CN105486655B - The soil organism rapid detection method of model is intelligently identified based on infrared spectroscopy - Google Patents
The soil organism rapid detection method of model is intelligently identified based on infrared spectroscopy Download PDFInfo
- Publication number
- CN105486655B CN105486655B CN201510816704.0A CN201510816704A CN105486655B CN 105486655 B CN105486655 B CN 105486655B CN 201510816704 A CN201510816704 A CN 201510816704A CN 105486655 B CN105486655 B CN 105486655B
- Authority
- CN
- China
- Prior art keywords
- sample
- model
- tested
- soil
- spectrum
- 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.)
- Active
Links
- 239000002689 soil Substances 0.000 title claims abstract description 57
- 238000001514 detection method Methods 0.000 title claims abstract description 12
- 238000004566 IR spectroscopy Methods 0.000 title claims abstract description 10
- 238000000034 method Methods 0.000 claims abstract description 57
- 238000001228 spectrum Methods 0.000 claims abstract description 52
- 239000000126 substance Substances 0.000 claims abstract description 23
- 238000012549 training Methods 0.000 claims abstract description 18
- 239000011159 matrix material Substances 0.000 claims abstract description 16
- 239000005416 organic matter Substances 0.000 claims abstract description 15
- 230000008569 process Effects 0.000 claims abstract description 10
- 230000003595 spectral effect Effects 0.000 claims abstract description 9
- 238000009614 chemical analysis method Methods 0.000 claims abstract description 6
- 238000010238 partial least squares regression Methods 0.000 claims description 19
- 238000004611 spectroscopical analysis Methods 0.000 claims description 14
- 238000011156 evaluation Methods 0.000 claims description 8
- 239000000835 fiber Substances 0.000 claims description 8
- 238000004476 mid-IR spectroscopy Methods 0.000 claims description 8
- KMUONIBRACKNSN-UHFFFAOYSA-N potassium dichromate Chemical compound [K+].[K+].[O-][Cr](=O)(=O)O[Cr]([O-])(=O)=O KMUONIBRACKNSN-UHFFFAOYSA-N 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 239000011521 glass Substances 0.000 claims description 6
- 230000015572 biosynthetic process Effects 0.000 claims description 5
- 238000005457 optimization Methods 0.000 claims description 5
- 238000004867 photoacoustic spectroscopy Methods 0.000 claims description 5
- 238000005070 sampling Methods 0.000 claims description 5
- 238000003786 synthesis reaction Methods 0.000 claims description 5
- 238000004737 colorimetric analysis Methods 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 4
- 230000036571 hydration Effects 0.000 claims description 4
- 238000006703 hydration reaction Methods 0.000 claims description 4
- 238000003705 background correction Methods 0.000 claims description 3
- 239000006229 carbon black Substances 0.000 claims description 3
- 238000002790 cross-validation Methods 0.000 claims description 3
- 230000001174 ascending effect Effects 0.000 abstract 1
- 238000001914 filtration Methods 0.000 description 8
- 238000001834 photoacoustic spectrum Methods 0.000 description 6
- 239000004016 soil organic matter Substances 0.000 description 5
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 4
- CURLTUGMZLYLDI-UHFFFAOYSA-N Carbon dioxide Chemical compound O=C=O CURLTUGMZLYLDI-UHFFFAOYSA-N 0.000 description 4
- 229910052799 carbon Inorganic materials 0.000 description 4
- 239000001307 helium Substances 0.000 description 4
- 229910052734 helium Inorganic materials 0.000 description 4
- SWQJXJOGLNCZEY-UHFFFAOYSA-N helium atom Chemical compound [He] SWQJXJOGLNCZEY-UHFFFAOYSA-N 0.000 description 4
- 238000005259 measurement Methods 0.000 description 4
- 239000007789 gas Substances 0.000 description 3
- 235000014121 butter Nutrition 0.000 description 2
- 229910002092 carbon dioxide Inorganic materials 0.000 description 2
- 239000001569 carbon dioxide Substances 0.000 description 2
- 239000000446 fuel Substances 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000003153 chemical reaction reagent Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000001066 destructive effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000035558 fertility Effects 0.000 description 1
- 230000004720 fertilization Effects 0.000 description 1
- 238000002329 infrared spectrum Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003647 oxidation Effects 0.000 description 1
- 238000007254 oxidation reaction Methods 0.000 description 1
- 239000001301 oxygen Substances 0.000 description 1
- 229910052760 oxygen Inorganic materials 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 238000004451 qualitative analysis Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 230000009919 sequestration Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/3563—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light for analysing solids; Preparation of samples therefor
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The soil organism rapid detection method of model: acquisition pedotheque is intelligently identified based on infrared spectroscopy；Mid-infrared light acousto-optic spectrum information is acquired, continuous several times scanning takes average spectrum；Using the content of organic matter of chemical analysis methods pedotheque；Spectral information is pre-processed；Acquire the mid-infrared light acousto-optic spectrum information of soil sample to be measured；It takes average spectrum and pre-processes；Measure the chemical reference value of sample to be tested；Mahalanobis distance between sample to be tested and training sample is calculated, it is ascending to establish new sequence matrix；The characteristics of new sequence matrix is according to sample to be tested establishes the prediction model of different modeling numbers；Model parameter is evaluated, best modeled number is selected and is optimized, intelligent soil identification model is obtained.The present invention establishes the intelligence identification model for meeting its spectral signature " one model of a sample " for each sample spectra, and Soil Background interference and variability, which can be effectively reduced, to be influenced, and intelligence identifies that model universality is strong, and model is steady, effectively improves accuracy of forecast.
