Soil nutrient spectrum detection and regulation method and system

Abstract

An embodiment of the present invention provides a soil nutrient spectrum detection and regulation method and system, belonging to the technical field of soil nutrient spectrum detection and regulation. The method includes: collecting soil multispectral raw data and environmental data, generating a calibration matrix and nonlinear correction parameters, performing optical reflectance conversion, and synchronously fusing the time-domain reflectometry dielectric constant and the spectral moisture index to obtain a fused moisture content; selecting a nutrient diagnostic characteristic band pair to obtain a moisture-suppressed spectral index; using a pre-trained transfer learning model to predict soil nutrient content, obtaining a soil nutrient content prediction value, and performing fertilization regulation based on the fertilization decision risk coefficient. The present invention addresses extreme soil moisture measurement deviations through soil type adaptive moisture fusion technology; through multi-source risk quantification decision-making and PID adaptive regulation, it can dynamically optimize fertilizer application and ensure execution safety, forming a precision agriculture closed loop of detection-decision-execution.

Description

Technical Field

[0001] The present invention relates to the technical field of soil nutrient spectrum detection and regulation, and in particular to a soil nutrient spectrum detection and regulation method and system. Background Art

[0002] Near-infrared spectroscopy has been developed over decades in the field of soil nutrient analysis, establishing a core technical framework based on reflectance at characteristic wavelengths and chemometric models. Early research (Stenberg et al., 2010, Advances in Agronomy) established the sensitive response mechanisms of the 400-2500 nm wavelength band to key nutrients such as nitrogen, phosphorus, and potassium, promoting the commercialization of static laboratory-based testing equipment. With the increasing demand for precision agriculture, mobile spectral detection systems have gradually integrated GPS positioning (with an accuracy of ±5 m) and multi-source data fusion technologies, enabling semi-quantitative analysis at the field scale. The recent introduction of deep learning methods has significantly improved modeling accuracy, with the nitrogen prediction error of a 1D-CNN model on a standardized dataset now reduced to within 7.5%. However, existing technical systems still suffer from systemic flaws when dealing with dynamic farmland environments: water disturbance compensation models rely on empirical coefficients, resulting in insufficient generalization; sensor drift lacks an online calibration mechanism; and the decision-to-execution process suffers from minute-long delays. Summary of the Invention

[0003] The purpose of the present invention is to provide a soil nutrient spectrum detection and regulation method and system for realizing accurate detection of farmland soil nutrients and intelligent fertilization regulation.

[0004] To achieve the above-mentioned objectives, the present invention provides a soil nutrient spectral detection and regulation method, comprising: collecting soil multispectral raw data and environmental data, combining with the reflectance of a standard whiteboard, performing dynamic sensor calibration, and generating a calibration matrix and nonlinear correction parameters; performing optical reflectance conversion based on the calibration matrix and the nonlinear correction parameters, and synchronously fusing the time-domain reflectometry dielectric constant and the spectral moisture index to obtain a fused moisture content; selecting a nutrient diagnostic feature band pair based on the fused moisture content and soil type to obtain a moisture suppression spectral index; predicting soil nutrient content based on the moisture suppression spectral index using a pre-trained transfer learning model to obtain a predicted soil nutrient content value; calculating a fertilization decision risk coefficient based on the predicted soil nutrient content value, combined with weather forecasts and vegetation indices, and performing fertilization regulation based on the fertilization decision risk coefficient.

[0005] Optionally, the method collects soil multispectral raw data and environmental data, combines them with the reflectance of a standard whiteboard, performs dynamic calibration of the sensor, and generates a calibration matrix and nonlinear correction parameters, including: collecting whiteboard reflection signals, and calculating basic reflectance response values ​​based on the whiteboard reflection signals; constructing an initial calibration matrix based on the basic reflectance response values; calculating the linear error of the initial calibration matrix, and obtaining linear correction parameters based on the linear error; correcting the initial calibration matrix based on the linear correction parameters to obtain a calibration matrix; collecting multiple groups of basic reflectance response values, calculating the nonlinear response error based on the multiple groups of basic reflectance response values, and generating nonlinear correction parameters based on the nonlinear response error.

