| At present, during the process of fruit planting, management, transportation, storage and sale, it is hard for fruit growers and sellers to make a nondestructive detection of fruit quality anytime and anywhere since there is a lack of reasonable and effective rapid detecting instrument. This largely hinders the pre-harvest and post-harvest management of fruit and even has an influence on the quality of fruit, thus leads to a serious waste. Visible/Near infrared spectroscopy has been widely applied to the fast nondestructive testing of the internal quality of agricultural products due to its characteristics of fast, convenient and green.This research mainly took apple as the object and aimed to develop a portable instrument for fruit sugar measuring using Visible/NIR spectroscopy, which can be used in the area of fruit garden management and fruit saling service in supermarket to get the weight and sugar information at the same time. Firstly, combined with Visible/NIR spectroscopy, optical fiber sensing technology and chemometrics methods, preliminary experiment platform was established and specified with optimal spectra acquisition conditions by comparing the performance of the prediction models based on the spectra collected in different lighting modes?different detectors and different integration time, respectively. Secondly, the optimal spectra acquisition conditions were used to build the structure of the portable instrument, and develop the control system and host software, then completed the preliminary research of the portable Visible/NIR instrument. In the end, experiment verification was made to test the instrument.In this paper, the main research contents and conclusions are as follows:(1) In the study of the influence factors and methods of spectra collection, a preliminary experiment platform was built and used to find out the best methods and factors of spectra collection by comparing the model performances based on different spectra collection modes. In the study of lighting mode, the performance of one-side irradiation pattern is superior to that of two-side irradiation pattern. Meanwhile taking one-side irradiation pattern can make the instrument more compact and dramatically reducing the heat. In the study of different detectors mode, the performances using Maya2000Pro spectrometer is superior to QE65Pro. In the study of different integration time, the spectra collected by an integration time of 200ms obtained the best prediction.(2) A portable fruit sugar-measuring instrument was developed through the designing of structure, building of the control system and the programming of the host software, which realized the collection of weight and spectral information of fruit at the same time. In the structure part, the weight collection system and spectrum collection system were combined together. In the hardware part, a weighing module and steering control system were built which can communicate sucessfully with host embedded computer. Through controlling of the wheel by steering control system, it achieve the freedom switch among a dark spectrum collection mode, filter mode and fruit detection mode. In the software part, a host computer system was developed to communicate with each module, to proces the information of fruit spectrum and weight, and to predict sugar content. At last, finishing the combination of instrument appearance, hardware and software system.(3) Prediction models were established and optimized through wavelength selection which can improve the prediction accuracy. Three wavelength selection methods, including GA (Genetic Algorithm), SPA (Successive Projection Algorithm) and CARS (Competitive Adaptive Reweighted Sampling), were used to extrate the useful information from the full-range spectra based on which the PLSR models were built. A comparison of the performances of CARS-PLS, GA-PLS, SPA-PLS models showed that CARS was better than the others with rcal, RMSEC, rpre and RMSEP being 0.8571,0.4594°Brix,0.8391 and 0.4545°Brix, respectively. Only 28 wavelengths were selected by CARS and used to establish the apple sugar content prediction model for the instrument. |
PDF Full Text Request