CN103743705A - Rapid detection method for sorghum halepense and similar species - Google Patents
Rapid detection method for sorghum halepense and similar species Download PDFInfo
- Publication number
- CN103743705A CN103743705A CN201410042766.6A CN201410042766A CN103743705A CN 103743705 A CN103743705 A CN 103743705A CN 201410042766 A CN201410042766 A CN 201410042766A CN 103743705 A CN103743705 A CN 103743705A
- Authority
- CN
- China
- Prior art keywords
- near infrared
- spectrum data
- original spectrum
- sorghum
- quick
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses a rapid detection method for sorghum halepense and similar species based on a near infrared spectrum. The method comprises the following steps: adopting the near infrared spectrum to obtain spectral reflectance characteristic curves of three kinds of sorghum seeds of sorghum, sorghum halepense and sorghum sudanense between 325-1075 nm, adopting a partial least squares method for pattern character analysis, and adopting a cross validation method for judgment to determine that the optimal number of principal components is 9; after character extraction is completed, using the 9 kinds of principal components as the input variable of neural network, establishing three layers of BP (Back Propagation) neural networks, and calculating the category of each seed. The method combines the near infrared spectrum analysis with a chemical measurement method to rapidly and accurately judge the category of the three kinds of sorghum seeds, the detection time is greatly shortened, the detection efficiency is improved, and the detection cost is reduced.
Description
Technical field
The present invention relates to inspection and quarantining for import/export field, particularly the method for quick of a kind of false Chinese sorghum and approximate species thereof.
Background technology
Sorghum (Sorghum) has various plants to be widely cultivated as grain, feed and each arid area, the alive boundary of industrial raw material.But in sorghum, also have quite a few to belong to instruction plant or have the kind of invasion property, wherein the most representative have a false Chinese sorghum (S.halepense (L.Pers)).False jowar is one of the world's ten large malignant weeds, and it can be bred jointly with root-like stock and seed, destroys China's vegetation diversity ability strong, and it endangers greatly, breeding is fast, difficult control, is the inward plant quarantine harmful organism of China.False Chinese sorghum former only in Fujian, Guangdong, Taiwan introduces a fine variety distribution, but along with the continuous increase of food import quantity, in Hainan, province such as more than 10, Guangxi etc. and municipality directly under the Central Government's report once had the distribution of false Chinese sorghum.Chinese sorghum (Sorghumbicolor), black jowar (S.almum) and sudangrass (S.sudanense) are the sibling specieses of false Chinese sorghum, quite similar with false Chinese sorghum in form, although form is approximate, but quarantine status is not quite similar, black jowar belongs to invasion property kind, jowar and sudangrass belong to cultivating economic crop and can be used as herbage, do not show invading property of people.Therefore, correctly identify sorghum plant, on port quarantine, there is significant application value.The existing evaluation sorting technique for sorghum plant is to distinguish based on cytology method and DNA detection, these methods length consuming time, and operating process is complicated.The invention provides a kind of fast and convenient harmless analytical approach differentiates the plant of sorghum.
" fingerprint " characteristic that infrared spectrum technology has with it, is more and more subject to people's attention in recent years.Infra-red sepectrometry is a kind of fast and convenient harmless analytical approach, can reflect that frequency multiplication and sum of fundamental frequencies that the bases such as organic principle, particularly C-H, O-H, N-H of material inside close absorb, and can be used for the content of quantitative measurement material organic substance.Infra-red sepectrometry has been widely used in the industries such as agricultural, food, feed, medicine, petrochemical complex.Lot of domestic and international scholar utilizes near-infrared spectrum technique to carry out the discriminating of material kind.In the research of sorghum seed fast detecting, have a wide range of applications potentiality.
Summary of the invention
The problem existing in order to solve existing port quarantine detection technique, the invention provides a kind of fast and convenient harmless classification authentication method similar sorghum seed is differentiated, this invention has significant application value on port quarantine.
