CN107643264A

CN107643264A - A kind of wheat quality near infrared detection system

Info

Publication number: CN107643264A
Application number: CN201610578779.4A
Authority: CN
Inventors: 由国峰
Original assignee: Individual
Current assignee: Individual
Priority date: 2016-07-21
Filing date: 2016-07-21
Publication date: 2018-01-30

Abstract

Detect wheat quality existing quick nondestructive, invented a kind of wheat quality near infrared detection system, tested the accuracy, repeatability and stability of the system, the modeling result R2 of MPA spectrometers is 95.99%, RESECV 0.293, RDP 5；The R2 of forecast set checking model is 98.31%, RMSEP 0.165.Experiment shows：Model obtained by wheat quality near infrared detection system has good predictability, stability and repeatability；Gained spectral wavelength has reappearance with absorbance；Its model is better than individual spectrum to the prediction effect of averaged spectrum；The system working stability, function admirable, can be applied to wheat quality quality testing.

Description

A kind of wheat quality near infrared detection system

Art

The present invention relates to a kind of detecting system, more particularly to a kind of wheat quality near infrared detection system.

Background technology

Quick detection wheat quality quality is advantageous to the breeding, trade and processing of wheat, is of great immediate significance.Closely The theoretical foundation foot of infra-red sepectrometry records the almost all such as C mono- H, O-H, N-H in sample by the near infrared spectrum of sample The information of hydric group.Compared to traditional analysis, the quality of wheat can easily and fast, be nondestructively detected, is national mark Wheat quality detection method as defined in standard.

At present, quick, easy to operate, cheap special near infrared spectrometer is new as the development of grain quality quick detection Direction.External many well-known spectral instrument companies have produced the special cereal spectrometric instrument of maturation, as Zeltex is public ZX-50SRT type Portable near infrared seed quality analyzers of department etc..The domestic research in the field lags far behind.

The content of the invention

The purpose of the present invention is to detect wheat quality with showing quick nondestructive, devises a kind of wheat quality near infrared detection system System.

The technical solution adopted for the present invention to solve the technical problems is：

Wheat quality near infrared detection system is by fixed grating optical system, CCD drivings and signal acquiring system, CCD temperature controls The parts such as system processed, light-source control system, rotation bench control system, computer software form.

The CCD temperature control systems control accuracy of described system is ± 0.1 DEG C, and the control accuracy of light intensity is 0.02%, Instrument signal to noise ratio is 4000:1.

Described Systematic selection PLS modeling, crosscheck method are examined to determine best model, finally with pre- Survey collection and evaluate its estimated performance.Evaluation model, which calibrates effect and the index of predictive ability, R2, RMSECV, RMSEP, RPD.

Described R2 represents the percentage of variable occur in true component value, and its value is closer to 1, and prediction content is closer to true Value, the predictive ability of model are better.

Described RMSECV and RMSEP is illustrated respectively in predicted value and standard value in the case of cross validation and external certificate Between departure degree, for the quality of judgment models, its value is smaller, and model is better.

PD described in R is used for verifying the stability and predictive ability of model, works as RPD>When 3, model stability and predictability Can be good.

The beneficial effects of the invention are as follows：

Wheat quality near infrared detection system has good predictability, repeatability and stability, can apply to wheat quality Quantitative detection with analysis.

Brief description of the drawings

The present invention is further described with reference to the accompanying drawings and examples.

Fig. 1 is system construction drawing.

Fig. 2 is near infrared detection network analysis flow.

Embodiment

As shown in figure 1, wheat quality near infrared detection system is by fixed grating optical system, CCD drivings and signal acquisition system The part such as system, CCD temperature control systems, light-source control system, rotation bench control system, computer software forms.System CCD temperature control systems control accuracy be ± 0.1 DEG C, the control accuracy of light intensity is 0.02%, instrument signal to noise ratio 4000:1. The operation principle of system is：Light-source control system flows through the current stability of light source to keep the steady of light source intensity by control It is fixed, stable near infrared light is provided for system.Light is imported after optically focused collimates by optical fiber, shines directly into the sample of turntable And pass through sample.Light after diffusing transmission is exported by optical fiber, again after optically focused collimation, penetrates the slit of beam splitting system, and reflected Dispersion on to fixed grating, and monochromatic light is formed, finally it is reflected on Linear CCD Detector, forms continuous near infrared light Spectrum.Programmable logic design provides pulse sequence driving ccd detector and converts optical signals into electric signal, by a/d converter Gather and be transferred to PLD, then be transferred to host computer.Last upper computer software system is preserved and handled to spectral signal.This System uses single-chip microcomputer and PLD as control centre, the a/d converter and D/A converter of external 16, completes the inside of the system The control of circuit, and realize the communication with host computer.

As shown in Fig. 2 the main flow of near infrared spectroscopic method includes establishing model and prediction sample, basic step is such as Under.(1) representational modeling sample is screened, content to be measured is measured using National Standard Method, as modeling standard value, and with closely red External spectrum instrument scans to obtain sample spectra.(2) chemo metric software is used, selects suitable pretreatment and modeling method processing Spectrum, establish the NIR Spectroscopy Analysis Model of function admirable.(3) near infrared spectrum of testing sample is scanned, and uses model Analyzed, obtain the component content to be measured of testing sample.For wheat nutritional ingredient species more than and sample size be less than build The situation of moding amount number, PLS modeling is selected, crosscheck method is examined to determine best model, finally with prediction Collection evaluates its estimated performance.Evaluation model, which calibrates effect and the index of predictive ability, R2, RMSECV, RMSEP, RPD.R2 is represented There is the percentage of variable in true component value, its value closer to 1, closer to true value, the predictive ability of model get over by prediction content It is good；RMSECV and RMSEP is illustrated respectively in the deviation between the predicted value and standard value in the case of cross validation and external certificate Degree, for the quality of judgment models, its value is smaller, and model is better；RPD is used for verifying the stability and predictive ability of model, Work as RPD>When 3, model stability and estimated performance are good.

Claims

1. a kind of wheat quality near infrared detection system is by fixed grating optical system, CCD drivings and signal acquiring system, CCD The parts such as temperature control system, light-source control system, rotation bench control system, computer software form.

2. wheat quality near infrared detection system according to claim 1, it is characterized in that the CCD temperature controls of described system System control accuracy processed is ± 0.1 DEG C, and the control accuracy of light intensity is 0.02%, instrument signal to noise ratio 4000:1.

3. wheat quality near infrared detection system according to claim 1, it is characterized in that described Systematic selection is partially minimum Square law is modeled, and crosscheck method is examined to determine best model, finally evaluates its estimated performance with forecast set；Evaluation model is determined The index of mark effect and predictive ability has R2, RMSECV, RMSEP, RPD.

4. wheat quality near infrared detection system according to claim 1, it is characterized in that described R2 represents true component Occurs the percentage of variable in value, for its value closer to 1, prediction content is better closer to true value, the predictive ability of model.

5. wheat quality near infrared detection system according to claim 1, it is characterized in that described RMSECV and RMSEP points The departure degree between the predicted value and standard value in the case of cross validation and external certificate is not represented, for judgment models Quality, its value is smaller, and model is better.

6. wheat quality near infrared detection system according to claim 1, it is characterized in that the PD described in R is used for verifying model Stability and predictive ability, work as RPD>When 3, model stability and estimated performance are good.