Black soil degradation real-time monitoring and early warning system and method based on multi-source remote sensing and Internet of Things

Abstract

The invention discloses a black soil degradation real-time monitoring and early warning system and method based on multi-source remote sensing and the Internet of Things, and the system comprises a data sensing layer which is used for receiving the Internet of Things data, satellite image data and unmanned aerial vehicle image data of a research area, constructing a multi-stage land utilization transfer matrix and a geological profile in combination with prior historical data, and carrying out the real-time monitoring and early warning of the black soil degradation. Forming an ecological geological map; the data analysis layer is used for generating an agricultural machinery operation instruction by adopting an enhanced regression tree model fused with a soil science mechanism and a multi-objective optimization algorithm pair according to the ecological geological map so as to realize benefit-ecological balance; and the execution layer is used for receiving and analyzing the agricultural machine operation instruction, downloading and executing the operation parameters, and performing one-way hash on the data for storage after the operation is completed. Through accurate monitoring, optimized operation decision and efficient implementation, sustainable utilization and restoration of the land are supported, black soil resources are protected, the agricultural ecological environment quality is improved, and sustainable development of agriculture is promoted.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ecological restoration, and in particular to a black soil degradation real-time monitoring and early warning system and method based on multi-source remote sensing and Internet of Things. BACKGROUND

[0002] Black soil is an important agricultural soil type in China, with good fertility and water retention capacity, but in recent years, due to unreasonable land use, excessive tillage and environmental changes, black soil degradation has become increasingly serious. Black soil degradation not only affects the sustainability of agricultural production and reduces soil biodiversity, but also exacerbates soil erosion and environmental degradation. Therefore, how to timely and accurately monitor and warn black soil degradation has become an urgent issue in the fields of agriculture and environmental science. By using remote sensing technology, Internet of Things technology and intelligent analysis model, the real-time monitoring and accurate assessment of black soil degradation process can be realized, and scientific basis for farmland management can be provided, which has important significance for effectively preventing black soil degradation and improving agricultural production efficiency.

[0003] At present, most of the researches on soil degradation rely on traditional ground investigation and sample collection methods, which are often limited by space and time, and it is difficult to realize large-scale and real-time monitoring. In addition, although the existing remote sensing technology can provide a large amount of image data in black soil degradation monitoring, it often lacks the ability of fusion and accurate analysis with Internet of Things data, and it is difficult to provide dynamic and real-time degradation monitoring information. Although some remote sensing image processing systems have certain monitoring capabilities, their analysis results often lack the support of soil science mechanism and fail to deeply explore the internal rules of soil degradation. In actual agricultural operation, the generation of agricultural operation instructions is often based on experience or simple rules, lacking scientific decision support system. The existing technology generally faces problems such as data multi-source heterogeneity, poor timeliness and insufficient decision support.

[0004] Therefore, the present application proposes a black soil degradation real-time monitoring and early warning system and method based on multi-source remote sensing and Internet of Things, which has the advantages of efficient, accurate and real-time monitoring, and provides scientific basis and decision support for the prevention and control of black soil degradation. SUMMARY

[0005] Therefore, the present application provides a black soil degradation real-time monitoring and early warning system and method based on multi-source remote sensing and Internet of Things, which solves the problem of poor real-time monitoring capability in the prior art, and can provide efficient and accurate degradation monitoring and early warning scheme by integrating remote sensing, Internet of Things data and intelligent analysis.

