CN103616346A - Method for rapidly detecting quality of ample flow pears - Google Patents

Abstract

The invention provides a method for rapidly detecting the quality of ample flow pears. The method comprises the following steps of: performing blank calibration on a polytetrafluoroethylene white board with flat surface and measuring the near-infrared reflection spectrum of the ample flow pears by using a JDSU near-infrared micro-spectrometer model 1700, and expressing the Y value of the spectrum in absorbancy; preferring the absorbancy values of the waveforms 1001.015nm, 1050.57nm, 1118.708nm, 1174.457nm, 1292.15nm, 1453.203nm, 1552.313nm, 1570.896nm and 1645.228nm, which are expressed as A1001.015, A1050.57, A118.708, A1174.457, A1292.15, A1453.203, A1552.313, A1570.896 and A1645.228, respectively, and then calculating through a special formula. The method provided by the invention is capable of detecting the SSC (Soluble Solids Content) in the ample flow pears without damaging the sample.

Description

Abundance of water pears quality rapid detection method

Technical field

The invention belongs near infrared spectrum and detect analysis, relate in particular to a kind of lossless detection method of the abundance of water pears soluble solid based on near-infrared spectral analysis technology.

Background technology

Cultivated area, the output of China pear tree all rank first in the world.In each planting fruit-trees of China, pear tree area, output, inferior to apple, citrus, occupy the 3rd.The crisp sweet succulence of the operatic circle, its pol and degree of ripeness, quality are closely related, and fruit pol that degree of ripeness is high is high, quality is high, and mouthfeel is good; Pol that degree of ripeness is low is low, poor quality, mouthfeel are poor, and pol is one of important indicator of evaluating its quality.Pol can obtain by measuring soluble solid (SSC) content.

Abundance of water pears, original producton location Japan, is one of three kinds of water pears of Japan.Single fruit on average weighs 240 grams, delicious, succulence, excellent taste.Abundance of water pears strong adaptability, equal well-grown in sandy loam, sandy soil and clayed ground.How China introduce the plantation of abundance of water pears, and as the Cangshan in Transition In Jurong, Jiangsu, Henan Zhumadian, Guan County, Shandong, Longkou City, Laixi City, Linyi, three beneficial villa gardens, Hubei Huang Po etc., as can be seen here, plantation region area is wide.The quality of abundance of water pears is except outside the Pass having with kind, also with the factor analysis such as geography and climate, soil.The field management such as fertilising in addition,, training, irrigation all have material impact to the pol of abundance of water pears.Illumination is also a factor that affects abundance of water pears pol, even same pear tree, the pol of different parts pears also has difference.

Transition In Jurong, Jiangsu belongs to mountain and hill area, the township of the fruit that suitable planting fruit tree ，Shi China is famous.The abundance of water pears quality that Jurong produces is better, and size is well-balanced, and pol is high.Jurong City agribusiness is just making great efforts to be forged into fine work.In order to ensure the quality of sell goods, normally pol is picked out to vanning higher than the pears of 13Birx.Yet in actual mechanical process, run into a problem, cannot guarantee that the pol of abundance of water pears in every case is all higher than 13Birx.If every case loads 12 abundance of water pears, as long as occur that 2～3 lower than 13Birx, client just thinks the poor quality of this pears, is reluctant to accept, thereby can causes amount and the price of selling to reduce, and agriculture enterprise income declines.How head it off becomes a difficult problem.

Near-infrared spectrum analysis has advantages of that cost is low, efficiency is high.Only need to obtain reflection or the transmitted spectrum of sample, through model, calculate, just can obtain the component content in material.The formation of near infrared spectrum is to come from hydric group in material the sum of fundamental frequencies of near infrared light and frequency multiplication are absorbed, spectrum peak overlapping, thereby there is not the high peak that middle infrared spectrum is such, present steamed bun peak.By adopting the method for Multivariate Correction, set up spectral model, can realize quantitative, the qualitative analysis of physical property.The method of modeling comprises multiple regression, multiple stepwise regression, principal component regression (PCR), partial least squares regression (PLSR).In these methods, best with the effect of PLSR, therefore also the most frequently used.200910116733.0) etc. such as patent of invention " method of rapidly detecting tea quality through near infrared technology " (application number: disclose the nearly red spectral fast detecting tea leaf quality of application.Patent of invention " a kind of method of differentiating Xiangshui County's rice with near-infrared spectrum technique " (application number: 201010033381.5) disclose application near infrared spectrum and carried out agricultural product original producton location knowledge method for distinguishing.China's technical paper " near-infrared diffuse reflectance is for detection of the research of apple sugar content and effective acidity " (periodical: spectroscopy and spectral analysis, in November, 2005, author: Liu Yande; Ying Yibin; Fu Xiaping), spectral range: 12500～4000cm ^-1, adopt full spectrum to detect the pol of apple.Technical paper " Primary Study of near-infrared transmission spectroscopic assay crystal pears pol " (periodical: food industry science and technology, the 3rd phase in 2007, author: Zhang Nan; Cheng Yu comes; Li Donghua etc.), wavelength coverage 643.26～928.35nm, adopts PLS method to set up the method for fast detecting crystal pears pol.

