Study On The Influence Of Detection Speed In Quality Of The Fruits On The Applicability Of The Model By Near-infrared Transmission

China is a big Fruit production country, It is an important research of fruit internal quality quickly, accurately and non-destructive detection for improving the market competition. The study adopt near infrared transmission spectroscopy to detect the fruit ingredients, The near infrared transmission spectroscopy analysis technology was used for fruit ingredients analysis, in which the content of sugar are inspected. In order to provide new reference for the online non-destructive examination and identification for the fruit internal quality detection, several essential issues are discussion in terms of the applicability of segement calibration, the integration time selection in different velocities, the modeling method optimize in different velocities. The main contents and conclusions of the dissertation are as follows:1. According to the online testing requirement, a test for on line fruit internal quality determination was set up which can access accurate and effective near infrared spectroscopy, it provided hardware foundation for on line non-destructive testing of fruit internal quality; And determine testing instruments that can accurate measure apple sugar index, The results showed that using concentration residual distribution to eliminate abnormal samples can improve the precision of the model.2. The influence of the near-infrared spectral bands on the principal components was studied, and analysis the correlation between the model and the apple sugar content. Compare the applicability of different bands model The results showed that:In range of 650-850nm,It can establish the best model,determine the wavelength range of dynamic spectral.3. complete randomized experiment about speed and integration time were arranged for testing the model impact in different integration time, The results showed that:the size of integration time has little effect on the static spectral calibration; However, for dynamic spectral calibration, select the appropriate integration time that close to the time of the sample through the light can improve the accuracy of the model.4. The influence of the calibration effect and prediction results on different modeling method was studied, compare the difference of single-spectrum calibration and multi-spectral calibration. The results showed that:the optimized multi-speed spectral calibration method have improved the prediction accuracy, in order to testing the repeatability of the mathematical model, continuous acquisition spectrum in the short time, obtained the standard deviation of two random samples were 0.055 and 0.078 that meet the real-line testing requirements.