Today, the quality eva‎luation of fruit by consumers is not only limited to external components but also includes the internal quality components. Hence, much attention has been paid to non-destructive tests for the quick, non-destructive, and accurate eva‎luation of quality. Among non-destructive methods, NIR spectroscopy method has practical application and has been developed and widely used in recent years. In this thesis, firstly, a portable system was designed and developed to acquire the spectral data of fruits with different shapes and sizes. The major components of the system were included: main structure, illumination unit, spectroscopy unit, and processing and monitoring unit. To design the main structure, the final design of the system and its various parts were first simulated in CATIA software. The manufacturing drawings were then extracted and the main structure was made using materials as light as possible. During the manufacturing process, laser cuts, brake bending, and Argon welding were used to increase the accuracy. In the illumination unit, 4 halogen lamps (100 W) were used for radiating the samples in half-transmittance mode. The intensity and time of exposure of the lamps were adjusted by an electronic circuit, designed and developed for this purpose, in order to create sufficient light for fruits of different sizes. The spectroscopy unit consisted of a probe for collecting the lights passed through the sample, a PDA spectrometer, and the fiber optic interface. The probe was designed and developed to collect the transmitted light passed through the fruit and transfer it to the optical fiber connected to the spectrometer with minimal loss. The PDA spectrometer was equipped with a CCD detector capable to acquire the Vis/NIR spectra in the range of 400-1100 nm. The processing unit consisted of a Thin-Client compact computer made by DELL, with 4 GB of RAM, 64 GB of storage space, and a 2.4 GHz quad-core processor. The monitoring unit used a 7-inch touch screen to display the data obtained from the system. The system reliability was then validated by investigating its ability to estimate the main maturity parameters of the apple including firmness and soluble solids content (SSC). For this purpose, a total of 192 apple samples were harvested in three different harvesting times from a garden located in Aleshtar, Lorestan province. Three spectra were collected from each sample at an angle of 120 degrees around the stem-calyx axis of the fruit. After spectral data were recorded, the firmness and SSC of the samples were measured using reference tests. To analyze the obtained spectra, the appropriate preprocessing method for each parameter was selected and applied to the spectral data. The partial least squares (PLS) regression method was used to develop the predictive models. The results showed that to estimate the firmness, the best PLS model was obtained by using the 1st derivative preprocessing. This model could predict firmness values with a coefficient of determination in prediction (R2p) of 0.856, a root means square of error in prediction (RMSEP) of 5.47 N, and a residual predictive deviation (RPD) of 2.55. The best model for estimating SSC was obtained using a combination of smoothing and normalization pretreatments. The corresponding model was able to result in an R2p of 0.767, an RMSEP of 0.8 ◦Brix, and an RPD of 1.9. These results indicate the good potential of the developed system in predicting the maturity components of apple fruit.

احمد ميره اي

مجيد ناظري

مهرنوش جعفري، ميلاد فتحي