Although near infrared spectrum comprises a large amount of information, not every spectral variables all contains identical information, but some height, some is medium, some low or nothing.The spectral variables that does not contain information, is considered to noise conventionally, and these noises enter in model, can reduce the precision of prediction of model.Theoretical according to modern analysis and test, apply a small amount of spectral variables and set up model, improve on the one hand the precision of prediction, can increase on the other hand the robustness of model, adaptive faculty strengthens, and reduces the error rate of prediction.Patent of invention " multiband soil nitrogen pick-up unit and method based on near-infrared spectrum technique " (application number: the method that 201010297438.2) discloses the near infrared light fast detecting soil nitrogen of application 1683nm, 1516nm, 1407nm, 1306nm, 1199nm, 1080nm, 836nm wavelength for example.

For the abundance of water pears pol test problems based near infrared spectrum, obtaining a small amount of, to contain high information quantity near-infrared wavelength is the key of dealing with problems.

Summary of the invention

The technical matters solving: the invention provides a kind of abundance of water pears quality rapid detection method, do not destroying under the prerequisite of sample, can Non-Destructive Testing go out SSC content in abundance of water pears.

Technical scheme: abundance of water pears quality rapid detection method, step is: adopt miniature near infrared spectrometer 1700 types of JDSU, teflon blank with surfacing carries out blank calibration, measures the near-infrared reflection spectrum of abundance of water pears, and spectrum Y value represents in absorbance mode; Optimal wavelength is the absorbance under 1001.015nm, 1050.57nm, 1118.708nm, 1174.457nm, 1292.15nm, 1453.203nm, 1552.313nm, 1570.896nm, 1645.228nm, respectively with A _1001.015, A _1050.57, A _1118.708, A _1174.457, A _1292.15, A _1453.203, A _1552.313, A _1570.896and A _1645.228represent; The computing formula of the soluble solid of abundance of water pears (SSC) content is: SSC=A _1001.015* 294.26261+A _1050.57* 22.77511+A _1118.708* (174.32142)+A _1174.457* (457.0596)+A _1292.15* 311.57459+A _1453.203* 138.0716+A _1552.313* (3982.21870)+A _1570.896* 5129.33158+A _1645.228*-1276.30232+8.36941; The unit of SSC is Birx.

Beneficial effect: the present invention has the feature of near infrared spectrum fast detecting, is not destroying under the prerequisite of sample, can Non-Destructive Testing go out SSC content in abundance of water pears; This method has adopted less spectral variables, and the SSC content calculating by model is accurate, has reached the actual demand of producing.Owing to having adopted the less spectral variables modeling with a large amount of SSC information, thereby model is sane, does not have larger deviation.

Embodiment

Embodiment 1

(1) adopt the miniature near infrared spectrometer of JDSU (1700 type), the teflon of surfacing of take is reference, obtains spectrum, as with reference to spectrum, measures the near-infrared reflection spectrum of 10 abundance of water pears, and spectrum Y value represents in absorbance mode.

(2) obtaining wavelength is the absorbance under 1001.015nm, 1050.57nm, 1118.708nm, 1174.457nm, 1292.15nm, 1453.203nm, 1552.313nm, 1570.896nm, 1645.228nm, respectively with A _1001.015, A _1050.57, A _1118.708, A _1174.457, A _1292.15, A _1453.203, A _1552.313, A _1570.896and A _1645.228represent.