Description
Technical field
The invention belongs to the fields of applying IT extensively to agricultural development, and in particular to one kind intelligently identifies mould based on mid-infrared light sound establishment of spectrum
Quick, accurate, lossless measuring method of the type to detection soil organic matter content.
Background technique
The soil organism is the important component of soil and the important indicator of agricultural land soil fertility and soil quality,
The quick detection of soil organic matter content is of great significance for farmland Tree Precise Fertilization.
Measurement soil organic matter content traditional at present generally uses potassium bichromate titrimetric method, hydration heat potassium bichromate oxygen
Change-colorimetric method etc., these methods need to pre-process sample, are complicated for operation, higher cost, test period are long and needs
A large amount of chemical reagent, environment easy to pollute are not suitable for extensive quickly measurement and require.
Infrared spectrum technology has been used to the qualitative and quantitative analysis of soil in recent years.Mid-infrared light optoacoustic spectroscopy is a kind of base
In the infrared spectroscopy of modern optoacoustic conversion, principle is incident on infrared light in the sample cell of optoacoustic attachment, and sample is by red
Fuel factor is generated after outer light irradiation, the gas in photoacoustic cell is by heat wave is converted into after fuel factor, heat wave is by sensitive microphone
Detection, obtains infrared photoacoustic spectra.Infrared photoacoustic spectra test sample without pre-treatment, to sample nondestructive, can be achieved in-site detecting,
Good application potential is shown in agricultural.
Infrared spectroscopy is combined with chemometrics method, constructs prediction model, it can be achieved that soil organic matter content
Prediction, such mould is based on that given sample carries out model construction, model single stable, and calibration set and verifying collection is relatively fixed.
But soil texture is complicated, and background interference is big, and soil variation coefficient is big, and different sampled point soil have different characteristic, vulnerable to back
Scape interference effect, and the too little or too much accuracy and universality that can all influence model of modeling sample number.How by infrared spectroscopy
It is combined with Chemical Measurement, building and Optimized model, realizes that soil property is fast, accurately predicted, the soil organism is contained
The measurement of amount has great importance and the research emphasis and difficult point in the field.
Summary of the invention
In view of the above-mentioned problems, the present invention provides a kind of Intelligent chemical meterological modeling method, may be implemented organic to soil
Quick, accurate, the lossless detection of matter content.By detecting the infrared photoacoustic spectra information of pedotheque, by sample light to be measured
Spectrum is ranked up identification by similitude with known spectra sample, for each pedotheque different characteristic, selects most like
Soil spectrum matrix and best modeled number of samples establish PLSR model prediction, form the intelligence identification mould of " a same model "
Type can be effectively reduced Soil Background interference.This method is modeled using region whole soil sample Forecast of Spectra merely with previous
Method is compared, and model stability and precision of prediction greatly improve, and model universality is stronger, is realized to pedotheque organic matter
Fast, accurately detected.