False Chinese sorghum based near infrared spectrum and a method for quick for approximate species thereof, comprise the steps:
1) utilize the original spectrum data of near infrared spectrometer collection known sample, and to the original spectrum data pre-service collecting;
2) use genetic algorithm to extract the original spectrum data of 15 characteristic wavelengths of original spectrum data;
3) the original spectrum data of 15 characteristic wavelengths that step (2) extracted use partial least square method to carry out pattern feature analysis, complete feature extraction, finally set up three layers of BP neural network again through validation-cross diagnostic method successively;
4) three layers of BP neural network that adopt step (3) to set up are predicted unknown sample.
The wavelength coverage of described collection original spectrum data is 325-1075nm.
Described known sample is all 40-50mm with diameter, the double dish splendid attire of height 8-10mm, and each double dish is filled as an experiment sample.
Described near infrared spectrometer be placed in known sample directly over, apart from known sample surface 90-100mm; Analytical spectra district has adopted the sweep interval of part to analyze, and adopts the data of 500-750nm scope, and spaced points is 3, in interval every three wavelength points of 500-750nm, chooses a data point.
The pretreated method of described original spectrum data is followed successively by: first original spectrum data are carried out to SavitzkyGolayDerivatives processing; Adopt Savitzky-Golay smoothing method to process original spectrum data, selecting smoothly counts is 9 again; Finally original spectrum data are carried out to SNV processing.Use genetic algorithm to extract 15 characteristic wavelengths respectively: 543nm, 378nm, 747nm, 817nm, 730nm, 545nm, 610nm, 663nm, 915nm, 676nm, 582nm, 506nm, 756nm, 459nm, 446nm.
Described BP neural network is error back propagation network, and the method for setting up BP neural network is as follows: first adopt principal component analysis (PCA) to compress and dimensionality reduction the original spectrum data of 15 characteristic wavelengths that extract, the major component obtaining is inputted as BP neural network; BP neural network divides 3 layers to be input layer, hidden layer and output layer, adopts sigmoid excitation function, and uses improved BP algorithm---LevenberMarquardt method.
Described network input layer, hidden layer, output layer nodes are respectively 7,7,1, and wherein 7 of input layer nodes are analyzed the major component obtaining from PLS; Minimum training speed is 0.1, and training iterations is 1000 times.
Compared with prior art, the invention has the beneficial effects as follows: the existing evaluation sorting technique for sorghum plant is to distinguish based on cytology method and DNA detection, these methods length consuming time, operating process is complicated.The present invention's application near-infrared spectrum technique extracts 15 characteristic wavelengths in conjunction with characteristic wavelength extracting method, the characteristic wavelength of extraction respectively: 543nm, 378nm, 747nm, 817nm, 730nm, 545nm, 610nm, 663nm, 915nm, 676nm, 582nm, 506nm, 756nm, 459nm, 446nm, has set up the model that sorghum category is differentiated, this model is to the Relative Error of unknown sample all below 4%, and accuracy of identification is high, and the prediction effect of this model can meet application request.Illustrate use near infrared spectrum can be fast, accurately sorghum category is differentiated.
Accompanying drawing explanation
Fig. 1 is system chart of the present invention;
Fig. 2 is modeling collection sample classification result scatter diagram of the present invention.
Embodiment
As shown in Figure 1, a kind of false Chinese sorghum based near infrared spectrum of the present invention and the method for quick of approximate species thereof, comprise the steps:
1) utilize the original spectrum data of near infrared spectrometer collection known sample, and to the original spectrum data pre-service collecting;
2) use genetic algorithm to extract the original spectrum data of 15 characteristic wavelengths of original spectrum data;
3) the original spectrum data of 15 characteristic wavelengths that step (2) extracted use partial least square method to carry out pattern feature analysis, complete feature extraction, finally set up three layers of BP neural network again through validation-cross diagnostic method successively;
4) three layers of BP neural network that adopt step (3) to set up are predicted unknown sample.
The wavelength coverage of described collection original spectrum data is 325-1075nm.
