| As a characteristic fruit of Xinjiang,Hami melon well receives by people.In recent years,The annual export volume of Hami melon is increasing year by year,but its output value has not increased.The Hami melon directed into the market in the fields without any detection,and it will affect the quality and selling price of Hami melon,and can not do pricing.In this study,the characteristics fruit of Hami melon in Xinjiang was taken as the research object,the sugar content of Hami melon was used as the index,the online measurement system for sugar content of Hami melon was designed and constructed,based on visible near infrared spectroscopy technology,the system mainly contained the hardware and the design of the spectrum acquisition software.In order to verify the feasibility and effectiveness of the detection system that built by the design,a series of experiments were carried out based on this platform.At the same time,different modeling methods and different spectral pretreatment methods were used to model and analyze the spectrum and sugar content of Hami melon.In this paper,the main work and conclusions were as follows:(1)The online measurement system for sugar content of Hami melon was designed and constructed,which based on visible near infrared spectroscopy technology,and the system mainly contained two parts,which hardware and spectrum acquisition software.In the hardware part,the selection of the related instruments and the design of the related parts in the conveying device,illumination design and spectrum acquisition device were completed;The spectral acquisition software had completed the design of the relevant function modules of spectral acquisition.(2)To complete the test and integration of the online inspection system of visible near infrared spectrum of Hami melon sugar degree,and the online test was carried out.The online spectral data of Hami melon at three different speeds(0.1235m/s,0.15m/s and 0.19675m/s)were collected by this system,the collected online spectra of Hami melon were pretreated by means of multivariate scattering correction(MSC),standard normal variate transformation(SNV),derivative(SD)and their combination methods,the partial least squares(PLS)model was established to analyze the influence of different preprocessing methods on the modeling accuracy,comprehensive comparison of three kinds of PLS modeling speed of the results,it showed that the method of optimal pretreatment method combined with SNV and SD selection,optimal online spectrum acquisition speed was selected for 0.1235m/s.(3)According to the superior online speed,the SMLR model was established,reducing the number of bands as much as possible to provide support for the next rapid detection.The results showed that when the band number was between 10 and 15,the effect of the model was better when the number of bands was 15,the correlation coefficients rcv and cross validation root mean square error RMSECV were 0.6187 and 0.741 respectively.The 15 bands were chosen as 437.76 nm,476.33 nm,536.11 nm,580.47 nm,593.97 nm,619.04 nm,638.32 nm,659.54 nm,667.25 nm,673.03 nm,703.89 nm,752.10 nm,819.60 nm,865.88 nm,925.66 nm.(4)The feasibility of the near infrared online detection system for sugar content of Hami melon were verified by online test. |
PDF Full Text Request