[0006] Optionally, the synchronous fusion of the time domain reflectometry dielectric constant and the spectral moisture index to obtain the fused moisture content includes: separately calculating the time domain reflectometry moisture content and the spectral moisture index, and setting a soil compensation coefficient; and calculating the fused moisture content based on the time domain reflectometry moisture content, the spectral moisture index and the soil compensation coefficient.

[0007] Optionally, the method selects a nutrient diagnostic characteristic band pair based on the fused moisture content and soil type to obtain a moisture suppression spectral index, including: using the calibration matrix to convert the soil multi-spectral raw data into basic reflectance data, and fine-tuning the position of each spectral wavelength according to the environmental data; selecting a nutrient diagnostic characteristic band pair based on the fused moisture content and a preset wavelength sensitivity parameter; and using a moisture attenuation correction model to perform moisture interference correction to obtain a reflectance spectrum after moisture attenuation correction.

[0008] Optionally, the method of selecting a nutrient diagnostic characteristic band pair according to the fused moisture content and soil type to obtain a moisture suppression spectral index also includes: performing a moisture condition judgment; based on the moisture condition judgment result, selecting a nutrient diagnostic characteristic band pair according to the fused moisture content and soil type; extracting reflectance data of the selected nutrient diagnostic characteristic band pair, and calculating a basic spectral ratio index; and correcting the basic spectral ratio index based on the soil type to obtain a moisture suppression spectral index.

[0009] Optionally, based on the moisture suppression spectral index, a pre-trained transfer learning model is used to predict the soil nutrient content to obtain a predicted value of the soil nutrient content, including: standardizing the moisture suppression spectral index and adjusting the input dimension of the moisture suppression spectral index; inputting the moisture suppression spectral index after adjusting the input dimension and the environmental constraint parameters generated based on the soil type into the pre-trained transfer learning model; the pre-trained transfer learning model outputs a preliminary predicted value, and physically constrains the preliminary predicted value to obtain a predicted value of the soil nutrient content.

[0010] Optionally, the fertilization decision risk coefficient is calculated based on the predicted value of the soil nutrient content, combined with the weather forecast and the vegetation index, including: calculating the basic fertilizer amount, and correcting the basic fertilizer amount according to the soil type; adjusting the corrected basic fertilizer amount according to the weather forecast data; compensating the adjusted basic fertilizer amount based on the vegetation index to obtain a fertilizer compensation value.

[0011] Optionally, the calculation of the fertilization decision risk coefficient based on the predicted value of the soil nutrient content in combination with the weather forecast and the vegetation index also includes: calculating the confidence of the predicted value of the soil nutrient content, and converting the confidence of the predicted value of the soil nutrient content into a confidence risk; calculating the soil conductivity risk according to the soil type; calculating the vegetation stress risk according to the vegetation index; and fusing the confidence risk, the soil conductivity risk and the vegetation stress risk to obtain the fertilization decision risk coefficient.

[0012] Optionally, based on a pre-stored PID parameter table, the fertilization decision risk coefficient is mapped to the PID parameter table to obtain PID parameters; a fuse mechanism for excessive fertilizer quantity fuse and equipment failure fuse is formulated; based on the PID parameters, the fertilizer compensation value, and the fuse mechanism, a fertilization control signal is generated; and fertilization control is performed based on the fertilization control signal.

[0013] On the other hand, the present invention provides a soil nutrient spectrum detection and regulation system for implementing a soil nutrient spectrum detection and regulation method. The system includes a control module, which includes a memory, a processor, and a computer program stored in the memory and runnable on the processor. The processor executes the computer program to implement the soil nutrient spectrum detection and regulation method.

[0014] The above technical solution significantly improves the accuracy of spectral data through dynamic dual-domain calibration, and combines soil type adaptive moisture fusion technology to solve extreme soil moisture measurement deviations; uses moisture suppression band optimization and physical correction models to weaken moisture interference and generate a high signal-to-noise ratio nutrient index; based on transfer learning and physical constraints, it achieves small sample high-precision nutrient prediction; finally, through multi-source risk quantification decision-making and PID adaptive control, it can optimize the amount of fertilizer and ensure execution safety, forming a precision agriculture closed loop of detection-decision-execution.