[0006] To achieve the above technical purpose, the present application adopts the following technical scheme: In a first aspect, the present application provides a black soil degradation real-time monitoring and early warning system based on multi-source remote sensing and Internet of Things, comprising a data sensing layer, a data analysis layer and an execution layer connected in sequence; The data sensing layer is configured to receive Internet of Things data, satellite image data and unmanned aerial vehicle image data of a study area, combine prior historical data to construct a multi-period land use transfer matrix and a geological profile, and form an ecological geological map; The data analysis layer is configured to analyze the land use transfer matrix and the geological profile according to the ecological geological map, generate a farm machine operation instruction by using an enhanced regression tree model fusing soil mechanism and a multi-objective optimization algorithm, so as to realize benefit-ecological balance, and is further configured to issue the generated farm machine operation instruction and record operation data uploaded by the execution layer based on a blockchain technology; The execution layer is configured to receive and analyze the farm machine operation instruction, locate a corresponding degraded plot on an electronic map, download and execute operation parameters, and real-time return operation trajectory and secondary acquisition data of a sensor after operation is completed, and store the operation data in a unidirectional hash chain for evidence.

[0007] Further, the data sensing layer comprises a data acquisition unit and a geological analysis unit. The data acquisition unit is configured to receive satellite image data, DEM data and unmanned aerial vehicle image data through an FTP channel and an MQTT message channel, and receive Internet of Things data sent by an Internet of Things terminal arranged in a preset manner; The geological analysis unit is configured to perform object-oriented classification on satellite and unmanned aerial vehicle images by using a prior land use type system, generate multi-period land use vector data, superimpose data by using a preset tool to obtain a land use transfer matrix, and perform spatial registration on satellite image data and unmanned aerial vehicle image data and prior historical data, assign horizontal geological labels and vertical profile labels to a study area according to the land use transfer matrix, and form an ecological geological map.

[0008] Further, the data analysis layer comprises a feature extraction module, a degradation prediction module, an optimization decision module and an instruction generation module connected in sequence. The feature extraction module is configured to extract land degradation related features from the ecological geological map; The degradation prediction module is configured to input the land degradation related features into a pre-trained enhanced regression tree model fusing soil mechanism, output soil organic carbon change amount and land degradation probability in a preset time period, and identify potential degradation risks; The optimization decision module is configured to input the soil organic carbon change amount and the land degradation probability, adopt a NSGA-II multi-objective optimization algorithm, and perform Pareto optimization between economic benefit, ecological benefit and repair suitability to generate an optimal farm machine operation parameter set; An instruction generation module is configured to encapsulate the optimal agricultural machine operation parameter set into a JSON instruction, and to issue the instruction through an MQTT channel, and to write an instruction hash value into a blockchain to achieve tamper-proof evidence.

[0009] Further, the enhanced regression tree model of the fused pedology mechanism adopts a regression tree model taking soil organic carbon as a continuous dependent variable, taking a CART regression tree as a base learner, and embedding a pedology physical constraint term in a loss function.

[0010] Further, the pedology physical constraint term is a difference between an average theoretical value obtained by integrating a theoretical calculation value of soil organic carbon vertical distribution and an average measured value of soil organic carbon in the previous period, to obtain a predicted soil organic carbon change amount in a prediction period ; The calculation formula of the theoretical calculation value of soil organic carbon vertical distribution is: wherein, is a soil layer depth, is a vegetation input intensity factor, is an organic carbon decay coefficient with depth; is a mineral combination coefficient, is a soil clay volume fraction at depth d, is a bedrock weathering inert organic carbon background value.

[0011] Further, the loss function embedding the pedology physical constraint term is expressed by a formula as: In the formula, is a total loss function of the model, is an enhanced regression tree original mean square error loss, is a penalty weight coefficient, is a predicted soil organic carbon change amount in a prediction period, is a prediction time step, is a soil organic carbon loss threshold.

[0012] Further, the calculation method of the land degradation probability is: substituting the predicted soil organic carbon change amount in the prediction period into a degradation probability mapping formula for solving, the degradation probability mapping formula being: wherein, is a degradation probability, is a predicted soil organic carbon change amount in a prediction period, is a translation parameter, is a scale parameter, The base of the natural logarithm.

