| Soluble Solids Content(SSC) is an important internal quality index of watermelon,which determines watermelon's value. Non-destructive detection for SSC is significant to improve watermelon's internal quality. And it is urgent to enhance the industry of watermelon in China.Currently, Near-infrared(NIR) spectroscopy is widely applied to the non-destructive detection of fruit, which has advantages of fast, non-destruction,no sample preparation and good reproducibility. At this stage, the research subjects of NIR applied in fruit quality detection are mainly in middle and small size, with thin-skin fruit, such as apple and pear.Relatively few studies on big and thick-skin fruit, such as watermelon. And the transmission spectroscopy is used widely, which is not fit for portable detection. For the problem,“jingxiu” watermelon with relatively small-size and thin-skin is selected for the research. The near-infrared transmission spectroscopy technique is used to predict SSC of different positions of the watermelon's surface. The evaluation method of SSC in small-size watermelon is built and the fast detection system based on LabVIEW is designed. The main research contents and results are summarized as follows:(1) The penetration depth of near-infrared lights propagating in “jingxiu” watermelon is studied, which offers the theoretical basis for the design of the annular optical fiber probe.The diffuse reflection spectra are measured with removing the sample slices along the perpendicular bisector of the source-detector line. Based on the spectrum in the range of750~900nm, a series of penetration depth curves are fitted by partial least square(PLS). It is shown that, the light penetration depth increases with the source-detector distance increasing.With the deeper penetration depth, more information of the internal quality is carried by spectra data. However, the spectra signals are weaker. Based on these results, an annular optical fiber probe suitable for detecting internal information of watermelon is designed. The radius of the probe, namely the source-detector distance, is 20 mm, which ensures the near-infrared lights can penetrate adequately into watermelon and the detected spectrum can reflect the internal quality.(2) The influence of the detection positions on the accuracy of NIR prediction model forSSC of watermelon is studied. The exact prediction of SSC at different positions of watermelon is realized, thus improving the generality of the model. The prediction models of single detection position(equator, calyx and stem) and the model of three-combined detection positions for SSC are established by Partial least square(PLS). The results show that the accuracy of the prediction model based on three-combined detection positions is better than the model of single detection position. And the correlation coefficient of prediction(RP) is all above 0.85.The influence of the detection positions on the accuracy of NIR prediction model is greatly reduced.(3) The influences of the characteristic wavelength optimization algorithms on accuracy and efficiency of prediction model are investigated, thus optimizing NIR prediction model for SSC of watermelon. Successive projections algorithm(SPA) and Competitive adaptive reweighted sampling(CARS) are adopted to select effective variables of NIR spectroscopy for SSC of watermelon. Finally, PLS prediction models are established based on full-wave band and characteristic-wave band, respectively. The results show that the CARS-PLS prediction model is simplified significantly and the prediction accuracy for SSC is improved greatly. And the CARS-PLS model can achieve fast and precise detection for SSC of watermelon.(4) The fast detection system based on the optimized model and LabVIEW is developed.With the design of hardware and software, the spectra data are transmitted in real time to PC by RS-232 serial port. And the CARS-PLS model is adopted to predict SSC of watermelon.The fast and real-time detection for SSC of watermelon can be realized. |
PDF Full Text Request