Technical scheme is as follows:
It is a kind of intelligently to identify that the soil organism rapid detection method of model is quick, accurate, lossless based on infrared spectroscopy
Method for establishing model comprising the steps of:
(1): acquisition pedotheque divides training sample and sample to be tested；
(2): the sampling of soil training sample is placed in FT-mid-IR fiber optics spectroscopy instrument-optoacoustic attachment sample cell, acquires
Mid-infrared light acousto-optic spectrum information, continuous several times scanning, takes average spectrum；
(3): using the content of organic matter for the pedotheque for having surveyed spectrum in chemical analysis methods step (1)；
(4): the spectral information that step (2) acquire being pre-processed, is pre-processed using de-noising, smooth, standardization；
(5): soil sample sampling to be measured is placed in FT-mid-IR fiber optics spectroscopy instrument-optoacoustic attachment sample cell, acquires
Mid-infrared light acousto-optic spectrum information；Continuous several times scanning takes average spectrum, and by the method for step (4) to the light of sample to be tested
Spectrum is pre-processed.According to step (3) method, the chemical reference value of sample to be tested is measured；
(6): it extracts one every time and has pre-processed sample to be tested spectrum, be calculated and compared by mahalanobis distance method, and
New sequence matrix from small to large is established according to the mahalanobis distance between sample to be tested and training sample；
(7): the characteristics of new sequence matrix is according to sample to be tested is established respective by Partial Least Squares Regression (PLSR) method
Prediction model, and according to different modelings collect number obtain different PLSR models；
(8): passing through the standard deviation SD of index of correlation such as sample, coefficient R2, root-mean-square error RMSE and model are pre-
Performance synthesis evaluation index RPD is surveyed, and verifies the root-mean-square error RMSEP of model and the root-mean-square error of forecast sample
The different models that the ratio of PRMSECV establishes modeling collection numbers different in PLSR are evaluated；
(9): by step (6), the method for (7), (8), the foundation and optimization of model are carried out to each sample to be tested, is obtained
To the intelligent soil identification model of best " a same model ".
It brings sample to be tested spectrum into model, the content of organic matter is calculated and is predicted, obtains predicted value；Then it utilizes
Chemical reference value obtained by step (5) chemical method is compared with the intelligence identification resulting predicted value of model prediction, constructed
Intelligence identification model is almost the same to chemical reference value, and prediction result is reliable.
More optimized and more specifically, each step operation method of the present invention is as follows:
(1): acquisition pedotheque is air-dried, ground and is passed through 2mm hole sizer to the sample of acquisition.Divide training sample
And sample to be tested.
(2): soil training sample samples every part of 100-200mg, is placed in FT-mid-IR fiber optics spectroscopy instrument (Nicolet
6700, Thermo Fisher Scientific, USA) in-optoacoustic attachment (PA 300, MTEC, USA) sample cell, it is red in acquisition
Outer optoacoustic spectroscopy information, acquisition wave-length coverage are 4000-400cm-1, scanning resolution 4cm-1；Index glass rate is 0.3162cm-1S, 32 continuous scannings take average spectrum, carry out background correction using carbon black before test.
(3): using the content of organic matter for the pedotheque for having surveyed spectrum in chemical analysis methods step (1), preferred method
For hydration heat potassium dichromate oxidation-colorimetric method.
(4): the spectral information that step (2) acquire being pre-processed, using de-noising, smooth, standardization pretreatment (Du Chang
Text, " soil infrared photoacoustic spectra principle and application ", Science Press, Beijing, 2012 editions).The specific method is as follows: denoising smooth
Method is carried out using wavelet filtering, provides many digital filtering functions in Matlab software, is carried out using filtfilt function,
Syntactic structure are as follows:
[b, a]=butter (n, wn, ' low ')
SpectrumF=filtfilt (b, a, spectrum)
The butterworth filter that wherein variable a, the b one n rank cutoff frequency that has been footage application definition are wn, low table
Show that cutoff frequency is the low-pass filtering of wn, after spectrum Spectrum filtering, return value SpectrumF.
Standardization is using the normalization function provided in Matlab software:
[pn, ps]=mapminmas (spectrum)
Wherein pn is the data after standard, and ps is the data containing former data average and standard deviation information.
(5): acquiring soil sample to be measured and sample every part of 100-200mg, be placed in FT-mid-IR fiber optics spectroscopy instrument
In (Nicolet6700, Thermo Fisher Scientific, USA)-optoacoustic attachment (PA 300, MTEC, USA) sample cell,
Mid-infrared light acousto-optic spectrum information is acquired, acquisition wave-length coverage is 4000-400cm-1, scanning resolution 4cm-1；Index glass rate is
0.3162cm-1S, 32 continuous scannings take average spectrum, and are located in advance by the method for step (4) to the spectrum of sample to be tested
Reason.According to step (3) method, the chemical reference value of sample to be tested is measured.
(6): extract a sample to be tested every time, will pass through through sample to be tested spectrum that step (4) pre-processs geneva away from
It is calculated and compared from method.Mahalanobis distance D calculation formula between two samples is as follows:
Wherein x is the spectroscopic data vector of training sample, and y is the spectroscopic data vector of sample to be tested, and S is sample to be tested association
Variance matrix；The mahalanobis distance of sample to be tested and training sample is calculated, and mahalanobis distance is new according to sequence composition from small to large
Sequence sets, the calculation method can be calculated by Matlab 2013a software.
