CN114199791A - Monitoring method for predicting wheat nitrogen content through end-to-end deep learning - Google Patents
Monitoring method for predicting wheat nitrogen content through end-to-end deep learning Download PDFInfo
- Publication number
- CN114199791A CN114199791A CN202111578893.4A CN202111578893A CN114199791A CN 114199791 A CN114199791 A CN 114199791A CN 202111578893 A CN202111578893 A CN 202111578893A CN 114199791 A CN114199791 A CN 114199791A
- Authority
- CN
- China
- Prior art keywords
- wheat
- nitrogen content
- spectrum
- deep learning
- neural network
- 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.)
- Withdrawn
Links
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 title claims abstract description 94
- 241000209140 Triticum Species 0.000 title claims abstract description 80
- 235000021307 Triticum Nutrition 0.000 title claims abstract description 80
- 229910052757 nitrogen Inorganic materials 0.000 title claims abstract description 47
- 238000000034 method Methods 0.000 title claims abstract description 24
- 238000012544 monitoring process Methods 0.000 title claims abstract description 17
- 238000013135 deep learning Methods 0.000 title claims abstract description 16
- 238000001228 spectrum Methods 0.000 claims abstract description 50
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 18
- 238000003384 imaging method Methods 0.000 claims abstract description 15
- 238000012549 training Methods 0.000 claims abstract description 6
- 239000000523 sample Substances 0.000 claims description 18
- 230000003595 spectral effect Effects 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 5
- 102000004190 Enzymes Human genes 0.000 claims description 3
- 108090000790 Enzymes Proteins 0.000 claims description 3
- 238000004140 cleaning Methods 0.000 claims description 3
- 238000001035 drying Methods 0.000 claims description 3
- 238000005303 weighing Methods 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 claims description 2
- 238000012795 verification Methods 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 abstract description 4
- 238000011176 pooling Methods 0.000 abstract description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000004836 empirical method Methods 0.000 description 2
- 239000000618 nitrogen fertilizer Substances 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 241000282414 Homo sapiens Species 0.000 description 1
- 229910002651 NO3 Inorganic materials 0.000 description 1
- NHNBFGGVMKEFGY-UHFFFAOYSA-N Nitrate Chemical compound [O-][N+]([O-])=O NHNBFGGVMKEFGY-UHFFFAOYSA-N 0.000 description 1
- 238000012271 agricultural production Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 235000013339 cereals Nutrition 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 235000013305 food Nutrition 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000005431 greenhouse gas Substances 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000002329 infrared spectrum Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000002310 reflectometry Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N2201/00—Features of devices classified in G01N21/00
- G01N2201/12—Circuits of general importance; Signal processing
- G01N2201/129—Using chemometrical methods
- G01N2201/1296—Using chemometrical methods using neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention discloses a method for monitoring wheat nitrogen content through end-to-end deep learning prediction, which comprises the steps of reconstructing a wheat canopy spectrum in a collection region by using linear interpolation, simultaneously measuring the nitrogen content of representative wheat in the region, and training a convolutional neural network through the collected canopy spectrum and the measured nitrogen content to obtain a wheat nitrogen content prediction model. In the method for measuring the nitrogen content of the wheat, interpolation reconstruction is carried out on multi-source wheat canopy spectrum data collected by a single-point and imaging spectrometer, then the nitrogen content of the wheat is predicted by combining convolutional neural network modeling, preprocessing and characteristic waveband selection are not needed for the spectrum, end-to-end rapid prediction of a spectrum sample directly to a nitrogen content index is realized, preprocessing and characteristic waveband selection are not needed for the modeled spectrum, important information in the spectrum is extracted through a convolutional layer and a pooling layer of the convolutional neural network, the established model has better universality and robustness, and accurate prediction of the nitrogen content of the wheat can be realized through the model.
Description
Technical Field
The invention belongs to the technical field of crop growth monitoring, and particularly relates to a method for monitoring wheat nitrogen content through end-to-end deep learning prediction.
Background
Wheat is one of three grain crops in the world, and accounts for more than 20% of the global arable land area. China is large in population and food supply is not in demand. Therefore, wheat plays a strategic position of playing a key role in agricultural production. In recent years, agricultural names often increase the yield by applying excessive nitrogen fertilizer, which not only significantly affects the yield and quality of wheat, but also causes the utilization rate of the nitrogen fertilizer to be greatly reduced, a large amount of nitrate leaks downwards, the emission of greenhouse gas (N2O) is increased, and underground water is polluted, thereby seriously threatening the health of human beings.
The visible light and near infrared spectrum technology is a simple, fast and nondestructive detection technology. A model of physical and chemical indexes of the sample can be established through the transmissivity or reflectivity of the sample, various components and physical characteristic parameters in the sample can be detected simultaneously, and the analysis result can accurately approach the traditional determination method.
In recent years, with the development of spectroscopic technology (spectroscopic devices, detectors), there are more and more types of spectroscopic instruments, such as single-point, imaging, and the like. Different manufacturers use different detectors, and the spectral intervals are also inconsistent. Traditional empirical methods, linear (PLS, etc.) and nonlinear methods (SVR, RF, etc.) all require spectral preprocessing of the spectrum and the selection of the eigenbands, where preprocessing has a great influence on the modeling results and the selection of the eigenbands also loses important information of other bands. The established model is not very universal and robust.
Disclosure of Invention
The invention aims to provide an end-to-end deep learning wheat nitrogen content monitoring method, which aims to solve the technical problems that the spectrum pretreatment and the characteristic wave band selection are required to be carried out on the spectrum by the traditional empirical method, the linear (PLS and the like) method and the nonlinear (SVR, RF and the like) method, the pretreatment has great influence on the modeling result, and the selection of the characteristic wave band can lose important information of other wave bands.
In order to achieve the purpose, the specific technical scheme of the monitoring method for predicting the nitrogen content of the wheat through end-to-end deep learning is as follows:
an end-to-end deep learning wheat nitrogen content prediction monitoring method comprises the following steps:
step 2, selecting representative wheat in the selected area in the step 1, cleaning, separating, deactivating enzyme, drying, weighing, and measuring the actual nitrogen content of the wheat by using a flow analyzer;
step 3, reconstructing the wheat canopy spectrum collected in the step 1 by using linear interpolation, and then extracting a 400nm-1000nm wave band as a spectrum sample;
step 4, establishing a convolutional neural network, inputting the spectrum sample in the step 3 into the convolutional neural network for training to obtain an output value;
step 5, carrying out convolution neural network training iteration for 500 times in the step 4, and selecting one of output values which is closest to the value of the nitrogen content measured in the step 2 as a wheat nitrogen content prediction model to output;
and 6, subsequently, the nitrogen content of the wheat can be predicted only by reconstructing the collected wheat canopy spectrum by using linear interpolation and inputting the reconstructed wheat canopy spectrum into the wheat nitrogen content prediction model obtained in the step 5.
Preferably, in order to improve the accuracy of collecting the wheat canopy spectrum, when the wheat canopy spectrum is collected by using the single-point spectrometer or the imaging spectrometer in step 1, the probe is required to be vertically downward in sunny and windless weather, and the distance between the probe and the wheat is 1 m.
Preferably, in order to improve the accuracy of the single-point spectrometer for collecting the wheat canopy spectrum, in step 1, a white board correction is required before the single-point spectrometer is used for collecting the wheat canopy spectrum each time.
Preferably, in order to improve the accuracy of the imaging spectrometer for collecting the wheat canopy spectrum, each spectral image collected by the imaging spectrometer in step 1 needs to be subjected to black and white correction.
Preferably, in order to improve the accuracy of the obtained wheat nitrogen content prediction model, 1/3 is selected as a verification set from the spectrum sample selected in step 3, and is used for verifying the accuracy of the wheat nitrogen content prediction model obtained in step 5.
The method for monitoring the nitrogen content of the wheat by end-to-end deep learning prediction has the following advantages:
1. the multisource wheat canopy spectrum data collected by the single-point imaging spectrometer and the imaging spectrometer are subjected to interpolation reconstruction and then combined with convolutional neural network modeling to predict the nitrogen content of the wheat, the spectrum is not required to be preprocessed and the characteristic wave band is not required to be selected, and the end-to-end rapid prediction of the spectrum sample directly to the nitrogen content index is realized;
2. the spectrum of the model is not required to be preprocessed and the characteristic wave band is selected, important information in the spectrum is extracted through a convolutional neural network convolutional layer and a pooling layer, and the established model has better universality and robustness.
Drawings
FIG. 1 is a flow chart of a monitoring method for predicting wheat nitrogen content through end-to-end deep learning according to the present invention;
FIG. 2 is a raw spectrum of the present invention;
FIG. 3 is a spectral plot after interpolation reconstruction in accordance with the present invention;
FIG. 4 is a schematic diagram of a convolutional neural network of the present invention;
Detailed Description
In order to better understand the purpose, structure and function of the present invention, the following describes a monitoring method for predicting wheat nitrogen content through end-to-end deep learning in further detail with reference to the accompanying drawings.
As shown in fig. 1, an end-to-end deep learning wheat nitrogen content monitoring method comprises the following steps:
step 2, selecting representative wheat in the selected area in the step 1, cleaning, separating, deactivating enzyme, drying, weighing, and measuring the actual nitrogen content of the wheat by using a flow analyzer;
step 3, reconstructing the wheat canopy spectrum collected in the step 1 and shown in the figure 2 by using linear interpolation, and then extracting a 400nm-1000nm wave band as a spectrum sample shown in the figure 3 to ensure that the number of points contained in each spectrum sample is consistent;
step 4, establishing a convolutional neural network, as shown in fig. 4, wherein the convolutional neural network comprises 3 convolutional layers, 3 pooling layers and two full-connection layers, inputting the spectrum sample in the step 3 into the convolutional neural network to obtain an output value, comparing the output value with the actual nitrogen content of the wheat measured in the step 2, then performing a back propagation process of the convolutional neural network, calculating an error between the output value and the actual nitrogen content of the wheat, and updating the convolutional neural network according to the error;
step 5, iterating the convolutional neural network training process in the step 4 for 500 times, and selecting the convolutional neural network which is closest to the actual nitrogen content value of the wheat measured in the step 2 from the output values as a wheat nitrogen content prediction model to output;
and 6, subsequently, the nitrogen content of the wheat can be predicted only by reconstructing the collected wheat canopy spectrum by using linear interpolation and inputting the reconstructed wheat canopy spectrum into the wheat nitrogen content prediction model obtained in the step 5.
In the method for measuring the nitrogen content of the wheat, interpolation reconstruction is carried out on multi-source wheat canopy spectrum data collected by a single-point imaging spectrometer and the multi-source wheat canopy spectrum data collected by an imaging spectrometer and then the wheat nitrogen content is predicted by combining convolutional neural network modeling.
It is to be understood that the present invention has been described with reference to certain embodiments, and that various changes in the features and embodiments, or equivalent substitutions may be made therein by those skilled in the art without departing from the spirit and scope of the invention. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the invention without departing from the essential scope thereof. Therefore, it is intended that the invention not be limited to the particular embodiment disclosed, but that the invention will include all embodiments falling within the scope of the appended claims.
Claims (5)
1. An end-to-end deep learning wheat nitrogen content prediction monitoring method is characterized by comprising the following steps:
step 1, selecting wheat in an area, and respectively collecting wheat canopy spectra for multiple times by using a single-point spectrometer and an imaging spectrometer, and taking an average value;
step 2, selecting representative wheat in the selected area in the step 1, cleaning, separating, deactivating enzyme, drying, weighing, and measuring the actual nitrogen content of the wheat by using a flow analyzer;
step 3, reconstructing the wheat canopy spectrum collected in the step 1 by using linear interpolation, and then extracting a 400nm-1000nm wave band as a spectrum sample;
step 4, establishing a convolutional neural network, inputting the spectrum sample in the step 3 into the convolutional neural network for training to obtain an output value;
step 5, carrying out convolution neural network training iteration for 500 times in the step 4, and selecting one of output values which is closest to the value of the nitrogen content measured in the step 2 as a wheat nitrogen content prediction model to output;
and 6, subsequently, the nitrogen content of the wheat can be predicted only by reconstructing the collected wheat canopy spectrum by using linear interpolation and inputting the reconstructed wheat canopy spectrum into the wheat nitrogen content prediction model obtained in the step 5.
2. The method for monitoring the nitrogen content of the wheat through end-to-end deep learning and prediction according to claim 1, wherein in the step 1, when a single-point spectrometer or an imaging spectrometer is used for collecting the wheat canopy spectrum, the probe is required to be vertically downward in clear and calm weather, and the distance between the probe and the wheat is 1 m.
3. The method for monitoring the nitrogen content of the wheat through end-to-end deep learning prediction according to claim 1, wherein a white board correction is required before a single-point spectrometer is used for collecting a wheat canopy spectrum in the step 1.
4. The method for monitoring wheat nitrogen content through end-to-end deep learning prediction as claimed in claim 1, wherein each spectral image acquired by using an imaging spectrometer in the step 1 needs to be corrected in black and white.
5. The method for monitoring the nitrogen content of the wheat through end-to-end deep learning prediction according to claim 1, wherein 1/3 is selected as a verification set from the spectrum samples selected in the step 3, and is used for verifying the accuracy of the wheat nitrogen content prediction model obtained in the step 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111578893.4A CN114199791A (en) | 2021-12-22 | 2021-12-22 | Monitoring method for predicting wheat nitrogen content through end-to-end deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111578893.4A CN114199791A (en) | 2021-12-22 | 2021-12-22 | Monitoring method for predicting wheat nitrogen content through end-to-end deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114199791A true CN114199791A (en) | 2022-03-18 |
Family
ID=80655879
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111578893.4A Withdrawn CN114199791A (en) | 2021-12-22 | 2021-12-22 | Monitoring method for predicting wheat nitrogen content through end-to-end deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114199791A (en) |
-
2021
- 2021-12-22 CN CN202111578893.4A patent/CN114199791A/en not_active Withdrawn
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107796764B (en) | Method for constructing wheat leaf area index estimation model based on three-band vegetation index | |
| CN102788752B (en) | Non-destructive detection device and method of internal information of crops based on spectrum technology | |
| CN108875913B (en) | Tricholoma matsutake rapid nondestructive testing system and method based on convolutional neural network | |
| CN110376167A (en) | Rice leaf nitrogen content monitoring method based on unmanned plane EO-1 hyperion | |
| CN111855589A (en) | Remote sensing inversion model and method for rice leaf nitrogen accumulation | |
| CN110567892B (en) | Summer corn nitrogen hyperspectral prediction method based on critical nitrogen concentration | |
| CN107607486B (en) | Soil total nitrogen detection method and device | |
| CN111855591A (en) | Rice overground part carbon-nitrogen ratio remote sensing inversion model and method | |
| CN111855590A (en) | Remote sensing inversion model and method for rice leaf starch accumulation | |
| CN111829965A (en) | Remote sensing inversion model and method for starch accumulation amount of rice overground part | |
| CN113111799B (en) | Cultivated land soil fertility level monitoring method based on ensemble empirical mode decomposition | |
| CN113436153A (en) | Method for predicting carbon components of undisturbed soil profile based on hyperspectral imaging and support vector machine technology | |
| CN112270131A (en) | Remote sensing inversion model and method for rice leaf area index based on ARD regression algorithm | |
| CN112129709A (en) | Apple tree canopy scale nitrogen content diagnosis method | |
| CN105608296B (en) | A kind of blade potassium concn inversion method based on lichee canopy spectra | |
| CN114676636A (en) | Grassland area soil moisture rapid inversion method integrating vegetation and habitat characteristics | |
| CN111855593A (en) | Remote sensing inversion model and method for starch content of rice leaf | |
| CN112362812A (en) | Remote sensing inversion model and method for rice leaf chlorophyll carotenoid content ratio based on Lars algorithm | |
| CN115598162B (en) | Soil heavy metal content detection method based on stacked model | |
| Chen et al. | Hyperspectral detection of sugar content for sugar-sweetened apples based on sample grouping and SPA feature selecting methods | |
| CN111855592A (en) | Remote sensing inversion model and method for upper dry matter weight in unit area of rice | |
| CN114486786A (en) | Soil organic matter measuring method and measuring system | |
| CN113552096A (en) | Spectrum-based pineapple leaf nitrogen content estimation method | |
| CN111595806A (en) | Method for monitoring soil carbon component by using mid-infrared diffuse reflection spectrum | |
| CN115358321A (en) | Hyperspectral soil quick-acting potassium content prediction method based on convolutional neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20220318 |