[0015] Other features and advantages of the present invention will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the following detailed description, they are used to explain the present invention but do not constitute a limitation of the present invention. In the accompanying drawings:

[0017] Figure 1 It is a flow chart of soil nutrient spectrum detection and regulation method.

[0018] Figure 2 This is a flow chart for generating moisture suppression spectral index. DETAILED DESCRIPTION

[0019] The following is combined with Figure 1 -Attached Figure 2 The specific implementation of the embodiment of the present invention is described in detail. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0020] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of national laws and regulations. In the embodiments of this application, certain software, components, models, and other existing solutions in the industry may be mentioned. These should be considered as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of this application, but it does not mean that the applicant has or will necessarily use such solutions.

[0021] In the process of realizing the present invention, the inventors of the present application discovered that the existing technology has the following defects: the water interference compensation relies on the empirical coefficient, resulting in distortion of the spectral analysis of different types of soils; the characteristic wavelength selection and solidification cannot respond to the jump of water content, and the nutrient detection stability is poor; the decision-making and execution links are separated, resulting in significant response delays.

[0022] Example 1

[0023] Reference Figure 1-Figure 2 , which is the first embodiment of the present invention, provides a soil nutrient spectrum detection and regulation method, comprising:

[0024] S100: Collects soil multispectral raw data and environmental data, combines them with the reflectance of a standard whiteboard, performs dynamic sensor calibration, and generates a calibration matrix and nonlinear correction parameters.

[0025] Furthermore, a whiteboard reflection signal is collected, and a basic reflectivity response value is calculated based on the whiteboard reflection signal; an initial calibration matrix is ​​constructed based on the basic reflectivity response value; a linear error of the initial calibration matrix is ​​calculated, and a linear correction parameter is obtained based on the linear error; the initial calibration matrix is ​​corrected based on the linear correction parameter to obtain a calibration matrix; multiple groups of basic reflectivity response values ​​are collected, and a nonlinear response error is calculated based on the multiple groups of basic reflectivity response values, and a nonlinear correction parameter is generated based on the nonlinear response error.

[0026] Specifically, during the initial spectral acquisition phase, a standard reflective whiteboard (reflectivity ≥ 99%) was used to establish a baseline reference system. The sensor was fixed vertically above the whiteboard, equipped with a halogen light source at a 45° incident angle to eliminate scattered interference. Interval scanning was performed in the 400-2500nm band to acquire the raw signal. The base reflectance response value was calculated, and the initial calibration matrix was then calculated based on this base reflectance response value.

[0027] Furthermore, during the linear calibration phase, a 510nm calibration filter is inserted to quantify linearity errors. A least-squares optimization algorithm is used to solve the Hadamard compensation coefficients. The linear error of the initial calibration matrix is ​​calculated based on this, and linear correction parameters are obtained based on the linear error. The initial calibration matrix is ​​corrected using the linear correction parameters to obtain a calibration matrix. Multiple sets of base reflectance response values ​​are collected, and nonlinear response errors are calculated based on these values. Nonlinear correction parameters are then generated based on the nonlinear response errors. Finally, the linear and nonlinear correction parameters are integrated using a tensor fusion engine to generate a full-dimensional calibration matrix.

[0028] Preferably, the calibration matrix is ​​optimized through Hilbert space orthogonal decomposition to achieve error spectrum optimization, ensuring that the spectrum conversion error output by the dynamic dual-domain calibration technology complies with the third category mobile device standard of the VDI / VDE 2620 specification.

[0029] Preferably, by combining a standard whiteboard benchmark, linear calibration (quantifying and compensating for linear errors) and nonlinear calibration (fitting nonlinear errors based on multiple sets of data), and utilizing tensor fusion and Hilbert space optimization technology, the measurement accuracy and stability of the spectral sensor are significantly improved, and the original error is greatly reduced to meet the industrial-grade mobile equipment standard (VDI / VDE2620). This effectively overcomes the interference of complex field environments (such as vibration, temperature and humidity changes) on the acquisition of original spectral data, laying a high-precision and reliable data foundation for subsequent analysis.