Described known sample is all 40-50mm with diameter, the double dish splendid attire of height 8-10mm, and each double dish is filled as an experiment sample.
Described near infrared spectrometer be placed in known sample directly over, apart from known sample surface 90-100mm; Analytical spectra district has adopted the sweep interval of part to analyze, and adopts the data of 500-750nm scope, and spaced points is 3, in interval every three wavelength points of 500-750nm, chooses a data point.
The pretreated method of described original spectrum data is followed successively by: first original spectrum data are carried out to Savitzky GolayDerivatives processing; Adopt Savitzky-Golay smoothing method to process original spectrum data, selecting smoothly counts is 9 again; Finally original spectrum data are carried out to SNV processing.
Use genetic algorithm to extract 15 characteristic wavelengths respectively: 543nm, 378nm, 747nm, 817nm, 730nm, 545nm, 610nm, 663nm, 915nm, 676nm, 582nm, 506nm, 756nm, 459nm, 446nm.
Described BP neural network is error back propagation network, and the method for setting up BP neural network is as follows: first adopt principal component analysis (PCA) to compress and dimensionality reduction the original spectrum data of 15 characteristic wavelengths that extract, the major component obtaining is inputted as BP neural network; BP neural network divides 3 layers to be input layer, hidden layer and output layer, adopts sigmoid excitation function, and uses improved BP algorithm---LevenberMarquardt method.
Described network input layer, hidden layer, output layer nodes are respectively 7,7,1, and wherein 7 of input layer nodes are analyzed the major component obtaining from PLS; Minimum training speed is 0.1, and training iterations is 1000 times.
Embodiment
Acquisition target is Chinese sorghum, false Chinese sorghum, three kinds of sorghum seeds of sudangrass.This detection method is also applicable to other seed of sorghum.
The spectroscopic data of acquisition range between 325-1075nm, spectrum sample is spaced apart 1.5nm, scanning times 30 times, probe field angle is 20 degree.Light source adopts the 14.5V Halogen lamp LED supporting with spectrometer.
Three kinds of sorghum seeds are totally 120 samples (40 of each kinds).Various sample standard deviation diameters are 40-50mm, the double dish splendid attire of height 8-10mm.Equidistant in order to reduce experimental error assurance testee and instrument, each double dish is filled as an experiment sample.Each kind is respectively made 40 samples, amounts to 120 samples.All experiments were sample is divided into modeling collection and forecast set at random, and modeling collection has 90 samples (30 of each kinds), and forecast set has 30 samples (10 of each kinds).Spectrometer is tested after whiteboard calibration.Spectrometer is placed in the top of sample, apart from rice surface 90-100mm, to each scan sample 20 times, averages.
In order to remove the impacts such as random noise, baseline wander, light scattering, sample be inhomogeneous, need to carry out pre-service to the spectroscopic data collecting.First data are carried out to SavitzkyGolayDerivatives single order differential and process, smoothly count and be set to 6, in order to remove the irrelevant drift of co-wavelength.Adopt again MovingAverageSmoothing smoothing method, smoothly count and be set to 9, in order to improve the signal to noise ratio (S/N ratio) of analytic signal, the high frequency noise that filtering various factors produces effectively.Finally data are carried out to polynary scatter correction processing, MSC can remove the mirror-reflection of sample in near-infrared diffuse reflection spectrum and the noise that unevenness causes, and eliminates the not repeated of the baseline of diffuse reflection spectrum and spectrum.Spectroscopic data after processing is as the input of genetic algorithm.
Genetic algorithm control setting parameter: initial population 100, genetic iteration number of times 50, crossover probability 0.8, variation probability 0.1.Pretreated spectroscopic data is carried out to wavelength screening with GA, extract altogether 15 characteristic wavelengths as PLS input variable.7 major components above that draw through principal component analysis (PCA) have comprised most of spectral information.Therefore, the input variable using these 7 characteristic variables as BP neural network, network input layer, hidden layer, output layer nodes are respectively 7,7,1.Minimum training speed is 0.1, and setting training iterations is 1000 times, and input sample is carried out to standardization.Training set and forecast set sample number are 90 and 30.Predicting the outcome of 30 unknown sample shown to this 3 kinds, fit residual error and be the prediction relative deviation of 0.0013,30 sample all below 4%, the accuracy of prediction can meet actual requirement.