(7): the characteristics of new sequence matrix is according to sample to be tested is established respective by Partial Least Squares Regression (PLSR) method
Prediction model.In the prediction model of each sample to be measured, the number of samples of modeling is respectively 25,30,35,40
A ... ..., being divided into number with is 5 incremental, and so on, until the maximum number of light matrix is that modeling collection inputs, respectively
Using Leave-one-out cross validation method, the PLSR prediction model of different modeling collection numbers is obtained.
(8): passing through the standard deviation SD of index of correlation such as sample, coefficient R2, root-mean-square error RMSE and model are pre-
Performance synthesis evaluation index RPD is surveyed, and verifies the root-mean-square error RMSEP of model and the root-mean-square error of forecast sample
The model that the ratio of PRMSECV establishes different modeling numbers is evaluated.
Model foundation and calculating process are by Matlab 2013a software progress (Mathwork, USA), and detailed process is such as
Under:
Wherein y is the test chemical value of sample；Y' is sample predictions value,For the average value of sample chemical reference value, n is
Sample number；I is from first soil sample to n-th of counting.
Wherein R2It is smaller closer to 1, RMSE, illustrate that the estimated performance of model is better.As RPD > 2, it is believed that model quality
It is excellent；As 1.5 < PRD < 2, it is believed that model is acceptable；As RPD < 1.5, then it is assumed that model it is poor it is unacceptable (Du Changwen,
" soil infrared photoacoustic spectra principle and application ", Science Press, Beijing, 2012 editions).By calculating, can be obtained by not same
Prediction model constructed by this number, when the coefficient R of prediction model2Close to 1, root-mean-square error RMSE close to 0, mould
Type estimated performance comprehensive evaluation index RPD is greater than 2, RMSEP/RMSECV less than 1.2, and the sample canonical under the number of samples
When poor SD is sufficiently large, the best modeled number of samples for establishing prediction model under this condition can be established.
(9): by step (6), the method for (7), (8), the foundation and optimization of model are carried out to each sample to be tested, is obtained
To the intelligent soil identification model of " a same model ".It brings sample to be tested spectrum into model, the content of organic matter is calculated
With prediction, predicted value is obtained；Then using obtained by chemical reference value obtained by step (5) chemical method and intelligence identification model prediction
Predicted value be compared, constructed intelligence identification model is almost the same to chemical reference value, and prediction result is reliable.
The beneficial effects of the present invention are:
The present invention provides a kind of intelligent identification method according to soil characteristic modeling and forecasting, its main feature is that being directed to each sample
Product establishment of spectrum meets the intelligence identification model of its spectral signature " one model of a sample ", can be effectively reduced Soil Background interference
And variability influences, intelligently identification model universality is strong for this, and model is steady, effectively improves accuracy of forecast.
Detailed description of the invention
Fig. 1 is soil intelligently operational flowchart of the identification model to content of organic matter fast non-destructive detection method.
Fig. 2 is the standard deviation SD of index of correlation such as sample, coefficient R2, root-mean-square error RMSE and model prediction
Energy comprehensive evaluation index RPD, and verify the root-mean-square error RMSEP of model and the root-mean-square error of forecast sample
The model evaluation that the ratio of PRMSECV establishes different modeling numbers.
Fig. 3 is intelligently to identify that the soil organic matter content that model foundation prediction model obtains is joined with chemistry in present embodiment
Examine the result figure of value.
Specific embodiment
The following examples are intended to illustrate the invention, but the range being not intended to limit the invention.
Unless otherwise specified, the conventional means that technological means all in embodiment is well known to those skilled in the art.
Embodiment 1
(1): 933 parts of pedotheque of acquisition is 0-20cm topsoil, is air-dried, ground simultaneously to the sample of acquisition
Pass through 2mm hole sizer.Wherein it is used as training sample for 711 parts, 222 parts are used as sample to be tested.
(2): soil training sample samples every part of 100-200mg, and present embodiment uses FT-mid-IR fiber optics spectroscopy
Instrument (Nicolet 6700, Thermo Fisher Scientific, USA)-optoacoustic attachment (PA 300, MTEC, USA) sample
Pond, sample, which is put into photoacoustic cell, is no more than 2/3, and sample cell is closed after sample is placed, high-purity helium valve is opened, with height
Pure helium is purged, and steam and carbon dioxide gas in photoacoustic cell are removed.Mid-infrared light acousto-optic spectrum information is acquired, wavelength is acquired
Range is 4000-400cm-1, scanning resolution 4cm-1；Index glass rate is 0.3162cm-1S, 32 times continuous scanning takes average light
Spectrum carries out background correction using carbon black before test.