[0030] S200: performing optical reflectance conversion based on the calibration matrix and the nonlinear correction parameters, and synchronously fusing the time domain reflectometry dielectric constant and the spectral moisture index to obtain a fused moisture content.

[0031] Furthermore, the time domain reflectometry moisture content and the spectral moisture index are calculated respectively, and a soil compensation coefficient is set; based on the time domain reflectometry moisture content, the spectral moisture index and the soil compensation coefficient, the fused moisture content is calculated.

[0032] Specifically, the physical quantity directly output by the spectral sensor, which is proportional to the incident light intensity, is converted into a basic reflectance through a calibration matrix. A nonlinear function (such as a polynomial) is used to compensate for signal distortion to obtain a corrected reflectance spectrum. The dielectric constant measurement value is obtained through a time domain reflectometry (TDR) soil probe, and the physical moisture content is obtained through a universal conversion model. The spectral moisture index is calculated based on the corrected reflectance spectrum. The soil compensation coefficient is set according to the soil type. , soil compensation coefficient Refer to Table 1 for settings.

[0033] Table 1 Soil compensation coefficient reference table

[0034]

[0035] Furthermore, the calculation formula of fusion moisture content is as follows:

[0036]

[0037]

[0038] in, represents the fusion moisture content, represents the soil compensation coefficient, represents the TDR physical moisture content, p represents the spectral slope conversion coefficient, SWI represents the spectral moisture index, q represents the spectral intercept conversion coefficient, Represents the spectral inversion water content.

[0039] Preferably, it combines the high-precision time domain reflectometry (TDR) physical moisture content and the spectral inversion moisture content based on the spectral moisture index, and introduces a compensation coefficient that is dynamically adjusted according to the soil type (sand, loam, clay). Weighted fusion is performed to overcome the inherent defects and errors of a single method in specific soils (such as weak spectral response of sand and overly strong spectral response of clay), and a fused moisture content value with higher accuracy, stronger adaptability, and more true reflection of the actual moisture condition of the soil is obtained, while reducing dependence on a single high-cost TDR probe.

[0040] S300: Selecting a nutrient diagnosis characteristic band pair according to the fused moisture content and soil type to obtain a moisture suppression spectral index.

[0041] Furthermore, the calibration matrix is ​​used to convert the soil multispectral raw data into basic reflectance data, and the position of each spectral wavelength is fine-tuned according to the environmental data; based on the fused moisture content and the preset wavelength sensitivity parameters, the nutrient diagnostic characteristic band pair is selected; and the moisture attenuation correction model is used to perform moisture interference correction to obtain the reflectance spectrum after moisture attenuation correction.

[0042] Specifically, the soil multispectral raw signal is converted based on the calibration matrix. For example, the analog electrical signal output by the sensor is converted into a digital signal, and then linear / nonlinear transformation is performed through the calibration matrix to generate basic reflectance data. Subsequently, each spectral wavelength is fine-tuned in combination with environmental data such as temperature and air pressure, such as temperature drift compensation. The temperature drift compensation formula is as follows:

[0043]

[0044] in, represents the wavelength shift, Indicates the temperature drift coefficient, T indicates the current ambient temperature, Indicates the reference temperature, which is calibrated by the sensor at the factory. Indicates the pressure compensation coefficient, P indicates the current atmospheric pressure, Indicates standard atmospheric pressure.

[0045] Furthermore, the calculated wavelength offset The whole value is added to each wavelength point of the original spectrum, and then the reflectance value is re-interpolated to achieve synchronous fine-tuning of the spectrum wavelength.

[0046] Furthermore, according to the fusion moisture content and preset wavelength sensitivity parameters to select nutrient diagnostic characteristic band pairs, for example, when When the value is less than 0.15, the 1400-2500nm segment is selected. The target nutrient band is determined based on the soil type (the 560nm / 720nm nitrogen diagnostic band pair is preferred for sandy soil, and the 1650nm / 1720nm organic matter band pair is preferred for clay). The reflectance of the selected band is then corrected using the moisture attenuation correction model to obtain the reflectance spectrum after moisture attenuation correction. The expression of the moisture attenuation correction model is as follows:

[0047]

[0048] in, represents the reflectance after correction of moisture, represents the original reflectivity after wavelength fine-tuning, represents the water extinction coefficient, represents the volumetric moisture content of the integrated soil, Indicates wavelength.