As shown in Figure 2, application near-infrared spectrum technique has been set up the model that sorghum category is differentiated, the prediction effect of this model can meet application request, and to the Relative Error of unknown sample, all below 4%, accuracy of identification is high.Illustrate use near infrared spectrum can be fast, accurately sorghum category is differentiated.Adopt genetic algorithm to extract characteristic wavelength, and carry out principal component analysis (PCA) by partial least square method, finally in conjunction with BP neural network, forecast sample is predicted, accuracy of identification has obtained large increase.The major component that employing obtains from PLS analyzes, as the input of BP neural network, has not only reduced calculated amount, has accelerated training speed, simultaneously because removed spectrum interfere information, has also improved the accuracy of prediction.Therefore, application genetic algorithm, partial least square method can quick and precisely be differentiated sorghum category in conjunction with mode identification method and the near-infrared spectrum technique of BP neural network.
Claims (8)
1. the false Chinese sorghum based near infrared spectrum and a method for quick for approximate species thereof, its feature comprises the steps:
1) utilize the original spectrum data of near infrared spectrometer collection known sample, and to the original spectrum data pre-service collecting;
2) use genetic algorithm to extract the original spectrum data of 15 characteristic wavelengths of original spectrum data;
3) the original spectrum data of 15 characteristic wavelengths that step (2) extracted use partial least square method to carry out pattern feature analysis, complete feature extraction, finally set up three layers of BP neural network again through validation-cross diagnostic method successively;
4) three layers of BP neural network that adopt step (3) to set up are predicted unknown sample.
2. a kind of false Chinese sorghum based near infrared spectrum according to claim 1 and the method for quick of approximate species thereof, is characterized in that: the wavelength coverage of described collection original spectrum data is 325-1075nm.
3. a kind of false Chinese sorghum based near infrared spectrum according to claim 1 and the method for quick of approximate species thereof, it is characterized in that: described known sample is all 40-50mm with diameter, the double dish splendid attire of height 8-10mm, each double dish is filled as an experiment sample.
4. a kind of false Chinese sorghum based near infrared spectrum according to claim 1 and the method for quick of approximate species thereof, is characterized in that: described near infrared spectrometer be placed in known sample directly over, apart from known sample surface 90-100mm; Analytical spectra district has adopted the sweep interval of part to analyze, and adopts the data of 500-750nm scope, and spaced points is 3, in interval every three wavelength points of 500-750nm, chooses a data point.
5. a kind of false Chinese sorghum based near infrared spectrum according to claim 1 and the method for quick of approximate species thereof, is characterized in that: the pretreated method of described original spectrum data is followed successively by: first original spectrum data are carried out to SavitzkyGolayDerivatives processing; Adopt Savitzky-Golay smoothing method to process original spectrum data, selecting smoothly counts is 9 again; Finally original spectrum data are carried out to SNV processing.
6. a kind of false Chinese sorghum based near infrared spectrum according to claim 1 and the method for quick of approximate species thereof, is characterized in that: use genetic algorithm to extract 15 characteristic wavelengths respectively: 543nm, 378nm, 747nm, 817nm, 730nm, 545nm, 610nm, 663nm, 915nm, 676nm, 582nm, 506nm, 756nm, 459nm, 446nm.
7. a kind of false Chinese sorghum based near infrared spectrum according to claim 1 and the method for quick of approximate species thereof, it is characterized in that: described BP neural network is error back propagation network, the method of setting up BP neural network is as follows: first adopt principal component analysis (PCA) to compress and dimensionality reduction the original spectrum data of 15 characteristic wavelengths that extract, the major component obtaining is inputted as BP neural network; BP neural network divides 3 layers to be input layer, hidden layer and output layer, adopts sigmoid excitation function, and uses improved BP algorithm---LevenberMarquardt method.