(3) computing formula of the soluble solid of abundance of water pears (SSC) content is:

SSC=A _1001.015×294.26261+A _1050.57×22.77511+A _1118.708×（-174.32142）+A _1174.457×（-457.0596）+A _1292.15×311.57459+A _1453.203×138.0716+A _1552.313×（-3982.21870）+A _1570.896×5129.33158+A _1645.228×-1276.30232+8.36941

Testing result is as shown in table 1, with prediction, represents.Adopt the method diminishing to get pear juice simultaneously, utilize Abbe refractometer to detect pol, result is with actual measurement expression, and difference is between the two error.Coefficient R=0.951 of actual measurement and prediction, root-mean-square error RMSPE=0.473.The higher expression accuracy of detection of R value is higher, and the less expression accuracy of detection of RMSPE is higher.

Wherein the computing formula of coefficient R is:

$R=\sqrt{(1-\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{(y_i-{y^{\text{'}}}_i)}^2}{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{(y_i-y_m)}^2})}$

N-sample number;

Y _ithe predicted value of the-the i sample;

Y' _ithe actual measurement chemical score of the-the i sample;

Y _mthe mean value of-all samples actual measurement chemical score.

$\mathrm{RMSEP}=\sqrt{\frac{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{(y_i-{\hat{y}}_i)}^2}{n}}$

In formula, y _iwith

be respectively actual measurement collection and the forecast set of i sample in forecast set, n is forecast set sample number.

The a collection of abundance of water pears of table 1 testing result

Abundance of water pears sequence number	Actual measurement	Prediction	Error
				1	11.25	11.03	0.22
2	11.50	11.14	0.36
				3	12.25	11.74	0.51
4	13.00	13.30	-0.30
				5	13.10	13.47	-0.37
6	13.75	13.34	0.41
				7	14.25	14.70	-0.45
8	14.50	14.05	0.45
				9	14.75	14.11	0.64
10	14.80	15.46	-0.66

With full wave 125 spectroscopic datas, set up PLS model, predict the outcome not good, coefficient R=0.80, root-mean-square error RMSPE=1.173.

Employing, without information variable elimination-offset minimum binary (UVE-PLS) method, preferentially goes out 32 spectral variables, and carries out PLS modeling, coefficient R=0.850 of prediction, root-mean-square error RMSPE=0.787.

Adopt genetic algorithm, optimize 24 spectral variables, set up PLS model, coefficient R=0.843 of prediction, root-mean-square error RMSPE=0.823.

Optimize with full spectrum of wavelengths and UVE-PLS, genetic algorithm the model that spectral variables sets up and compare, the method that the application provides has obtained unforeseeable effect, has very high detection pol.

The present invention is that the miniature near infrared spectrometer (1700 type) based on JDSU company is quick, the method for Non-Destructive Testing abundance of water pears pol.Spectral wavelength scope 908.1nm～1676.2nm, has 125 wavelength variablees, and because resolution is low, spectral information is overlapping serious, therefore, picks out the spectral variables modeling that quantity of information is high, and for improving, precision of prediction is significant.The invention is characterized in and only apply 9 spectroscopic datas under specific near-infrared wavelength, set up mathematical model.By repeatedly testing, this model is higher than the precision of prediction of the long drag of all-wave, meets the demand of production practices completely.

Claims (1)

1. abundance of water pears quality rapid detection method, it is characterized in that step is: adopt miniature near infrared spectrometer 1700 types of JDSU, teflon blank with surfacing carries out blank calibration, measures the near-infrared reflection spectrum of abundance of water pears, and spectrum Y value represents in absorbance mode; Optimal wavelength is the absorbance under 1001.015nm, 1050.57nm, 1118.708nm, 1174.457nm, 1292.15nm, 1453.203nm, 1552.313nm, 1570.896nm, 1645.228nm, respectively with A _1001.015, A _1050.57, A _1118.708, A _1174.457, A _1292.15, A _1453.203, A _1552.313, A _1570.896and A _1645.228represent; The computing formula of the soluble solid of abundance of water pears (SSC) content is: SSC=A _1001.015* 294.26261+A _1050.57* 22.77511+A _1118.708* (174.32142)+A _1174.457* (457.0596)+A _1292.15* 311.57459+A _1453.203* 138.0716+A _1552.313* (3982.21870)+A _1570.896* 5129.33158+A _1645.228*-1276.30232+8.36941; The unit of SSC is Birx.