(3): using the content of organic matter for the pedotheque for having surveyed spectrum in chemical analysis methods step (1), preferred method
For hydration heat potassium dichromate oxidation-colorimetric method, is measured by the method as organic carbon quality, is converted by formula,
Wherein O.M is the mass fraction of the soil organism；m1For soil carbon sequestration (mg)；M is quality of soil sample；1.724 being
Organic carbon converts the coefficient (calculating by the mean carbon content of the soil organism into 58%) of organic matter；1.32 oxidation correction systems
Number.By calculating, and record the chemical reference value of the content of organic matter of each sample.
(4): by the spectral information progress de-noising of step (2) acquisition, smooth, (Du Changwen, " soil is red for standardization pretreatment
Outer optoacoustic spectroscopy principle and application ", Science Press, Beijing, 2012 editions).The specific method is as follows:
Denoising smooth method is carried out using wavelet filtering, provides many digital filtering functions in Matlab software, is used
Filtfilt function carries out, syntactic structure are as follows:
[b, a]=butter (n, wn, ' low ')
SpectrumF=filtfilt (b, a, spectrum)
The butterworth filter that wherein variable a, the b one n rank cutoff frequency that has been footage application definition are wn, low table
Show that cutoff frequency is the low-pass filtering of wn, after spectrum Spectrum filtering, return value SpectrumF.
Standardization is using the normalization function provided in Matlab software:
[pn, ps]=mapminmas (spectrum)
Wherein pn is the data after standard, and ps is the data containing former data average and standard deviation information.
(5): totally 222 parts of soil sample to be measured, sampling every part of 100-200mg, be placed in FT-mid-IR fiber optics spectroscopy instrument
(Nicolet 6700, Thermo Fisher Scientific, USA)-optoacoustic attachment (PA 300, MTEC, USA) sample cell
In, sample, which is put into photoacoustic cell, is no more than 2/3, and sample cell is closed after sample is placed, high-purity helium valve is opened, with height
Pure helium is purged, and steam and carbon dioxide gas in photoacoustic cell are removed.Mid-infrared light acousto-optic spectrum information is acquired, wavelength is acquired
Range is 4000-400cm-1, scanning resolution 4cm-1；Index glass rate is 0.3162cm-1S, 32 times continuous scanning takes average light
Spectrum, and the spectrum of sample to be tested is pre-processed by the method for step (4), according to step (3) method, measure sample to be tested
Chemical reference value.
(6): extract a sample to be tested every time, will pass through through sample to be tested spectrum that step (4) pre-processs geneva away from
It is calculated and compared from method.Mahalanobis distance D calculation formula between two samples is as follows:
Wherein x is the spectroscopic data vector of training sample, and y is the spectroscopic data vector of sample to be tested, and S is sample to be tested association
Variance matrix；The mahalanobis distance of sample to be tested and training sample is calculated, and mahalanobis distance is new according to sequence composition from small to large
Sequence sets, the calculation method can be calculated by Matlab 2013a software.
(7): the characteristics of new sequence matrix is according to sample to be tested is established respective by Partial Least Squares Regression (PLSR) method
Prediction model.In the prediction model of each sample to be measured, the number of samples of modeling is respectively 25,30,35,40
A ... ..., being divided into number with is 5 incremental, and so on, until maximum number 930 of light matrix input for modeling collection,
Leave-one-out cross validation method is respectively adopted, obtains the PLSR prediction model of different modeling collection numbers.
(8): passing through the standard deviation SD of index of correlation such as sample, coefficient R2, root-mean-square error RMSE and model are pre-
Performance synthesis evaluation index RPD is surveyed, and verifies the root-mean-square error RMSEP of model and the root-mean-square error of forecast sample
The model that the ratio of PRMSECV establishes different modeling numbers is evaluated.
Model foundation and calculating process are by Matlab 2013a software progress (Mathwork, USA), and detailed process is such as
Under:
Wherein y is the test chemical value of sample；Y' is sample predictions value,For the mean value of sample chemical reference value, n is sample
Product capacity；I is from first soil sample to n-th of counting.
Wherein R2It is smaller closer to 1, RMSE, illustrate that the estimated performance of model is better.As RPD > 2, it is believed that model quality
It is excellent；As 1.5 < PRD < 2, it is believed that model is acceptable；As RPD < 1.5, then it is assumed that model it is poor it is unacceptable (Du Changwen,
" soil infrared photoacoustic spectra principle and application ", Science Press, Beijing, 2012 editions).By calculating, so that prediction model
Coefficient R2Close to 1, root-mean-square error RMSE close to 0, model prediction performance synthesis evaluation index RPD be greater than 2,
RMSEP/RMSECV is less than 1.2, and the sample standard deviation SD under the number of samples is sufficiently large, and prediction model is established in final establishment
Best modeled number of samples.Different samples have the independent prediction for meeting respective Spectral characteristics of soil and best modeled sample number
Model.
