Research On Orange Sorting Technology Based On Optical-Mechanical-Electrical Detection And Development Of Portable Instrument

The grading selection of internal quality and the classification selection of varieties are the important tasks in the transportation of citrus fruits.However,the traditional method needs to destroy the sample,it wastes time and energy and is challenging.Compared with traditional methods,the visible-near infrared spectroscopy method not only has the characteristics of non-destructive,green and non-pollution,but also has high stability and accuracy.In this paper,the internal quality of Gannan navel orange and three varieties of oranges include Jiangxi navel orange,Hunan Bingtang orange and Guangdong sweet orange are studied and analyzed based on optical-mechanical-electrical detection equipment,explore new methods,to achieve more stable and accurate internal quality detection of Gannan navel orange and classification and identification of different varieties of oranges,and based on the research results,a portable optical-mechanical-electrical testing instrument was designed for non-destructive detect the internal quality of Gannan navel orange.The specific research contents and conclusions of this paper were described as below:(1)Non-destructive detection of total acidity and soluble solid content in Gannan navel orange based on visible-near infrared spectroscopy method.The raw spectral data are preprocessed based on the savitzky-golay(SG)smoothing method,and then three algorithms are used including random frog(RF),genetic algorithm(GA)and successive projections algorithm(SPA)to extract characteristic wavelengths,and then,based on the raw spectral data and the extracted characteristic wavelength data,combined with partial least squares regression(PLSR),principal component regression(PCR),least square support vector machine(LS-SVM)and multiple linear regression(MLR),various prediction models for the total acidity and soluble solid content of Gannan navel oranges were established.The results shown that the prediction effect based on SG+GA+LS-SVM model is the best,for total acidity,the value of RMSEP is 0.016%,the value of_p~2 is 0.9834,and the value of RPD is7.76.For soluble solid content,the value of RMSEP is 0.395°Brix,the value of_p~2 is 0.9282,and the value of RPD is 3.73,effectively improve the detection accuracy.(2)Rapid classification and identification of different varieties of oranges based on visible-near infrared spectroscopy method.The raw spectral data are preprocessed based on the savitzky-golay(SG)smoothing method,and then two algorithms are used including genetic algorithm(GA)and successive projections algorithm(SPA)to extract characteristic wavelengths,and then,based on the raw spectral data and the characteristic wavelengths extracted by genetic algorithm(GA)and successive projections algorithm(SPA)are used to establish the classification prediction model between different varieties of oranges by combined with three analysis methods are used including soft independent modeling of class analogy(SIMCA),support vector machine(SVM)and convolutional neural network(CNN).The results shown that the classification prediction effect of using raw spectral data combined with CNN is the best,and the model operation achieves 98.85%classification accuracy only in the 39nd round of operation,meanwhile,the loss function value is 0.0123,which has a good stability,compared with common classification methods,it can achieve accurate classification while avoid multiple complex operation links such as data preprocessing and characteristic wavelengths extraction in the experiment process.(3)A portable spectrometer was designed based on visible-near infrared spectroscopy to detect the internal quality of Gannan navel oranges.The instrument consists of main components such as USB4000 spectrometer,touch screen,fiber optic,lens,light source and power supply and the functional operation programming of Gannan navel orange detection is realized based on JAVA language.The detection accuracy of the designed instrument was verified by detection 15 Gannan navel orange samples with the designed portable instrument,and the results shown that the designed portable optical-mechanical-electrical detection instrument can realize non-destructive detection of the internal quality of Gannan navel oranges,it has good practical application value.