8. a kind of false Chinese sorghum based near infrared spectrum according to claim 7 and the method for quick of approximate species thereof, it is characterized in that: described network input layer, hidden layer, output layer nodes are respectively 7,7,1, wherein 7 of input layer nodes are analyzed the major component obtaining from PLS; Minimum training speed is 0.1, and training iterations is 1000 times.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410042766.6A CN103743705A (en) | 2014-01-28 | 2014-01-28 | Rapid detection method for sorghum halepense and similar species |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410042766.6A CN103743705A (en) | 2014-01-28 | 2014-01-28 | Rapid detection method for sorghum halepense and similar species |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103743705A true CN103743705A (en) | 2014-04-23 |
Family
ID=50500743
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410042766.6A Pending CN103743705A (en) | 2014-01-28 | 2014-01-28 | Rapid detection method for sorghum halepense and similar species |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103743705A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104062262A (en) * | 2014-07-09 | 2014-09-24 | 中国科学院半导体研究所 | Crop seed variety authenticity identification method based on near infrared spectrum |
| CN106404710A (en) * | 2015-07-15 | 2017-02-15 | 重庆医科大学 | Pharmaceutical powder auxiliary material near-infrared spectrum rapid non-destructive identification method |
| CN104374738B (en) * | 2014-10-30 | 2017-03-08 | 中国科学院半导体研究所 | A kind of method for qualitative analysis improving identification result based on near-infrared |
| CN106610377A (en) * | 2016-11-14 | 2017-05-03 | 北京农业信息技术研究中心 | Seed spectral detection method and system |
| CN107024446A (en) * | 2016-01-29 | 2017-08-08 | 九芝堂股份有限公司 | A kind of Liuwei Dihuang Wan small honey pill crude drug powder multiple index quick detecting method |
| CN115047129A (en) * | 2022-06-16 | 2022-09-13 | 贵州茅台酒股份有限公司 | Sorghum variety identification method based on volatile substance composition characteristics |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002090299A (en) * | 2000-09-11 | 2002-03-27 | Opt Giken Kk | Method for distinguishing grade of high-molecular material |
| CN101393122A (en) * | 2008-10-31 | 2009-03-25 | 中国农业大学 | Honey quality rapid detection method |
| CN101738373A (en) * | 2008-11-24 | 2010-06-16 | 中国农业大学 | Method for distinguishing varieties of crop seeds |
| CN102830087A (en) * | 2011-09-26 | 2012-12-19 | 武汉工业学院 | Method for quickly identifying food waste oils based on near infrared spectroscopy |
-
2014
- 2014-01-28 CN CN201410042766.6A patent/CN103743705A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002090299A (en) * | 2000-09-11 | 2002-03-27 | Opt Giken Kk | Method for distinguishing grade of high-molecular material |
| CN101393122A (en) * | 2008-10-31 | 2009-03-25 | 中国农业大学 | Honey quality rapid detection method |
| CN101738373A (en) * | 2008-11-24 | 2010-06-16 | 中国农业大学 | Method for distinguishing varieties of crop seeds |
| CN102830087A (en) * | 2011-09-26 | 2012-12-19 | 武汉工业学院 | Method for quickly identifying food waste oils based on near infrared spectroscopy |
Non-Patent Citations (3)
| Title |
|---|
| 岑海燕等: "基于光谱技术的杨梅汁品种快速鉴别方法的研究", 《光谱学与光谱分析》, vol. 27, no. 3, 31 March 2007 (2007-03-31), pages 503 - 506 * |
| 林萍等: "基于可见近红外光谱的糖类别快速鉴别研究", 《光谱学与光谱分析》, vol. 29, no. 2, 28 February 2009 (2009-02-28), pages 383 - 385 * |
| 赵弘等: "基于Levenberg-Marquardt算法的神经网络监督控制", 《西安交通大学学报》, vol. 36, no. 5, 31 May 2002 (2002-05-31) * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104062262A (en) * | 2014-07-09 | 2014-09-24 | 中国科学院半导体研究所 | Crop seed variety authenticity identification method based on near infrared spectrum |
| CN104374738B (en) * | 2014-10-30 | 2017-03-08 | 中国科学院半导体研究所 | A kind of method for qualitative analysis improving identification result based on near-infrared |