By taking wherein three samples to be tested as an example (table 1), when it is 60 that sample 1, which models calibration set number, prediction model is related
Coefficient is that 0.8698, RMSEP value is 0.17g kg-1, RPD value is that 2.77, RMSEP/RMSECV ratio is 0.17, meets optimal mould
Shape parameter；When the modeling calibration set number of sample 2 is 65, prediction model related coefficient is that 0.8848, RMSEP value is 0.44g kg-1,
RPD value is that 3.05, RMSEP/RMSECV ratio is 0.40, meets optimal models；When the modeling calibration set number of sample 3 is 65, in advance
It is 0.38g kg that survey model related coefficient, which is 0.8924, RMSEP value,-1, RPD value is that 2.41, RMSEP/RMSECV ratio is 0.42,
Meet optimal model parameters；When other being selected to model number, each parameter of gained model is not up to its optimal value (see Fig. 2).
1 sample to be tested collection optimum modeling parameter of table
(9): by step (6), the method for (7), (8), the foundation and optimization of model are carried out to each sample to be tested, is obtained
To the intelligent soil identification model of " a same model ".The R of verifying model is obtained by optimized parameter optimization2It is 0.9082,
RMSEP is 1.6509g kg-1, RPD value is 3.01.Bring sample to be tested spectrum into model, to the content of organic matter carry out calculate with
Prediction, obtains predicted value；Then resulting using chemical reference value obtained by step (5) chemical method and intelligence identification model prediction
Predicted value is compared, and Fig. 3 indicates the training sample true value of the model and the scatter plot distributions of predicted value, and sample to be tested is substantially
Fit line two sides are distributed in, without biggish prediction deviation, it is believed that the intelligent soil identification model stability of " a same model " can
It leans on.
Therefore, during soil organism quick predict, using the intelligent soil identification model of " a same model ", behaviour
It is simple to make process, precision of prediction and universality can be improved.
Each sample to be tested is established the intelligence for meeting its spectral signature " one model of a sample " by present embodiment first
Identify model, Soil Background interference and variability, which is effectively reduced, to be influenced, and detects prediction soil then in conjunction with Partial Least Squares Regression
The content of organic matter.Intelligently identification model universality is strong for this, and model is steady, effectively improves accuracy of forecast, and in accurate agricultural
Organic matter information quick obtaining provides research ideas and methods and supports.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, without departing from the technical principles of the invention, several improvement can also be made and connect retouching, these improvements and modifications
Also it should be regarded as protection scope of the present invention.
Claims (3)
1. a kind of soil organism rapid detection method for intelligently identifying model based on infrared spectroscopy, which is characterized in that step is such as
Under:
(1): acquisition pedotheque divides training sample and sample to be tested；
(2): the sampling of soil training sample is placed in FT-mid-IR fiber optics spectroscopy instrument-optoacoustic attachment sample cell, red in acquisition
Outer optoacoustic spectroscopy information, continuous several times scanning, takes average spectrum；
(3): using the content of organic matter for the pedotheque for having surveyed spectrum in chemical analysis methods step (1)；
(4): the spectral information that step (2) acquire being pre-processed, is pre-processed using de-noising, smooth, standardization；
(5): soil sample sampling to be measured is placed in FT-mid-IR fiber optics spectroscopy instrument-optoacoustic attachment sample cell, red in acquisition
Outer optoacoustic spectroscopy information；Continuous several times scanning, takes average spectrum, and by the method for step (4) to the spectrum of sample to be tested into
Row pretreatment；According to step (3) method, the chemical reference value of sample to be tested is measured；
(6): it extracts one every time and has pre-processed sample to be tested spectrum, be calculated and compared by mahalanobis distance method, and according to
Mahalanobis distance between sample to be tested and training sample establishes new sequence matrix from small to large；
(7): the characteristics of new sequence matrix is according to sample to be tested is established respective pre- by Partial Least Squares Regression (PLSR) method
Model is surveyed, and collects number according to different modelings and obtains different PLSR models；
(8): being evaluated by the different models that index of correlation establishes modeling collection numbers different in PLSR, wherein correlation refers to
It is designated as standard deviation SD, the coefficient R of sample2, root-mean-square error RMSE and model prediction performance synthesis evaluation index
RPD, and verify the ratio of the root-mean-square error RMSEP of model and the root-mean-square error PRMSECV of forecast sample；