[0049] Furthermore, a moisture condition judgment is performed; based on the moisture condition judgment result, a nutrient diagnostic characteristic band pair is selected according to the fused moisture content and soil type; the reflectance data of the selected nutrient diagnostic characteristic band pair is extracted, and a basic spectral ratio index is calculated; and the basic spectral ratio index is corrected based on the soil type to obtain a moisture suppression spectral index.

[0050] Specifically, when the fusion moisture content When the moisture content is less than 0.15, it is judged as drought state; when 0.15≤ fusion moisture content When the fusion moisture content is less than 0.35, it is judged as the best detection state; when ... When the value is greater than 0.35, it is judged as an over-humidity state.

[0051] Furthermore, the optimal nutrient diagnostic characteristic band pairs were determined based on the moisture condition judgment results and soil type, as shown in Table 2.

[0052] Table 2 Optimal nutrient diagnostic characteristic band pairs

[0053]

[0054] Furthermore, the moisture-corrected reflectance value corresponding to the selected band is extracted from the moisture-attenuation-corrected reflectance spectrum, and the basic spectrum ratio index is calculated. The calculation formula of the basic spectrum ratio index is as follows:

[0055]

[0056] in, represents the basic spectral ratio index, and These are all selected nutrient diagnostic characteristic band pairs. Taking the sandy soil in drought state as an example, is the reflectance after water correction at a wavelength of 560nm, It is the reflectance after correction of moisture at a wavelength of 720nm.

[0057] Furthermore, the basic spectral ratio index is finally adjusted according to the soil type, for example, sandy soil is +0.03, clay is +0.05 to obtain the final moisture suppression spectral index.

[0058] Preferably, based on the fusion of moisture content and precise soil type information, the most sensitive nutrient diagnostic characteristic band pairs are dynamically selected, and the physical model (moisture attenuation correction model) is applied to the reflectivity of the selected bands to perform quantitative correction of moisture interference, which significantly weakens the absorption masking effect of soil moisture on the characteristic spectral signals of target nutrients (such as nitrogen and organic matter); and the basic spectral index is fine-tuned according to the soil type, and finally a moisture-suppressed spectral index is obtained that can more purely reflect soil nutrient information and is effectively suppressed by the influence of moisture, which greatly improves the signal-to-noise ratio and accuracy of subsequent nutrient predictions.

[0059] S400: Based on the moisture suppression spectral index, use a pre-trained transfer learning model to predict soil nutrient content to obtain a predicted value of soil nutrient content.

[0060] Furthermore, the moisture suppression spectral index is standardized and the input dimension of the moisture suppression spectral index is adjusted; the moisture suppression spectral index after the input dimension adjustment and the environmental constraint parameters generated based on the soil type are input into a pre-trained transfer learning model; the pre-trained transfer learning model outputs a preliminary prediction value, and the preliminary prediction value is physically constrained to obtain a predicted value of soil nutrient content.

[0061] Specifically, the water-attenuation-corrected full-band reflectance spectra and the water suppression spectral index were normalized and reconstructed into a time-step × feature-dimension format. Based on existing databases such as the Chinese Soil Species Directory and the UNSODA (Unsaturated Soil Database), statistical or machine-learning mappings were established between soil type and key physical parameters. Environmental constraint parameters were generated based on soil type. The reconstructed water-attenuation-corrected full-band reflectance spectra, environmental constraint parameters, and water suppression spectral index were then fed into a pretrained transfer learning model. The pretrained transfer learning model kept the weights of 80% of the convolutional layers fixed (only the top fully connected layer was allowed to be fine-tuned) and output preliminary nutrient predictions. Physical constraints were applied to the preliminary predictions, such as forcing the predictions to fall within the physically feasible range, setting spatial variation thresholds based on soil type, and performing a moving average filter with a 9×9 grid window to eliminate abnormal fluctuations. This resulted in predicted soil nutrient content.

