| Summary: | 碩士 === 國立屏東科技大學 === 機械工程系 === 91 === Abstract
A near infrared detecting system was develop for evaluating sugar content of fruit in this study. This detecting system used a concave grating for simpled optical components; single component to have more sufficient splitting. The tungsten light source was illuminated on the fruit surface, then absorbed and reflected, and entered the concave grating of the monochromator through object lens focus. The light source was then focused on the sensitization component of the detector after splitting, and was analyzed with Modified Partial Least Squares Regression (MPLSR), Multiple Linear Regression (MLR) to establish the calibration equation for sugar content. Instrument functionality tests, including the light source selection and impact experiment, focusing point validation test for focal lens, ruled grating and concave grating splitting test, were conducted for future reference to online inspection system.
The results showed that the tungsten halogen lamp with ALU (aluminum) was more suitable for detection of near infrared ray zone. If the chromatic aberration factors based on various refractive indexes were not considered in the lens design, the intensity signal may be destructed and result in distortion of waveforms. In addition, the concave grating required simpled optical components; single component would be sufficient for splitting. Thus, the spectrometer developed with the concave grating would be more desirable for online detection. In the absorption spectrum test for fruit sugar content using the detection system constructed based on the abovementioned results, as well as the sugar content calibration equation for wax apple based on MPLSR five-factors model, the multiple correlation coefficient of calibration (rcal) was 0.949, the standard error of calibration (SEC) and standard error of cross validation (SECV) were 0.41 and 0.778°Brix respectively, and the calibration equation index (1-VR) was 0.641. The five optimal wavelengths in the MLR calibration equation for wax apple sugar content were 1228, 1038, 1447, 989, and 1176nm, where rcal and rval were 0.863 and 0.6833, respectively. SEC and SEP were 0.658 and 1.0413°Brix, respectively. In the simulation of classified detecting over 30 wax apples with this calibration equation, the selectivity was 80%.
|
|---|