(9): by step (6), the method for (7), (8), the foundation and optimization of model are carried out to each sample to be tested, is obtained most
The intelligent soil identification model of good " a same model "；
Acquisition mid-infrared light acousto-optic spectrum information in step (2), acquisition wave-length coverage are 4000-400cm-1, scanning resolution is
4cm-1；Index glass rate is 0.3162cm-1S, 32 times continuous scanning takes average spectrum；
Mahalanobis distance D calculation formula in step (6) between two samples is as follows:
Wherein x is the spectroscopic data vector of training sample, and y is the spectroscopic data vector of sample to be tested, and S is sample to be tested covariance
Matrix；The mahalanobis distance of sample to be tested and training sample is calculated, and mahalanobis distance is formed into new sequence according to sequence from small to large
Collection, the calculation method can be calculated by Matlab2013a software；
In step (7), the characteristics of new sequence matrix is according to sample to be tested, is established respective by Partial Least Squares Regression PLSR method
Prediction model；In the prediction model of each sample to be measured, the number of samples of modeling is respectively 25,30,35,40
A ... ..., being divided into number with is 5 incremental, and so on, until maximum number 930 of light matrix input for modeling collection,
Leave-one-out cross validation method is respectively adopted, establishes PLSR prediction model；
The acquisition of step (5) soil sample mid-infrared light acousto-optic spectrum information to be measured, acquisition wave-length coverage are 4000-400cm-1, sweep
Retouching resolution ratio is 4cm-1；Index glass rate is 0.3162cm-1S, 32 times continuous scanning takes average spectrum, to the spectrum of sample to be tested
It is pre-processed and measures chemical reference value；
The model established in step (8) to different modeling numbers is evaluated, and detailed process is as follows:
Wherein y is the test chemical value of sample；Y' is sample predictions value,For the mean value of sample chemical reference value, yiIn i table
Show that sample number, n are sample capacity；I is from first soil sample to n-th of counting.
2. the soil organism rapid detection method according to claim 1 that model is intelligently identified based on infrared spectroscopy,
It is characterized in that, background correction is carried out using carbon black before test in step (2).
3. the soil organism rapid detection method according to claim 1 or 2 that model is intelligently identified based on infrared spectroscopy,
It is characterized in that, " using the content of organic matter of chemical analysis methods pedotheque " in step (3), the method is hydration heat
Potassium dichromate oxidation-colorimetric method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510816704.0A CN105486655B (en) | 2015-11-23 | 2015-11-23 | The soil organism rapid detection method of model is intelligently identified based on infrared spectroscopy |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510816704.0A CN105486655B (en) | 2015-11-23 | 2015-11-23 | The soil organism rapid detection method of model is intelligently identified based on infrared spectroscopy |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105486655A CN105486655A (en) | 2016-04-13 |
| CN105486655B true CN105486655B (en) | 2018-12-18 |
Family
ID=55673779
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510816704.0A Active CN105486655B (en) | 2015-11-23 | 2015-11-23 | The soil organism rapid detection method of model is intelligently identified based on infrared spectroscopy |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105486655B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106198434A (en) * | 2016-07-26 | 2016-12-07 | 新疆农垦科学院 | A kind of remote-sensing monitoring method of the North SinKiang desert grey soil content of organic matter |
| CN107024445A (en) * | 2017-04-17 | 2017-08-08 | 中国科学院南京土壤研究所 | The modeling method and detection method of the quick detection of Nitrate in Vegetable |
| CN107132190A (en) * | 2017-04-21 | 2017-09-05 | 武汉大学 | A kind of soil organism spectra inversion model calibration samples collection construction method |
| CN108254333A (en) * | 2017-12-14 | 2018-07-06 | 中国科学院南京土壤研究所 | Nitrogenous pesticide residue rapid detection method based on infrared photoacoustic spectra |
| RU2698954C1 (en) * | 2018-07-31 | 2019-09-02 | федеральное государственное бюджетное образовательное учреждение высшего образования "Самарский государственный технический университет" | Method of quantitative determination of humus |
| CN109766909B (en) * | 2018-11-29 | 2023-10-17 | 温州大学 | Analysis method for aging behavior of shore environment microplastic based on spectrogram fusion |
| CN111125885A (en) * | 2019-12-03 | 2020-05-08 | 杭州电子科技大学 | ASF correction table construction method based on improved kriging interpolation algorithm |
| CN111488926B (en) * | 2020-04-07 | 2024-03-29 | 中国科学院南京土壤研究所 | Soil organic matter determination method based on optimization model |
| CN114034684B (en) * | 2021-10-13 | 2023-05-30 | 中国科学院南京土壤研究所 | Soil attribute rapid determination method based on one-stop spectrum model |
