in interactance measurement mode with R2p of 0 509 RMSEP of 1 71 Brix andthe RPD of 1 39 The best result for the b was obtained in interactance mode with R2p of 0 849 RMSEPof 2 34 and RPD of 2 61 In qualitative analysis the best results were obtained by using the normalizationpreprocessing in both measurement modes when the SIMCA method was used for classification Theintact and infested classes were distinguished with the accuracies of 100 and 85 2 respectively at thetest set validation and the total accuracy was obtained as 95 10 In this method the interactance modedid not lead to the acceptable results The best PLS DA model could detect the intact and infested sampleswith a total accuracy of 94 Despite the lower total accuracy the PLS DA was superior over SIMCAsince its results had more consistency for both intact and infested classes Key Words Vis NIR spectroscopy Quince Partial least squares Soft independent modeling of classanalogy Partial least squares discriminant analysis

احمد ميره‌اي

جهانگير خواجه‌علي، مجيد ناظري

مرتضي صادقي، نفيسه پورجواد