| CN106404710A (en) * | 2015-07-15 | 2017-02-15 | 重庆医科大学 | Pharmaceutical powder auxiliary material near-infrared spectrum rapid non-destructive identification method |
| CN107024446A (en) * | 2016-01-29 | 2017-08-08 | 九芝堂股份有限公司 | A kind of Liuwei Dihuang Wan small honey pill crude drug powder multiple index quick detecting method |
| CN106610377A (en) * | 2016-11-14 | 2017-05-03 | 北京农业信息技术研究中心 | Seed spectral detection method and system |
| CN106610377B (en) * | 2016-11-14 | 2019-11-15 | 北京农业信息技术研究中心 | Seed spectral method of detection and system |
| CN115047129A (en) * | 2022-06-16 | 2022-09-13 | 贵州茅台酒股份有限公司 | Sorghum variety identification method based on volatile substance composition characteristics |
| CN115047129B (en) * | 2022-06-16 | 2023-07-04 | 贵州茅台酒股份有限公司 | Sorghum variety identification method based on volatile matter composition characteristics |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103743705A (en) | Rapid detection method for sorghum halepense and similar species | |
| Wang et al. | Technology using near infrared spectroscopic and multivariate analysis to determine the soluble solids content of citrus fruit | |
| Zimmerman et al. | Analysis of allergenic pollen by FTIR microspectroscopy | |
| CN103234922A (en) | Rapid soil organic matter detection method based on large sample soil visible-near infrared spectrum classification | |
| CN103134765A (en) | Chinese medicine sample authenticity preliminary screening method based on terahertz time-domain spectrum | |
| CN102937575B (en) | Watermelon sugar degree rapid modeling method based on secondary spectrum recombination | |
| CN104376325A (en) | Method for building near-infrared qualitative analysis model | |
| CN112669915B (en) | Pear nondestructive testing method based on neural network and near infrared spectrum | |
| CN104215591A (en) | Damage-free visible-near infrared light spectrum detecting method | |
| CN106124445A (en) | A kind of quick, Undamaged determination genetically engineered soybean method | |
| CN104374739A (en) | Identification method for authenticity of varieties of seeds on basis of near-infrared quantitative analysis | |
| CN105139022A (en) | Plastic identification model establishment method via near-infrared hyper-spectral image technology | |
| CN105740898A (en) | Method for establishing classification models by utilizing spectral feature vectors and principal component analysis | |
| KR20220038602A (en) | Methods and systems for identifying microorganisms | |
| Yun et al. | Identification of tea based on CARS‐SWR variable optimization of visible/near‐infrared spectrum | |
| CN105717066A (en) | Near-infrared spectrum recognition model based on weighting association coefficients | |
| CN107328735A (en) | Rape species discrimination method based on terahertz light spectral technology | |
| CN111537469A (en) | Apple quality rapid nondestructive testing method based on near-infrared technology | |
| CN102128805A (en) | Method and device for near infrared spectrum wavelength selection and quick quantitative analysis of fruit | |
| Wang et al. | Intelligent detection of hard seeds of snap bean based on hyperspectral imaging | |
| CN102135496A (en) | Infrared spectrum quantitative analysis method and infrared spectrum quantitative analysis device based on multi-scale regression | |
| CN105911000A (en) | Characteristic wave band based blood spot egg on-line detection method | |
| CN113030011A (en) | Rapid nondestructive testing method and system for sugar content of fruits | |
| CN102967557B (en) | Biological aerosol water-content determination based on near infrared spectroscopy and classification method thereof | |
| Liu et al. | Research on the online rapid sensing method of moisture content in famous green tea spreading |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20140423 |