| CN115931828B (en) * | 2023-02-17 | 2023-06-16 | 华谱智能科技(天津)有限公司 | Component analysis and prediction method, unit and system suitable for complex soil matrix |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009058443A (en) * | 2007-08-31 | 2009-03-19 | National Institute For Environmental Studies | Method for automatically categorizing rhizosphere element in soil |
| WO2010148458A1 (en) * | 2009-06-25 | 2010-12-29 | Commonwealth Scientific And Industrial Research Organisation | Method of detecting contaminants |
| CN103884661A (en) * | 2014-02-21 | 2014-06-25 | 浙江大学 | Soil total nitrogen real-time detection method based on soil visible-near infrared spectrum library |
| CN103954586A (en) * | 2014-05-13 | 2014-07-30 | 泰顺派友科技服务有限公司 | Method for rapidly predicting content of soil organic matters based on eleven spectrum wavelet coefficients |
| CN104990894A (en) * | 2015-07-09 | 2015-10-21 | 南京富岛信息工程有限公司 | Detection method of gasoline properties based on weighted absorbance and similar samples |
-
2015
- 2015-11-23 CN CN201510816704.0A patent/CN105486655B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009058443A (en) * | 2007-08-31 | 2009-03-19 | National Institute For Environmental Studies | Method for automatically categorizing rhizosphere element in soil |
| WO2010148458A1 (en) * | 2009-06-25 | 2010-12-29 | Commonwealth Scientific And Industrial Research Organisation | Method of detecting contaminants |
| CN103884661A (en) * | 2014-02-21 | 2014-06-25 | 浙江大学 | Soil total nitrogen real-time detection method based on soil visible-near infrared spectrum library |
| CN103954586A (en) * | 2014-05-13 | 2014-07-30 | 泰顺派友科技服务有限公司 | Method for rapidly predicting content of soil organic matters based on eleven spectrum wavelet coefficients |
| CN104990894A (en) * | 2015-07-09 | 2015-10-21 | 南京富岛信息工程有限公司 | Detection method of gasoline properties based on weighted absorbance and similar samples |
Non-Patent Citations (2)
| Title |
|---|
| 应用红外光声光谱技术及支持向量机模型测定土壤有机质含量;曾胤 等;《土壤学报》;20141130;第51卷(第6期);第1263-1265页第1节 * |
| 甘肃黑河流域上游森林地上生物量的多光谱遥感估测;郭云 等;《林业科学》;20150131;第51卷(第1期);第140-149页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105486655A (en) | 2016-04-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105486655B (en) | The soil organism rapid detection method of model is intelligently identified based on infrared spectroscopy | |
| CN104897607B (en) | Portable near infrared spectrum food modeling and quick detection integral method and system | |
| CN110174359B (en) | Aviation hyperspectral image soil heavy metal concentration assessment method based on Gaussian process regression | |
| CN106841103A (en) | Near infrared spectrum detects fruit internal quality method and dedicated test system | |
| CN108801934A (en) | A kind of modeling method of soil organic carbon EO-1 hyperion prediction model | |
| CN103196838B (en) | Hyperspectral remote sensing monitoring method for coastal estuary eutrophication | |
| CN108956505B (en) | Method and device for detecting concentration of chlorophyll a in small water body based on Sentinel-2 image | |
| CN110531054B (en) | Soil organic carbon prediction uncertainty estimation method based on Bootstrap sampling | |
| CN109031344A (en) | A kind of method of Full wave shape laser radar and high-spectral data joint inversion forest structural variable | |
| CN109669023A (en) | A kind of soil attribute prediction technique based on Multi-sensor Fusion | |
| CN111488926B (en) | Soil organic matter determination method based on optimization model | |
| CN111855608B (en) | Near-infrared nondestructive detection method for apple acidity based on fusion characteristic wavelength selection algorithm | |
| CN110118742A (en) | Land table Vegetation canopy ecology water content remote sensing inversion method based on spectrum analysis | |
| CN102636450A (en) | Method for detecting wolfberry polyose content in Chinese wolfberry in a nondestructive way based on near infrared spectrum technology | |
| CN111965140B (en) | Wavelength point recombination method based on characteristic peak | |
| CN105784672A (en) | Drug detector standardization method based on dual-tree complex wavelet algorithm | |
| CN102072767A (en) | Wavelength similarity consensus regression-based infrared spectrum quantitative analysis method and device | |
| CN102841069B (en) | Method for rapidly identifying types of crude oil by using mid-infrared spectrum | |
| CN109060771A (en) | A kind of common recognition model building method based on spectrum different characteristic collection | |
| CN106126879B (en) | A kind of soil near-infrared spectrum analysis prediction technique based on rarefaction representation technology | |
| CN102322957A (en) | Spectrum drifting detection method for interference type hyperspectral imager | |
| CN111141809B (en) | Soil nutrient ion content detection method based on non-contact type conductivity signal | |
| CN205506679U (en) | Portable near infrared spectroscopy detecting system | |
| CN206788033U (en) | A kind of dedicated test system of near infrared spectrum detection fruit internal quality | |
| CN104181125A (en) | Method for rapidly determining Kol-bach value of beer malt |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |