Portable soil total nitrogen on-line detection device
By constructing a total nitrogen prediction model library and integrating a small laboratory testing structure with a mobile field testing structure in a mobile soil total nitrogen online detection device, the issues of versatility and efficiency of the device in different regions were resolved, enabling real-time detection and simplified procedures.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANDONG UNIV OF TECH
- Filing Date
- 2025-01-13
- Publication Date
- 2026-07-14
Smart Images
Figure CN122385491A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of total nitrogen detection technology, specifically to a mobile online soil total nitrogen detection device. Background Technology
[0002] Traditional methods for detecting total nitrogen in soil primarily rely on chemical approaches, which are cumbersome, time-consuming, inefficient, and polluting, failing to meet the demands of modern agriculture. In contrast, spectroscopic analysis, a modern analytical technique, integrates the advantages of spectroscopy, chemometrics, and computer technology, enabling rapid, efficient, and non-destructive detection of total nitrogen content in soil. Spectroscopic analysis of soil total nitrogen content is based on the different reflective effects of chemical bonds and functional groups within the material on different wavelengths of light, thus allowing for the analysis of total nitrogen content values.
[0003] However, most soil total nitrogen detection devices currently developed based on spectral principles require the establishment of a prediction model based on the current soil spectral data for each detection of soil total nitrogen content. This prediction model is then used to predict the current soil total nitrogen content. During the prediction, a custom light source and a custom bandpass filter are used to obtain the spectral characteristic data of the current soil at the characteristic wavelength as the model input.
[0004] Due to the differences between soils, on the one hand, the prediction model established for the current soil is difficult to reuse in soils in other regions, resulting in poor equipment versatility; on the other hand, when predicting soils in different regions, it is necessary to change the customized light source and customized bandpass filter to obtain the spectral characteristic data of the corresponding soil, which makes the equipment detection steps cumbersome and the detection efficiency low.
[0005] In summary, existing soil total nitrogen testing equipment suffers from poor versatility, cumbersome testing procedures, and low testing efficiency when dealing with soil total nitrogen testing in different regions. This is because the prediction models are difficult to reuse and the optical elements used to acquire spectral characteristic data need to be constantly replaced. Summary of the Invention
[0006] This application provides a portable online soil total nitrogen detection device to solve the technical problems of existing soil total nitrogen detection devices when facing soil total nitrogen detection in different regions. These problems include poor device versatility, cumbersome detection steps, and low detection efficiency because the prediction model is difficult to reuse and the optical elements for acquiring spectral feature data need to be constantly replaced.
[0007] This application provides a mobile online soil total nitrogen detection device, including: a small laboratory detection structure and a field mobile detection structure; The small laboratory testing structure is installed on the Daejeon mobile testing structure; In laboratory testing mode, the small laboratory testing structure is used for: First full-spectral reflectance data of soil samples were collected; The first total nitrogen prediction model for the region to which the soil sample belongs is called from the total nitrogen prediction model library. The first reflectance feature data is extracted from the first full-spectral reflectance data, and the first reflectance feature data is input into the first total nitrogen prediction model to obtain the total nitrogen content value of the soil sample. In the real-time detection mode of Daegu, the Daegu motion detection structure is used for: The small laboratory testing structure is moved in the field, and field soil of a preset depth is provided for the small laboratory testing structure. The small laboratory testing structure is used for: Collect the second full-spectral reflectance data of the field soil; The second total nitrogen prediction model for the field soil region in the total nitrogen prediction model library is called, the second reflectance feature data is extracted from the second full-spectral reflectance data, and the second reflectance feature data is input into the second total nitrogen prediction model to obtain the total nitrogen content value of the field soil. The total nitrogen prediction model library is constructed based on total nitrogen prediction models for soils in multiple regions. The total nitrogen prediction model for soils in any region is trained based on historical reflectance characteristic data and historical measured values of total nitrogen content in the soils of that region.
[0008] In one embodiment, the small laboratory testing structure includes a detection probe, a miniature spectrometer, and a host computer system; The field mobile detection structure includes a mobile platform and a tillage plow. The tillage plow is located at the bottom of the mobile platform, and a through hole is provided between the tillage plow and the detection probe. In the laboratory testing mode: The detection probe is used to collect the first reflected light signal of the soil sample under a single-hole cold light source and transmit the first reflected light signal to the miniature spectrometer; The miniature spectrometer is used to convert the first reflected light signal into the first full-spectrum reflectance data and transmit the first full-spectrum reflectance data to the host computer system. The host computer system is used to call the first total nitrogen prediction model of the region to which the soil sample belongs in the total nitrogen prediction model library, extract the first reflectance feature data from the first full-spectral reflectance data, and input the first reflectance feature data into the first total nitrogen prediction model to obtain the total nitrogen content value of the soil sample; In Daejeon real-time detection mode: The mobile platform is used to move the small laboratory testing structure in the field, and the deep-loosening plow is used to cut into the field soil at the preset depth; The detection probe is used to extend from the through hole into the interior of the deep plow, collect the second reflected light signal of the field soil under a single-hole cold light source, and transmit the second reflected light signal to the micro spectrometer; The miniature spectrometer is used to convert the second reflected light signal into the second full-spectrum reflectance data and transmit the second full-spectrum reflectance data to the host computer system. The host computer system is used to call the second total nitrogen prediction model of the field soil region in the total nitrogen prediction model library, extract the second reflectance feature data from the second full-spectral reflectance data, and input the second reflectance feature data into the second total nitrogen prediction model to obtain the total nitrogen content value of the field soil.
[0009] In one embodiment, the small laboratory testing structure further includes a first motor, a ball screw, and multiple bevel gears; The multiple bevel gears are connected in sequence by changing their directions to form a connecting gear; One end of the connecting gear is fixedly connected to the first motor, the other end of the connecting gear is fixedly connected to one end of the ball screw, and one end of the detection probe is fixedly connected to the other end of the ball screw; The first motor, the connecting gear, and the ball screw are used to drive the transmission in sequence to adjust the height of the detection probe.
[0010] In one embodiment, the Daejeon movement detection structure further includes an electric actuator; One end of the electric push rod is fixedly connected to the slack plow, and the other end of the electric push rod is fixedly connected to the mobile platform; The electric push rod is used to push the deep-plowing plow to cut into the field soil at the preset depth.
[0011] In one embodiment, the field moving detection structure further includes a second motor, a crank rocker arm, and a soil scraper; The second motor, the crank rocker arm, and the soil scraper are disposed inside the loosening plow; One end of the crank rocker arm is fixedly connected to the second motor, and the other end of the crank rocker arm is fixedly connected to the soil scraper brush; The second motor and the crank rocker arm are used to drive the scraper brush to clean the detection probe that extends into the interior of the loosening plow.
[0012] In one embodiment, extracting first reflectance feature data from the first full-spectrum reflectance data includes: Based on the first characteristic wavelength input by the user, when the first full-spectrum reflectance data is updated, the first reflectance feature data corresponding to the first characteristic wavelength is extracted from the first full-spectrum reflectance data.
[0013] In one embodiment, extracting second reflectance feature data from the second full-spectrum reflectance data includes: Based on the second characteristic wavelength input by the user, when the second full-spectrum reflectance data is updated, the second reflectance feature data corresponding to the second characteristic wavelength is extracted from the second full-spectrum reflectance data.
[0014] In one embodiment, the host computer system is further configured to: Obtain the GPS information of the Daejeon mobile detection structure during its movement; The GPS information and its corresponding total nitrogen content value are displayed and saved as a txt file.
[0015] In one embodiment, the host computer system is further configured to: Mark the geographical location corresponding to the GPS information on the electronic map; A total nitrogen detection trajectory is generated based on the geographical location.
[0016] In one embodiment, the total nitrogen prediction model for soil in any region is obtained based on the following steps: Collect target soil samples from the described area in the field; The total nitrogen content of the target soil was determined using chemical assay methods. The target soil was collected using a hyperspectral imager, which collected first reflectance data in a first wavelength range and second reflectance data in a second wavelength range; the first wavelength range was 400 nm to 1100 nm, and the second wavelength range was 1000 nm to 2500 nm. After correcting the reflection data within the overlapping wavelength range based on the first reflection data and the second reflection data, the corrected third reflection data within the first wavelength range is obtained; The third reflection data and the measured total nitrogen content are used as the model dataset, and the model dataset is divided into a training set and a prediction set; The third reflection data in the training set is denoised to obtain denoised reflection data; Feature wavelengths are selected from the noise-reduced reflection data, and the reflection data corresponding to the feature wavelengths are extracted to obtain the target feature reflection data; The target feature reflectance data and the measured total nitrogen content are used as a new training set. The new training set is used to train any regression model until the regression model converges, thus obtaining the prediction model to be verified. The prediction set is input into the prediction model to be verified to obtain the predicted value of total nitrogen content; If the error between the predicted total nitrogen content and the measured total nitrogen content does not meet expectations, the noise reduction method and / or feature wavelength screening method are changed, and the process of denoising the third reflection data in the training set is repeated until the error between the predicted total nitrogen content and the measured total nitrogen content meets expectations. At this point, the prediction model to be verified is determined as the total nitrogen prediction model.
[0017] The mobile soil total nitrogen online detection device provided in this application includes: a small laboratory detection structure and a field mobile detection structure. The small laboratory detection structure is installed on the field mobile detection structure. In laboratory detection mode, the small laboratory detection structure is used to collect the first full-spectral reflectance data of the soil sample, call the first total nitrogen prediction model of the region to which the soil sample belongs in the total nitrogen prediction model library, extract the first reflectance feature data from the first full-spectral reflectance data, and input the first reflectance feature data into the first total nitrogen prediction model to obtain the total nitrogen content value of the soil sample. In field real-time detection mode, the field mobile detection structure is used to move the small laboratory detection structure in the field and provide the small laboratory detection structure with field soil at a preset depth. The small laboratory detection structure is used to collect the second full-spectral reflectance data of the field soil, call the second total nitrogen prediction model of the region to which the field soil belongs in the total nitrogen prediction model library, extract the second reflectance feature data from the second full-spectral reflectance data, and input the second reflectance feature data into the second total nitrogen prediction model to obtain the total nitrogen content value of the field soil. The total nitrogen prediction model library is constructed based on total nitrogen prediction models for soils from multiple regions. The total nitrogen prediction model for any given region is trained using historical reflectance characteristic data and historical measured values of total nitrogen content in that region. The device in this application provides two detection modes. Firstly, regardless of whether it's in a small-scale laboratory testing mode or a real-time field testing mode, a total nitrogen prediction model matching the current region's soil spectral characteristics can be dynamically loaded from the pre-built total nitrogen prediction model library, eliminating the need to construct a specific prediction model and improving the device's versatility in different regions. Secondly, regardless of whether it's in a small-scale laboratory testing mode or a real-time field testing mode, reflectance characteristic data is not directly acquired using a customized light source and bandpass filter during the spectral data acquisition stage. Instead, full-spectral reflectance data of the soil is acquired first, and algorithms can then be used to extract the full-spectral reflectance data from the data. By extracting reflection characteristic data, there is no need to use custom light sources and custom bandpass filters, and therefore no need to replace these optical components to obtain reflection characteristic data of soils in different regions, simplifying the equipment detection steps and improving equipment detection efficiency. On the other hand, in the real-time field detection mode, the field mobile detection structure can drive the small laboratory detection structure to move in the field and provide the small laboratory detection structure with a preset depth of field soil for detection. Therefore, during the movement, the small laboratory detection structure can detect the total nitrogen content value of field soil in real time, without waiting for manual sampling and then sending it to the laboratory for testing, further improving the equipment detection efficiency.
[0018] In summary, the device proposed in this application can effectively improve the versatility of the device, simplify the detection steps, and improve detection efficiency by pre-constructing a total nitrogen prediction model library and providing dynamic calling and loading functions, collecting soil full-spectrum reflectance data and then using algorithms to extract reflectance feature data, and adopting a real-time mobile detection method that integrates a small laboratory detection structure and a field mobile detection structure. Attached Figure Description
[0019] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1 This is one of the structural schematic diagrams of the portable online soil total nitrogen detection device provided in the embodiments of this application; Figure 2 This is the second structural schematic diagram of the portable online soil total nitrogen detection device provided in the embodiments of this application; Figure 3 This is a system architecture diagram of the portable online soil total nitrogen detection device provided in the embodiments of this application; Figure 4 This is a schematic diagram of the small laboratory testing structure of the portable online soil total nitrogen detection device provided in this application embodiment; Figure 5 This is a top view of the mobile soil total nitrogen online detection device provided in this application embodiment when the detection probe extends to the loosening plow; Figure 6 This is a flowchart illustrating the method for constructing a total nitrogen prediction model in a portable online soil total nitrogen detection device provided in this application embodiment; Figure 7 This is a system flowchart of the portable online soil total nitrogen detection device provided in the embodiments of this application; Figure 8 This is a correlation coefficient diagram between the detected values and actual measured values of the portable online soil total nitrogen detection device provided in this application embodiment.
[0021] Figure label: 1-Small laboratory testing structure; 11-Detection probe; 12-First motor; 13-Ball screw; 14-Bevel gear; 15-Motor driver; 16-Box; 2-Field mobile testing structure; 21-Mobile platform; 22-Deepening plow; 23-Electric actuator; 24-Second motor; 25-Crank rocker arm; 26-Soil scraper; 27-Three-point suspension mechanism; 28-Handwheel; 29-Universal caster; 3-Soil sample. Detailed Implementation
[0022] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0023] It should be noted that in the description of the embodiments of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. The terms "upper," "lower," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application. Unless otherwise expressly specified and limited, the terms "installed," "connected," and "linked" should be interpreted broadly, for example, they can be fixed connections, detachable connections, or integral connections; they can be mechanical connections or electrical connections; they can be direct connections or indirect connections through an intermediate medium; and they can be internal connections between two elements. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.
[0024] The terms "first," "second," etc., used in this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, a first object can be one or more. Furthermore, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects have an "or" relationship.
[0025] Reference Figures 1 to 2 This application provides a mobile online soil total nitrogen detection device, which may include: a small laboratory detection structure 1 and a field mobile detection structure 2; The small laboratory testing structure 1 is installed on the Daejeon mobile testing structure 2; In laboratory testing mode, the small laboratory testing structure 1 is used for: First full-spectral reflectance data of soil samples were collected; The first total nitrogen prediction model for the region to which the soil sample belongs is called from the total nitrogen prediction model library. The first reflectance feature data is extracted from the first full-spectral reflectance data and input into the first total nitrogen prediction model to obtain the total nitrogen content value of the soil sample. In the real-time detection mode of Daejeon, the Daejeon motion detection structure 2 is used for: The small laboratory testing structure 1 is moved in the field and provides a preset depth of field soil for the small laboratory testing structure 1. Small laboratory testing structure 1, used for: Collect second full-spectral reflectance data of field soil; The second total nitrogen prediction model for the field soil region in the total nitrogen prediction model library is called. The second reflectance feature data is extracted from the second full-spectral reflectance data and input into the second total nitrogen prediction model to obtain the total nitrogen content value of the field soil. The total nitrogen prediction model library is built based on total nitrogen prediction models for soils in multiple regions. The total nitrogen prediction model for soils in any region is trained based on historical reflectance characteristic data and historical measured values of total nitrogen content in that region.
[0026] The portable online soil total nitrogen detection device provided in this embodiment has two advantages. First, regardless of whether it is in a small laboratory testing mode or a real-time field testing mode, it can dynamically call and load a total nitrogen prediction model that matches the spectral characteristics of the soil in the current region from a pre-built total nitrogen prediction model library, eliminating the need to build a specific prediction model and improving the device's versatility in different regions. Second, regardless of whether it is in a small laboratory testing mode or a real-time field testing mode, it does not directly acquire reflectance characteristic data using a customized light source and a customized bandpass filter during the spectral data acquisition stage. Instead, it first acquires the full spectral reflectance data of the soil, and then uses algorithms to extract the full spectral reflectance data. By extracting reflection feature data from the radiation data, there is no need to use custom light sources and custom bandpass filters, and therefore no need to replace these optical components to obtain reflection feature data of soil in different regions, simplifying the equipment detection steps and improving equipment detection efficiency. On the other hand, in the real-time field detection mode, the field mobile detection structure can drive the small laboratory detection structure to move in the field and provide the small laboratory detection structure with a preset depth of field soil for detection. Therefore, during the movement, the small laboratory detection structure can detect the total nitrogen content value of the field soil in real time, without waiting for manual sampling and then sending it to the laboratory for testing, further improving the equipment detection efficiency.
[0027] In summary, the device in this embodiment effectively improves the versatility of the equipment and simplifies the detection steps by pre-constructing a total nitrogen prediction model library and providing dynamic calling and loading functions, collecting soil full-spectrum reflectance data and then using algorithms to extract reflectance feature data, and adopting a real-time mobile detection method that integrates a small laboratory detection structure and a field mobile detection structure.
[0028] Reference Figures 1 to 5 In one embodiment, the small laboratory testing structure 1 includes a detection probe 11, a miniature spectrometer, and a host computer system; The field mobile detection structure 2 includes a mobile platform 21 and a tillage plow 22. The tillage plow 22 is located at the bottom of the mobile platform 21, and a through hole is provided between the tillage plow 22 and the detection probe 11. In laboratory testing mode: The detection probe 11 is used to collect the reflected light signal of soil sample 3 under a single-hole cold light source and transmit the reflected light signal to a miniature spectrometer; The miniature spectrometer is used to convert the reflected light signal into first full-spectrum reflectance data and transmit the first full-spectrum reflectance data to the host computer system. The host computer system is used to call the first total nitrogen prediction model of the region to which the soil sample belongs in the total nitrogen prediction model library, extract the first reflectance feature data from the first full-spectral reflectance data, and input the first reflectance feature data into the first total nitrogen prediction model to obtain the total nitrogen content value of the soil sample; In Daejeon real-time detection mode: The mobile platform 21 is used to move the small laboratory testing structure 1 in the field, and the deep-loosening plow 22 is used to cut into the field soil at a preset depth. The detection probe 11 is used to extend from the through hole into the interior of the deep plow 22, collect the second reflected light signal of the field soil under a single-hole cold light source, and transmit the second reflected light signal to the micro spectrometer; The miniature spectrometer is used to convert the second reflected light signal into second full-spectrum reflectance data and transmit the second full-spectrum reflectance data to the host computer system. The host computer system is used to call the second total nitrogen prediction model of the field soil region in the total nitrogen prediction model library, extract the second reflectance feature data from the second full-spectral reflectance data, and input the second reflectance feature data into the second total nitrogen prediction model to obtain the total nitrogen content value of the field soil.
[0029] The detection probe 11 is composed of a Y-shaped optical fiber and sapphire glass. A single-aperture cold light source is connected to one end of the branch of the Y-shaped optical fiber. The emitted light is injected into the detection probe 11 through one end of the branch and is perpendicularly irradiated onto the soil to be tested at the bottom of the detection probe 11, i.e., the soil sample or field soil. The micro spectrometer is connected to the other end of the branch of the Y-shaped optical fiber, receives and decomposes the light signal transmitted by the detection probe 11, and transmits the obtained full-spectrum reflectance data to the host computer system.
[0030] The single-aperture cold light source can be a single-aperture halogen cold light source, and the miniature spectrometer can be a visible-shortwave near-infrared miniature spectrometer. Soil testing is conducted in a dark environment, whether in laboratory testing mode or in real-time field testing mode.
[0031] This embodiment uses a single-hole halogen cold light source as the detection light source, which has better output stability than traditional LED light sources, lower operating costs than laser light sources, can work continuously in temperatures ranging from -10 degrees Celsius to -45 degrees Celsius, has a long service life, and is easy to replace; at the same time, a visible-shortwave near-infrared miniature spectrometer is used to receive light signals, which can complete the acquisition of full-spectrum reflectance data from 400 nanometers to 1100 nanometers.
[0032] Reference Figure 4 In one embodiment, the small laboratory testing structure 1 further includes a first motor 12, a ball screw 13, and a plurality of bevel gears 14; Multiple bevel gears 14 are connected in sequence by changing their orientation to form a connecting gear; One end of the connecting gear is fixedly connected to the first motor 12, and the other end of the connecting gear is fixedly connected to one end of the ball screw 13. One end of the detection probe 11 is fixedly connected to the other end of the ball screw 13. The first motor 12, connecting gear and ball screw 13 are used for sequential transmission to adjust the height of the detection probe 11.
[0033] Furthermore, the small laboratory testing structure 1 also includes a motor driver 15.
[0034] The motor driver 15 drives the first motor 12 to rotate, which in turn drives the connecting gear to rotate. The connecting gear then drives the ball screw 13 to rise and fall, thereby adjusting the height of the detection probe 11 and, consequently, the distance between the detection probe 11 and the soil to be tested. The first motor 12 can be a stepper motor. The diameter of the bevel gear 14 can be selected according to specific needs and is not limited here. In this embodiment, the diameter of the bevel gear 14 can be set to 8 mm.
[0035] This embodiment employs multiple bevel gears in a reversing configuration, with the ball screw being raised and lowered under the drive of a first motor to adjust the distance between the detection probe and the soil to be tested. This avoids the space occupation caused by using traditional coupling transmissions and achieves miniaturization of the laboratory testing structure.
[0036] Reference Figure 1 The Daejeon mobile detection structure 2 also includes an electric actuator 23; One end of the electric actuator 23 is fixedly connected to the sloshing plow 22, and the other end of the electric actuator 23 is fixedly connected to the moving platform 21; The electric push rod 23 is used to push the loosening plow 22 to cut into the field soil at a preset depth.
[0037] This embodiment can automatically control the stroke of the electric pusher, thereby pushing the loosening plow to cut into the field soil at a preset depth, realizing the automatic acquisition of field soil at different depths. Compared with the traditional method of setting a fixed height borehole to adjust the plowing depth, which has problems such as discontinuous adjustment, inconvenience in practical application and insufficient accuracy, this embodiment can conveniently and continuously adjust the detection depth, thereby realizing continuous acquisition and real-time detection of field soil at different depths, and improving detection accuracy.
[0038] Reference Figure 5 In one embodiment, the field moving detection structure 2 further includes a second motor 24, a crank rocker arm 25, and a soil scraper 26; The second motor 24, crank rocker 25 and soil scraper 26 are installed inside the loosening plow 22; One end of the crank rocker arm 25 is fixedly connected to the second motor 24, and the other end of the crank rocker arm 25 is fixedly connected to the soil scraper 26. The second motor 24 and the crank rocker arm 25 are used to drive the scraper brush 26 to clean the detection probe 11 that extends into the interior of the loosening plow 22.
[0039] The scraper brush 26 can be made of soft rubber or other materials; no specific restrictions are made here.
[0040] In field testing practice, a small amount of field soil particles may adhere to the surface of the detection probe 11, affecting the detection effect. In this embodiment, the second motor 24 drives the crank rocker 25 to continuously cycle and reciprocate, driving the soil scraper 26 to clean the surface of the detection probe 11 repeatedly, which can avoid the adverse effects of field soil particles on the detection.
[0041] In one embodiment, extracting first reflectance feature data from first full-spectral reflectance data may include: Based on the first characteristic wavelength input by the user, when the first full-spectrum reflectance data is updated, the first reflectance feature data corresponding to the first characteristic wavelength is extracted from the first full-spectrum reflectance data; Similarly, extracting second reflectance feature data from the second full-spectrum reflectance data can include: Based on the second characteristic wavelength input by the user, when the second full-spectrum reflectance data is updated, the second reflectance feature data corresponding to the second characteristic wavelength is extracted from the second full-spectrum reflectance data.
[0042] This embodiment is based on extracting updated reflectance feature data based on the characteristic wavelength input by the user when the full-spectrum reflectance data is updated, and can detect the soil total nitrogen content value in real time in response to changes in reflectance feature data.
[0043] In one embodiment, the host computer system is also used for: The GPS information of the mobile detection structure in Datian during its movement is obtained. On the one hand, the GPS information and its corresponding total nitrogen content value are displayed and saved as a txt file. On the other hand, the geographical location corresponding to the GPS information is marked on an electronic map, and a total nitrogen detection trajectory is generated based on the geographical location.
[0044] This embodiment acquires GPS information, displays and saves it in combination with the corresponding total nitrogen content value, and generates a total nitrogen detection trajectory based on the corresponding geographical location, which facilitates subsequent data traceability and analysis.
[0045] Reference Figure 6 In one embodiment, a total nitrogen prediction model for soil in any region can be obtained based on the following steps: 601. Collect target soil samples in the field; 602. The total nitrogen content of the target soil was determined using chemical methods. 603. Use a hyperspectral analyzer to collect the first reflectance data of the target soil in the first wavelength range and the second reflectance data in the second wavelength range, respectively; The first wavelength range is 400 nanometers to 1100 nanometers, and the second wavelength range is 1000 nanometers to 2500 nanometers; 604. Based on the first reflection data and the second reflection data, the reflection data within the overlapping wavelength range is corrected, and the corrected third reflection data within the first wavelength range is obtained. 605. Use the third reflection data and the measured total nitrogen content as the model dataset, and divide the model dataset into a training set and a prediction set; 606. Denoise the third reflection data in the training set to obtain denoised reflection data; 607. Select characteristic wavelengths from the noise-reduced reflection data and extract the reflection data corresponding to the characteristic wavelengths to obtain the target characteristic reflection data; 608. Use the target feature reflectance data and the measured total nitrogen content as a new training set, and use the new training set to train any regression model until the regression model converges to obtain the prediction model to be verified. 609. Input the prediction set into the prediction model to be validated to obtain the predicted value of total nitrogen content; 610. If the error between the predicted value of total nitrogen content and the measured value of total nitrogen content does not meet expectations, then after changing the noise reduction method and / or the characteristic wavelength screening method, return to step 606. 611. If the error between the predicted value of total nitrogen content and the measured value of total nitrogen content meets the expectations, the prediction model to be verified at this time shall be determined as the total nitrogen prediction model.
[0046] In step 602, the target soil can be dried and then ground. Weigh 2.0 g of the dried and ground target soil, add a catalyst mixture of potassium sulfate and copper sulfate crystals, and put it into a test tube. Then add concentrated sulfuric acid to the test tube and place the test tube in a digestion furnace. After digestion, cool for 0.5 hours, and finally distill using a fully automatic Kjeldahl nitrogen analyzer to obtain the measured value of the total nitrogen content of the target soil.
[0047] In step 603, the hyperspectral imager can use two lenses to measure the first reflection data in the first wavelength range and the second reflection data in the second wavelength range, respectively.
[0048] In step 604, there is an overlapping wavelength range of 1000 nm to 1100 nm between the first wavelength range and the second wavelength range. The first reflection data and the second reflection data within the overlapping wavelength range can be used to correct each other to obtain corrected reflection data within the overlapping wavelength range. After integrating the corrected reflection data with the original first reflection data within the wavelength range of 400 nm to 1000 nm, the corrected third reflection data within the wavelength range of 400 nm to 1100 nm is obtained.
[0049] In step 605, the model dataset can be divided into a training set and a prediction set in a 3:1 ratio.
[0050] In step 606, the third reflection data may be affected by factors such as noise and baseline drift. Therefore, it is necessary to denoise the third reflection data in the training set to avoid affecting the model training effect.
[0051] The third reflection data can be denoised using one or a combination of two of the following methods: Savitzky-Gora filtering, standard normal variable transformation, multivariate scattering correction, and first derivative. No limitation is made here. In this embodiment, a combination algorithm of multivariate scattering correction and first derivative is used to denoise the third reflection data.
[0052] In step 607, there may be useless and redundant information in the noise-reduced reflection data. Therefore, a competitive adaptive weighted algorithm can be used to select multiple feature wavelengths that contribute significantly to the model prediction from these noise-reduced reflection data. Then, the reflection data corresponding to these feature wavelengths can be extracted from the noise-reduced reflection data as target reflection feature data. In this embodiment, 15 feature wavelengths can be selected.
[0053] In steps 608 to 611, the regression model can be a partial least squares regression model. The final total nitrogen prediction model has a goodness of fit of 0.8767 to the training set and a goodness of fit of 0.8122 to the prediction set.
[0054] This embodiment establishes a regression equation between the measured total nitrogen content of the target soil and the reflectance data of characteristic wavelengths in the visible-shortwave near-infrared spectrum using a partial least squares regression model, and finally obtains a total nitrogen prediction model. During the model construction process, data correction and noise reduction were performed, resulting in a final total nitrogen prediction model with good detection accuracy. Experiments have shown that the equipment detection accuracy can reach 0.89 in laboratory detection mode and 0.87 in field real-time detection mode.
[0055] Reference Figure 4 In one embodiment, the small laboratory testing structure 1 further includes a housing 16, within which all testing mechanisms are integrated. The overall dimensions of the housing 16 are 300 mm × 300 mm × 350 mm, exhibiting a small overall size and high integration, facilitating rapid laboratory testing. The testing process in laboratory testing mode can be briefly described as follows: 1. Control the detection probe 11 to move to its highest position; 2. Spread soil sample 3 evenly in the test dish, open the chamber door, place the test dish on the internal test platform, and close the chamber door; 3. Set the integration time, control the detection probe 11 to move to a certain distance above the soil sample 3, select single detection and start the detection, observe the spectral curve on the display interface, and determine whether the distance meets the detection conditions. 4. If the light intensity in the spectral curve is weak, it indicates that the detection distance is too large. The distance between the detection probe 11 and the soil sample 3 can be adjusted multiple times. When the light intensity is moderate, the distance at this time is determined to be the optimal distance for capturing full-spectral reflectance data. Select and load the appropriate total nitrogen prediction model to obtain the total nitrogen content value of the soil sample 3.
[0056] Reference Figures 1 to 2In one embodiment, the Daejeon mobile detection structure 2 further includes a three-point suspension mechanism 27, a handwheel 28, and a caster wheel 29; the three-point suspension mechanism 27 is mounted on the mobile platform 21, and the handwheel 28 and caster wheel 29 are mounted at the head of the mobile platform 21. The handwheel 28 is used to adjust the lifting and lowering of the caster wheel 29; the detection process in the Daejeon real-time detection mode can be briefly described as follows: 1. The control detection probe 11 extends into the tillage plow 22 and connects the three-point suspension mechanism 27 to the rear of the tractor; 2. Adjust handwheel 28 to raise caster wheel 29; 3. Set the points accumulation period; 4. The tractor tows the mobile platform 21 a certain distance, and at the same time, the electric push rod 23 is extended and retracted, so that the deep-loosening plow 22 cuts into the field soil to reach the corresponding detection depth; 5. Start the test, and the tractor begins to tow the equipment to officially begin work; 6. Obtain GPS information and total nitrogen content values, and record them in the corresponding area of the host computer system.
[0057] Reference Figure 3 and Figure 7 In one embodiment, the small laboratory testing structure also includes a lower-level computer system: The host computer system mainly uses self-developed software to set various parameters of the micro spectrometer, drive the micro spectrometer to capture soil spectral reflectance data, temporarily store the data in a virtual array container, call the model and extract reflectance feature data to obtain the total nitrogen content value, visualize and save the data, receive GPS information collected by the lower computer, draw the total nitrogen detection trajectory map and display and record it in combination with the total nitrogen content value; The lower-level computer system has three main functions: First, in laboratory testing mode, it controls the lifting system to adjust the distance between the detection probe and the soil sample to obtain the best soil spectral data; second, in real-time field testing mode, it obtains the current GPS information; third, in real-time field testing mode, it controls the electric push rod to make the loosening plow reach the testing depth; and fourth, it establishes a wireless connection with the upper-level computer system to realize data interaction.
[0058] Specifically, refer to Figure 7 The workflow of the host system and the slave system is as follows: Figure 7 The descriptions are as follows from left to right. The following numbers do not specify the workflow order of the host computer system and the slave computer system: 1. The host computer system sets the parameters of the miniature spectrometer through serial communication and determines whether the slave computer system is turned on. If it is turned on, it connects with the slave computer system. After the connection is successful, it sends action commands to the slave computer system and receives the GPS information returned by the slave computer system. 2. The host computer system dynamically loads the total nitrogen prediction model library. Users can subsequently choose to call the local model in the total nitrogen prediction model library. The original model can be replaced without modifying the software source code, thus improving the versatility of the equipment. The system receives the characteristic wavelengths manually input by the user and marks them. The spectral reflectance data corresponding to the marked characteristic wavelengths will be extracted separately with each update. 3. The host computer system initiates spectral measurement, receives the user-selected reflectance measurement mode, and acquires the standard white spectrum and dark spectrum. The standard white spectrum can be obtained using a standard white board with 99% reflectance, and the dark spectrum can be obtained using a standard black board with 0% reflectance. Black and white calibration is performed based on the standard white and dark spectra, enabling the host computer system to acquire spectral reflectance data. The integration time is set. If the user selects a single measurement, spectral reflectance data is acquired; if the user selects multiple measurements, the data acquisition interval for multiple measurements is set before acquiring spectral reflectance data, and the spectral reflectance data is saved to the data temporary storage area (i.e.,...). Figure 3 The system uses a virtual array container and a MySQL database. If the spectral reflectance data in the data buffer is updated, an asynchronous task is triggered. The host computer system calls the adaptation model and extracts the spectral reflectance data corresponding to the characteristic wavelength as the reflectance feature data input to the model to obtain the total nitrogen content value. At the same time, the extracted reflectance feature data is uploaded to the cloud monitoring system through the message queue telemetry transmission protocol. The total nitrogen content value is also uploaded to the cloud monitoring system through the message queue telemetry transmission protocol. The host computer system combines the total nitrogen content value with the GPS information returned by the slave system to form a data combination and displays it. The data combination is numbered and saved as a txt file. The data combination is saved to the MySQL database and uploaded to the cloud monitoring system through the message queue telemetry transmission protocol. The host computer system can obtain the GPS information returned by the slave system every 5 seconds to analyze the location information and call the electronic map to mark the location, thereby generating a detection trajectory. The GPS information returned by the slave system is also uploaded to the cloud monitoring system through the message queue telemetry transmission protocol.
[0059] Following the above workflow, the equipment described in this application can achieve better testing results, as per the reference. Figure 8 The correlation coefficient r between the soil total nitrogen content value (i.e., the detected value) and the actual measured value of soil total nitrogen content obtained by the equipment in this application can reach 0.8862, which shows high detection accuracy. The units for both the actual measured value and the detected value are grams per kilogram.
[0060] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.
Claims
1. A portable online soil total nitrogen detection device, characterized in that, include: Small-scale laboratory testing structures and large-scale mobile testing structures; The small laboratory testing structure is installed on the Daejeon mobile testing structure; In laboratory testing mode, the small laboratory testing structure is used for: First full-spectral reflectance data of soil samples were collected; The first total nitrogen prediction model for the region to which the soil sample belongs is called from the total nitrogen prediction model library. The first reflectance feature data is extracted from the first full-spectral reflectance data, and the first reflectance feature data is input into the first total nitrogen prediction model to obtain the total nitrogen content value of the soil sample. In the real-time detection mode of Daegu, the Daegu motion detection structure is used for: The small laboratory testing structure is moved in the field, and field soil of a preset depth is provided for the small laboratory testing structure. The small laboratory testing structure is used for: Collect the second full-spectral reflectance data of the field soil; The second total nitrogen prediction model for the field soil region in the total nitrogen prediction model library is called, the second reflectance feature data is extracted from the second full-spectral reflectance data, and the second reflectance feature data is input into the second total nitrogen prediction model to obtain the total nitrogen content value of the field soil. The total nitrogen prediction model library is constructed based on total nitrogen prediction models for soils in multiple regions. The total nitrogen prediction model for soils in any region is trained based on historical reflectance characteristic data and historical measured values of total nitrogen content in the soils of that region.
2. The portable online soil total nitrogen detection device according to claim 1, characterized in that, The small laboratory testing structure includes a detection probe, a miniature spectrometer, and a host computer system. The field mobile detection structure includes a mobile platform and a tillage plow. The tillage plow is located at the bottom of the mobile platform, and a through hole is provided between the tillage plow and the detection probe. In the laboratory testing mode: The detection probe is used to collect the first reflected light signal of the soil sample under a single-hole cold light source and transmit the first reflected light signal to the miniature spectrometer; The miniature spectrometer is used to convert the first reflected light signal into the first full-spectrum reflectance data and transmit the first full-spectrum reflectance data to the host computer system. The host computer system is used to call the first total nitrogen prediction model of the region to which the soil sample belongs in the total nitrogen prediction model library, extract the first reflectance feature data from the first full-spectral reflectance data, and input the first reflectance feature data into the first total nitrogen prediction model to obtain the total nitrogen content value of the soil sample; In Daejeon real-time detection mode: The mobile platform is used to move the small laboratory testing structure in the field, and the deep-loosening plow is used to cut into the field soil at the preset depth; The detection probe is used to extend from the through hole into the interior of the deep plow, collect the second reflected light signal of the field soil under a single-hole cold light source, and transmit the second reflected light signal to the micro spectrometer; The miniature spectrometer is used to convert the second reflected light signal into the second full-spectrum reflectance data and transmit the second full-spectrum reflectance data to the host computer system. The host computer system is used to call the second total nitrogen prediction model of the field soil region in the total nitrogen prediction model library, extract the second reflectance feature data from the second full-spectral reflectance data, and input the second reflectance feature data into the second total nitrogen prediction model to obtain the total nitrogen content value of the field soil.
3. The portable online soil total nitrogen detection device according to claim 2, characterized in that, The small laboratory testing structure also includes a first motor, a ball screw, and multiple bevel gears; The multiple bevel gears are connected in sequence by changing their directions to form a connecting gear; One end of the connecting gear is fixedly connected to the first motor, the other end of the connecting gear is fixedly connected to one end of the ball screw, and one end of the detection probe is fixedly connected to the other end of the ball screw; The first motor, the connecting gear, and the ball screw are used to drive the transmission in sequence to adjust the height of the detection probe.
4. The portable online soil total nitrogen detection device according to claim 2, characterized in that, The Daejeon mobile detection structure also includes an electric actuator; One end of the electric push rod is fixedly connected to the slack plow, and the other end of the electric push rod is fixedly connected to the mobile platform; The electric push rod is used to push the deep-plowing plow to cut into the field soil at the preset depth.
5. The portable online soil total nitrogen detection device according to claim 2, characterized in that, The field mobile detection structure also includes a second motor, a crank rocker arm, and a soil scraper; The second motor, the crank rocker arm, and the soil scraper are disposed inside the loosening plow; One end of the crank rocker arm is fixedly connected to the second motor, and the other end of the crank rocker arm is fixedly connected to the soil scraper brush; The second motor and the crank rocker arm are used to drive the scraper brush to clean the detection probe that extends into the interior of the loosening plow.
6. The portable online soil total nitrogen detection device according to claim 1, characterized in that, Extracting first reflectance feature data from the first full-spectrum reflectance data includes: Based on the first characteristic wavelength input by the user, when the first full-spectrum reflectance data is updated, the first reflectance feature data corresponding to the first characteristic wavelength is extracted from the first full-spectrum reflectance data.
7. The portable online soil total nitrogen detection device according to claim 1, characterized in that, Extracting second reflectance feature data from the second full-spectrum reflectance data includes: Based on the second characteristic wavelength input by the user, when the second full-spectrum reflectance data is updated, the second reflectance feature data corresponding to the second characteristic wavelength is extracted from the second full-spectrum reflectance data.
8. The portable online soil total nitrogen detection device according to claim 2, characterized in that, The host computer system is also used for: Obtain the GPS information of the Daejeon mobile detection structure during its movement; The GPS information and its corresponding total nitrogen content value are displayed and saved as a txt file.
9. The portable online soil total nitrogen detection device according to claim 8, characterized in that, The host computer system is also used for: Mark the geographical location corresponding to the GPS information on the electronic map; A total nitrogen detection trajectory is generated based on the geographical location.
10. The portable online soil total nitrogen detection device according to claim 1, characterized in that, The total nitrogen prediction model for soil in any region is obtained based on the following steps: Collect target soil samples from the described area in the field; The total nitrogen content of the target soil was determined using chemical assay methods. The target soil was collected using a hyperspectral imager, which collected first reflectance data in a first wavelength range and second reflectance data in a second wavelength range; the first wavelength range was 400 nm to 1100 nm, and the second wavelength range was 1000 nm to 2500 nm. After correcting the reflection data within the overlapping wavelength range based on the first reflection data and the second reflection data, the corrected third reflection data within the first wavelength range is obtained; The third reflection data and the measured total nitrogen content are used as the model dataset, and the model dataset is divided into a training set and a prediction set; The third reflection data in the training set is denoised to obtain denoised reflection data; Feature wavelengths are selected from the noise-reduced reflection data, and the reflection data corresponding to the feature wavelengths are extracted to obtain the target feature reflection data; The target feature reflectance data and the measured total nitrogen content are used as a new training set. The new training set is used to train any regression model until the regression model converges, thus obtaining the prediction model to be verified. The prediction set is input into the prediction model to be verified to obtain the predicted value of total nitrogen content; If the error between the predicted total nitrogen content and the measured total nitrogen content does not meet expectations, the noise reduction method and / or feature wavelength screening method are changed, and the process of denoising the third reflection data in the training set is repeated until the error between the predicted total nitrogen content and the measured total nitrogen content meets expectations. At this point, the prediction model to be verified is determined as the total nitrogen prediction model.