[0062] Preferably, the pre-trained transfer learning model is used to effectively solve the problem of scarce field nutrient annotation data in farmland. By reusing the prior knowledge of large soil databases (such as UNSODA), combined with local fine-tuning and multi-dimensional input (corrected spectra, environmental constraint parameters, moisture inhibition index), high-precision nutrient content prediction under small sample conditions is achieved; at the same time, physical constraints are imposed on the preliminary prediction results (such as feasible interval restrictions and spatial filtering), which effectively eliminates abnormal fluctuations and results that do not conform to physical laws, ensuring the stability and reliability of the predicted values.

[0063] S500: Calculating a fertilization decision risk coefficient based on the predicted soil nutrient content, combined with weather forecasts and vegetation indices, and performing fertilization control based on the fertilization decision risk coefficient.

[0064] Furthermore, the basic fertilizer amount is calculated and corrected according to the soil type; the corrected basic fertilizer amount is adjusted according to the weather forecast data; and the adjusted basic fertilizer amount is compensated based on the vegetation index to obtain a fertilizer compensation value.

[0065] Specifically, the calculation formula for basic fertilizer amount is as follows:

[0066]

[0067] in, represents the basic fertilizer amount, C represents the nitrogen requirement of the target crop, and N represents the predicted value of soil nutrient content. represents the soil bulk density, D represents the root layer depth, Indicates the fertilizer utilization rate in the season.

[0068] Furthermore, the basic fertilizer amount is modified according to the soil type, for example, ×1.2, clay is ×0.8.

[0069] Furthermore, the formula for adjusting the revised basic fertilizer amount according to the weather forecast data is as follows:

[0070]

[0071] in, represents the amount of fertilizer corrected for meteorological risk, Indicates the basic fertilizer amount after correction by soil type, represents the meteorological risk adjustment factor.

[0072] Furthermore, the formula for compensating the adjusted basic fertilizer amount based on the vegetation index is as follows:

[0073]

[0074] in, represents vegetation index compensation, represents the amount of fertilizer corrected for meteorological risk, Represents the vegetation stress compensation coefficient.

[0075] It should be noted that the vegetation index compensation is the final fertilizer compensation value.

[0076] Furthermore, the confidence of the predicted value of soil nutrient content is calculated, and the confidence of the predicted value of soil nutrient content is converted into confidence risk; the soil conductivity risk is calculated according to the soil type; the vegetation stress risk is calculated according to the vegetation index; the confidence risk, the soil conductivity risk and the vegetation stress risk are integrated to obtain the fertilization decision risk coefficient.

[0077] Furthermore, the residual distribution of the pre-trained transfer learning model is calculated, the confidence of the predicted value of soil nutrient content is calculated, and the confidence is converted into confidence risk. When the confidence is ≥0.9, the confidence risk is 0; when 0.8<confidence<0.9, the confidence risk is 0.5×(0.9-confidence); when the confidence is ≤0.8, the confidence risk is 0.5+2.5×(0.8-confidence).

[0078] Furthermore, a soil conductivity sensor is used to obtain the conductivity of the surface layer (0-20 cm). The soil conductivity risk is then calculated based on the soil type. For example, the conductivity risk calculation formula for sandy soil is as follows:

[0079]

[0080] in, It indicates the conductivity risk of sandy soil, and EC indicates the current measured value of soil conductivity.

[0081] Furthermore, calculating vegetation stress risk includes calculating NDVI stress ratio and red edge stress index.

[0082] The calculation formula of NDVI stress ratio is as follows:

[0083]

[0084] in, represents the normalized stress level, Indicates the best vegetation status index of the region in the current growing season. Indicates the most recent comprehensive vegetation index.

[0085] The calculation formula of red edge stress index is as follows:

[0086]

[0087] in, It represents the red edge stress index, 700nm is the starting point of the chlorophyll absorption edge, and 720nm is the cell structure scattering peak.

[0088] when When <1.1, the vegetation stress risk is 0.8× +0.2 (1.1- );

[0089] when When ≥1.1, the vegetation stress risk is 0.5× .

[0090] Furthermore, the confidence risk, the soil conductivity risk, and the vegetation stress risk are integrated to calculate the fertilization decision risk coefficient. The calculation formula of the fertilization decision risk coefficient is as follows:

[0091]

[0092] Where S is the risk coefficient of fertilization decision (0-1 scalar), 、 、 All are weights.

[0093] Furthermore, based on a pre-stored PID (Proportional-Integral-Derivative) benchmark parameter table, the fertilization decision risk coefficient is mapped to the PID parameter table to obtain PID parameters, and a fuse mechanism for excessive fertilizer quantity fuse and equipment failure fuse is formulated. Based on the PID parameters, the fertilizer compensation value, and the fuse mechanism, a fertilization control signal is generated; and fertilization control is performed based on the fertilization control signal.

[0094] Furthermore, based on a pre-stored PID parameter table, the fertilization decision risk coefficient is mapped to the PID parameter table to obtain PID parameters, and a fuse mechanism for excessive fertilizer quantity fuse and equipment failure fuse is formulated. Based on the PID parameters, the fertilizer compensation value, and the fuse mechanism, a fertilization control signal is generated; and fertilization control is performed based on the fertilization control signal.

[0095] Preferably, the fertilizer quantity out-of-tolerance fuse is that when the flow meter feedback error continues to be greater than 10% and exceeds 30 seconds, the actuator output power is forced to be reduced to 50%; the equipment failure fuse is that when the motor temperature is greater than 80°C or the pipeline pressure is less than 0.2MPa, the current channel valve is immediately closed and the entire machine is triggered to shut down.

[0096] Preferably, the fertilization decision risk coefficient is calculated through the integration of multi-dimensional risk assessment (model confidence risk, soil conductivity salinization risk, vegetation stress risk), providing a scientific basis for fertilization amount decision-making; based on this coefficient, the basic fertilizer amount is dynamically adjusted (taking into account soil type correction, meteorological risk, and vegetation stress compensation), and finally the optimized fertilizer compensation value is obtained, which significantly improves the fertilizer utilization efficiency and the matching degree with crop demand; the innovative fuse mechanism (fertilizer amount out-of-tolerance fuse, equipment failure fuse) combined with the adaptive PID control based on dynamic risk mapping ensures the safety and reliability of the fertilization execution process, and effectively prevents fertilizer waste and equipment damage.

[0097] The present invention also provides a soil nutrient spectrum detection and regulation system for implementing a soil nutrient spectrum detection and regulation method. The system includes a control module, which includes a memory, a processor, and a computer program stored in the memory and runnable on the processor. The processor executes the computer program to implement the soil nutrient spectrum detection and regulation method.

[0098] An embodiment of the present invention provides a storage medium having a program stored thereon, which implements the soil nutrient spectrum detection and regulation method when executed by a processor.

[0099] An embodiment of the present invention provides a processor, which is used to run a program, wherein the soil nutrient spectrum detection and regulation method is executed when the program is run.

[0100] An embodiment of the present invention provides a device comprising a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements a method for soil nutrient spectral detection and control. The device herein may be a server, a PC, a PAD, a mobile phone, or the like.

[0101] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a soil nutrient spectrum detection and regulation method.

[0102] Those skilled in the art will appreciate that the embodiments of the present application may provide methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0103] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0104] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0105] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0106] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0107] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0108] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can be implemented using any method or technology for information storage. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change RAM (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media, such as modulated data signals and carrier waves.

[0109] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0110] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included within the scope of the claims of the present application.

Claims

1. A soil nutrient spectrum detection and regulation method, characterized in that: include: Collect soil multispectral raw data and environmental data, combine them with the reflectance of a standard whiteboard, perform dynamic sensor calibration, and generate a calibration matrix and nonlinear correction parameters; Based on the calibration matrix and the nonlinear correction parameters, optical reflectance conversion is performed, and the time domain reflectance dielectric constant and the spectral moisture index are synchronously fused to obtain a fused moisture content, including respectively calculating the time domain reflectance moisture content and the spectral moisture index, and setting a soil compensation coefficient; calculating the fused moisture content based on the time domain reflectance moisture content, the spectral moisture index and the soil compensation coefficient; further comprising: performing moisture condition judgment; based on the moisture condition judgment result, selecting a nutrient diagnostic feature band pair according to the fused moisture content and soil type; extracting reflectance data of the selected nutrient diagnostic feature band pair, and calculating a basic spectral ratio index; and correcting the basic spectral ratio index based on the soil type to obtain a moisture suppression spectral index; Selecting a nutrient diagnostic characteristic band pair based on the fused moisture content and soil type to obtain a moisture suppression spectral index, including using the calibration matrix to convert the soil multispectral raw data into basic reflectance data, and fine-tuning the position of each spectral wavelength based on the environmental data; selecting a nutrient diagnostic characteristic band pair based on the fused moisture content and a preset wavelength sensitivity parameter; and using a moisture attenuation correction model to perform moisture interference correction to obtain a reflectance spectrum after moisture attenuation correction; Based on the moisture suppression spectral index, a pre-trained transfer learning model is used to predict soil nutrient content to obtain a predicted value of the soil nutrient content, including standardizing the moisture suppression spectral index and adjusting the input dimension of the moisture suppression spectral index; inputting the moisture suppression spectral index after the input dimension adjustment and environmental constraint parameters generated based on the soil type into the pre-trained transfer learning model; the pre-trained transfer learning model outputs a preliminary predicted value, and the preliminary predicted value is subjected to physical constraint processing to obtain a predicted value of the soil nutrient content; According to the predicted value of soil nutrient content, combined with weather forecast and vegetation index, a fertilization decision risk coefficient is calculated, and fertilization regulation is performed based on the fertilization decision risk coefficient.

2. The soil nutrient spectrum detection and control method according to claim 1, characterized in that: The method collects soil multispectral raw data and environmental data, combines them with the reflectance of a standard whiteboard, performs dynamic sensor calibration, and generates a calibration matrix and nonlinear correction parameters, including: collecting a whiteboard reflection signal, and calculating a basic reflectivity response value based on the whiteboard reflection signal; constructing an initial calibration matrix according to the basic reflectance response value; Calculating a linear error of the initial calibration matrix and obtaining a linear correction parameter according to the linear error; Correcting the initial calibration matrix based on the linear correction parameters to obtain a calibration matrix; A plurality of groups of basic reflectivity response values ​​are collected, a nonlinear response error is calculated based on the plurality of groups of basic reflectivity response values, and a nonlinear correction parameter is generated according to the nonlinear response error.

3. The soil nutrient spectrum detection and control method according to claim 1, characterized in that: The calculation of the fertilization decision risk coefficient based on the predicted soil nutrient content, combined with weather forecasts and vegetation indices, includes: Calculate the basic fertilizer amount and modify it according to the soil type; Adjusting the revised basic fertilizer amount according to the weather forecast data; The adjusted basic fertilizer amount is compensated based on the vegetation index to obtain a fertilizer compensation value.

4. The soil nutrient spectrum detection and control method according to claim 1, characterized in that: The step of calculating the risk coefficient of fertilization decision-making based on the predicted value of soil nutrient content in combination with weather forecast and vegetation index also includes: Calculating the confidence level of the predicted value of soil nutrient content, and converting the confidence level of the predicted value of soil nutrient content into a confidence risk; Calculate soil conductivity risk based on soil type; calculating vegetation stress risk according to the vegetation index; The confidence risk, the soil conductivity risk and the vegetation stress risk are integrated to obtain a fertilization decision risk coefficient.

5. The soil nutrient spectrum detection and control method according to claim 3, characterized in that: The fertilization control based on the fertilization decision risk coefficient includes: Based on a pre-stored PID parameter table, the fertilization decision risk coefficient is mapped to the PID parameter table to obtain PID parameters; Formulate a fusing mechanism for excessive fertilizer dosage and equipment failure; generating a fertilization control signal based on the PID parameter, the fertilizer compensation value, and the fuse mechanism; Fertilization control is performed based on the fertilization control signal.

6. A soil nutrient spectrum detection and control system, characterized in that: The system includes a control module, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program to implement the soil nutrient spectrum detection and regulation method according to any one of claims 1 to 5.

Images