[0013] Further, the optimization decision module comprises a target setting unit and a decision variable unit; The target setting unit is configured to calculate an economic benefit based on a seasonal crop market average price, an ecological benefit based on a carbon sink transaction guide price, and a restoration suitability parameter based on an average hardness of the land; The decision unit is configured to take the economic benefit, the ecological benefit, and the restoration suitability parameter as a fitness function of NSGA-II, perform Pareto optimization on three-dimensional decision variables including deep loosening depth, fertilizer amount, and irrigation amount, obtain a Pareto frontier, and screen a block-level optimal agricultural machine operation parameter set from the Pareto frontier.

[0014] Further, the screening of the block-level optimal agricultural machine operation parameter set from the Pareto frontier comprises: The individuals are sorted in descending order of crowding distance, and the individual with the largest crowding distance is taken as the optimal agricultural machine operation parameter set; if the crowding distances are the same, the individuals are sorted in descending order of economic benefit value, and the individual at the front of the sorting is taken as the optimal agricultural machine operation parameter set.

[0015] On the other hand, the present application provides a black soil degradation real-time monitoring and early warning method based on multi-source remote sensing and Internet of Things, which is realized by using the technical solution of the black soil degradation real-time monitoring and early warning system based on multi-source remote sensing and Internet of Things, and comprises the following steps: Receiving Internet of Things data, satellite image data, and unmanned aerial vehicle image data of a research area, and combining prior historical data to construct a multi-period land use transfer matrix and a block-level geological profile, and forming a block-level ecological geological map; Based on the block-level ecological geological map, an enhanced regression tree model taking soil organic carbon as a continuous dependent variable and embedding a soil science physical constraint term is called to output a soil organic carbon change amount and a land degradation probability in a preset time period; A NSGA-II multi-objective optimization algorithm is used to perform Pareto optimization among economic benefit, ecological benefit, and restoration suitability, to generate a block-level optimal agricultural machine operation parameter set, and to issue the agricultural machine operation instruction; The agricultural machine operation instruction is parsed and executed, the corresponding degraded block is located on an electronic map, the operation parameters are downloaded and executed, and after the operation is completed, the operation trajectory and secondary acquisition data of the sensor are returned in real time, and the operation data is written into a blockchain through one-way hashing; The returned data is used to update the block-level ecological geological map, to realize real-time closed-loop monitoring and early warning of black soil degradation.

[0016] Compared with the prior art, the present application has the following advantages: (1) The data analysis layer can accurately analyze land use transfer and geological changes by fusing the enhanced regression tree model of soil mechanism and the multi-objective optimization algorithm, and generate scientific and reasonable agricultural machinery operation instructions, so as to optimize operation parameters, improve agricultural machinery operation efficiency, reduce operation cost, and realize the double balance of benefits and ecology.

[0017] (2) The operation data is recorded and notarized by the blockchain technology, so that the data is tamper-proof, has high transparency and traceability, and the credibility of the operation data is improved, which provides a reliable basis for subsequent farmland management and decision-making.

[0018] (3) The system can monitor the black soil degradation process in real time by combining multi-source remote sensing data and Internet of Things sensing technology, and generate early warning information in time to provide a rapid response mechanism for soil management. (4) By accurately analyzing the land use change and black soil degradation, the balance between benefits and ecology is realized, which promotes the sustainable development of agriculture, reduces the negative impact of soil degradation on the ecological environment, and helps to protect the black soil resources in the long term.

[0019] The present application constructs a high-precision ecological geological map by multi-source data fusion, accurately predicts soil degradation by using the enhanced regression tree model of soil physical constraints, generates optimal agricultural machinery operation parameters by combining multi-objective optimization algorithm, realizes the balance between economic benefits and ecological benefits, and at the same time adopts blockchain to ensure data security, forms real-time closed-loop monitoring and early warning, greatly improves the scientific nature, efficiency and sustainability of black soil degradation management. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 The structure diagram of the black soil degradation real-time monitoring and early warning system based on multi-source remote sensing and Internet of Things provided by the present application is provided. Figure 2 The structure diagram of the data analysis layer provided by the present application is provided. Figure 3 The flowchart of the black soil degradation real-time monitoring and early warning method based on multi-source remote sensing and Internet of Things provided by the present application is provided. DETAILED DESCRIPTION

[0021] The preferred embodiments of the present application will be specifically described below in conjunction with the drawings, wherein the drawings constitute a part of the present application, and are used to illustrate the principles of the present application together with the embodiments of the present application, but are not used to limit the scope of the present application.

[0022] Please see Figure 1 , Figure 1 The structure diagram of the black soil degradation real-time monitoring and early warning system 100 based on multi-source remote sensing and Internet of Things provided by the present embodiment is shown, which includes a data sensing layer 101, a data analysis layer 102 and an execution layer 103 connected in sequence. The data perception layer 10 is configured to receive Internet of Things data, satellite image data and unmanned aerial vehicle image data of a study area, combine prior historical data to construct a multi-period land use transfer matrix and a geological profile, and form an ecological geological atlas; The data analysis layer 102 is configured to analyze the land use transfer matrix and the geological profile according to the ecological geological atlas, generate a farm machine operation instruction by using an enhanced regression tree model and a multi-objective optimization algorithm, and realize benefit-ecological balance; and is further configured to issue the generated farm machine operation instruction, and record operation data uploaded by the execution layer based on a blockchain technology; The execution layer 103 is configured to receive and analyze the farm machine operation instruction, locate a corresponding degraded land block on an electronic map, download and execute operation parameters, and return operation track and secondary collection data of a sensor in real time after the operation is completed, and store and record the operation data by one-way hashing.

[0023] The real-time monitoring and early warning system for black soil degradation based on multi-source remote sensing and Internet of Things provided by the embodiment realizes real-time monitoring and accurate early warning of black soil degradation by fusing multi-source remote sensing and Internet of Things data, optimizes generation and execution of a farm machine operation instruction, and improves the intelligent level of agricultural management. Meanwhile, the blockchain technology is used to ensure data security and transparency, which provides strong support for black soil protection and sustainable agricultural development, and has significant ecological and economic benefits.

[0024] As a preferred embodiment, the data perception layer comprises a data acquisition unit and a geological analysis unit. The data acquisition unit is configured to receive satellite image data, DEM data and unmanned aerial vehicle image data through an FTP channel and an MQTT message channel, and receive Internet of Things data sent by an Internet of Things terminal arranged in a preset manner. The geological analysis unit is configured to perform object-oriented classification on satellite and unmanned aerial vehicle images by using a prior land use type system, generate multi-period land use vector data, superimpose the data by using a preset tool to obtain a land use transfer matrix, and perform spatial registration on the satellite image data and the unmanned aerial vehicle image data and prior historical data, assign horizontal geological labels and vertical profile labels to a study area according to the land use transfer matrix, and form an ecological geological atlas.

[0025] Specifically, satellite data receives domestic GF-2 (1 m), GF-6 (16 m) and Landsat-8 / 9 (30 m) L2 level products through provincial geological cloud FTP; based on the interpretation experience of five periods from 1980 to 2020, unified radiation calibration, atmospheric correction, geometric precision correction (error ≤1 pixel) are carried out. The UAV image adopts 0.05 m resolution aerial photo, flight height 120 m, 80% lateral overlap; after aerial triangulation, DEM and orthophoto are generated; Internet of Things terminal: according to the relevant measurement specifications of "soil hardness, three parameters", seven field JSON (hardness, water content, Fe 2 ⁺, Mn 2 ⁺, temperature, conductivity, SOC) is returned every 30 minutes through the Internet of Things terminal, and is unified to the server through MQTT.

[0026] In ArcGIS Pro, the "Tabulate Area" tool is used to superimpose each period to generate a 6x6 transfer matrix, and the matrix is written into the PostgreSQL partition table in real time, with the primary key being "unique land code + starting year".

[0027] As a preferred embodiment, as shown in Figure 2 The data analysis layer includes a feature extraction module 201, a degradation prediction module 202, an optimization decision module 203 and an instruction generation module 204 connected in turn; The feature extraction module 201 is configured to extract land degradation related features from the ecological geological map; The degradation prediction module 202 is configured to input the land degradation related features into a pre-trained enhanced regression tree model based on the fusion pedology mechanism, and output the soil organic carbon change amount and the land degradation probability in a preset time period, and identify potential degradation risks; The optimization decision module 203 is configured to input the soil organic carbon change amount and the land degradation probability, and use the NSGA-II multi-objective optimization algorithm to perform Pareto optimization between economic benefit, ecological benefit and repair suitability, and generate an optimal agricultural machinery operation parameter set; The instruction generation module 204 is configured to encapsulate the optimal agricultural machinery operation parameter set into a JSON instruction, and issue the instruction through an MQTT channel, and write the instruction hash value into a blockchain to realize tamper-proof evidence.

[0028] As a specific embodiment, the land degradation related features include the following three categories, a total of 19 items: (1) Time series features (from land use transfer matrix), including: wetland→dryland conversion area ratio, wetland→paddy field conversion area ratio, dryland→paddy field conversion area ratio, dryland→wetland conversion area ratio, unused land→cultivated land conversion area ratio; (2) Horizontal geological features (from the plot horizontal geological label); including: soil-forming parent material type (boggy soil, meadow soil, marsh soil…), lithology code (river-lake clay, granodiorite residual…), soil type vulnerability classification value; (3) Vertical profile features (from the 0-60 cm measured curve), including: 0-10 cm soil hardness (kg cm⁻ 2 ), 10-20 cm soil hardness, 20-40 cm soil hardness, 40-60 cm soil hardness, 0-10 cm SOC (g kg⁻ 1 ), 10-20 cm SOC, 20-40 cm SOC, 40-60 cm SOC, groundwater depth (m), slope (°) and elevation.

[0029] The above fields constitute a set of 19 indicators required for the enhanced regression tree model.

[0030] As a preferred embodiment, the enhanced regression tree model of the integrated soil science mechanism adopts a regression tree model with soil organic carbon as a continuous dependent variable, CART regression tree as a base learner, and embedding a soil science physical constraint term in the loss function.

[0031] Specifically, the soil science physical constraint term is the difference ΔSOC between the average SOC theoretical value integrated from the theoretical calculation value of the vertical distribution of soil organic carbon in the 0-60 cm soil layer and the average SOC measured value in the previous period. By considering the soil science mechanism (vertical attenuation, parent material difference) as a constraint term in the loss function of the model, not just fitting statistical errors, the model can ensure that it follows the basic principles of soil science, such as the distribution characteristics of soil organic carbon in different soil layers, thereby enhancing the practical interpretability of the model.

[0032] As a preferred embodiment, the soil science physical constraint term is the difference between the average theoretical value integrated from the theoretical calculation value of the vertical distribution of soil organic carbon and the average measured value of soil organic carbon in the previous period, resulting in the predicted amount of change of soil organic carbon in the prediction period ; The calculation formula of the theoretical calculation value of the vertical distribution of soil organic carbon is: wherein, is the soil layer depth, is the vegetation input intensity factor, is the organic carbon decay coefficient with depth; is the mineral combination coefficient, is the soil clay volume fraction at depth d, is the bedrock weathering inert organic carbon background value.

[0033] As a specific embodiment, in practice, the β, γ, δ parameters are usually calibrated according to soil types: β = 0.42, γ = 1.15, ξ = 0.8 for white clay soil; β = 0.38, γ = 1.05, ξ = 0.6 for meadow soil; β = 0.35, γ = 0.95, ξ = 0.4 for marsh soil.

[0034] As a preferred embodiment, the loss function embedded with soil physics constraints is expressed by the formula: In the formula, is the total loss function of the model, is the original mean square error loss of the enhanced regression tree, is the penalty weight coefficient, is the predicted soil organic carbon change amount in the prediction period, is the prediction time step, is the soil organic carbon loss threshold.

[0035] As a preferred embodiment, the calculation method of the land degradation probability is to substitute the predicted soil organic carbon change amount in the prediction period into the degradation probability mapping formula for solving, the degradation probability mapping formula being: In the formula, is the degradation probability, is the predicted soil organic carbon change amount in the prediction period, is the translation parameter, is the scale parameter, is the base of the natural logarithm.

[0036] As a preferred embodiment, the optimization decision module includes a target setting unit and a decision unit; The target setting unit is configured to calculate economic benefits based on the average market price of crops in the season, ecological benefits based on the guidance price of carbon sink transactions, and repair suitability parameters based on the average hardness of the land. The decision unit is configured to use the economic benefits, ecological benefits, and repair suitability parameters as the fitness function of NSGA-II, perform Pareto optimization on three-dimensional decision variables including deep loosening depth, fertilizer amount, and irrigation amount, obtain a Pareto frontier, and screen a land-level optimal agricultural machinery operation parameter set from the Pareto frontier.

[0037] As a preferred embodiment, the filtering of the plot-level optimal agricultural machine operation parameter set in the Pareto frontier comprises: In descending order of crowding distance, the individual with the largest crowding distance is taken as the optimal agricultural machine operation parameter set; if the crowding distance is the same, the individual with the highest economic benefit value is taken as the optimal agricultural machine operation parameter set.

[0038] In some embodiments, after filtering the plot-level optimal agricultural machine operation parameter set, the SHAP value can be sorted to automatically generate a Chinese natural language report, for example: “Since the parent material of the plot is white paste soil, the hardness is 45 kg / cm 2 , and it is predicted that the SOC will decrease by 0.3% in 30 days, and it is recommended to deep till 35 cm + organic fertilizer 300 kg / acre.” As a specific embodiment, the execution layer includes a packaging unit, a publishing unit and a blockchain writing unit connected in turn; wherein the packaging unit packages the instruction according to the ISO 11783 Task Controller field format; the publishing unit is used to publish the operation instruction conforming to the MQTT protocol, adopts the Topic structure, and specifically is: / field / {plot ID} / cmd; the blockchain writing module writes the SHA-256 hash of JSON into the block by calling FISCO-BCOS SDK.

[0039] For example, the JSON packager generates {“plot ID”: “F123”, “deep tillage”: 35, “fertilization”: 300, “irrigation”: 20}; the MQTT publisher QoS=1, ensures that the agricultural machine automatically retransmits after offline caching; the blockchain writer returns txHash, which is bound to the plot ID for evidence storage.

[0040] As shown in Figure 3 , the embodiment of the present application also provides a black soil degradation real-time monitoring and early warning method based on multi-source remote sensing and Internet of Things, comprising: Step S301: receiving Internet of Things data, satellite image data and unmanned aerial vehicle image data of a study area, and combining prior historical data to construct a multi-period land use transfer matrix and a plot-level geological profile, and forming a plot-level ecological geological map; Step S302: based on the plot-level ecological geological map, calling an enhanced regression tree model with soil organic carbon as a continuous dependent variable and embedding a soil science physical constraint term, and outputting a soil organic carbon change amount and a land degradation probability in a preset time period; Step S303: performing Pareto optimization between economic benefit, ecological benefit and repair suitability by using an NSGA-II multi-objective optimization algorithm, generating a plot-level optimal agricultural machine operation parameter set, and issuing the agricultural machine operation instruction; Step S304: analyze and execute the agricultural operation instruction, locate the corresponding degraded plot on the electronic map, download and execute the operation parameters, and return the operation trajectory and secondary acquisition data of the sensor in real time after the operation is completed, and write the operation data into the blockchain through one-way hash; Step S305: use the returned data to update the plot-level ecological geological map to realize real-time closed-loop monitoring and early warning of black soil degradation.

[0041] The real-time monitoring and early warning method for black soil degradation based on multi-source remote sensing and the Internet of Things provided by the application can accurately predict the degradation of soil by multi-source data fusion and enhanced regression tree model, helping researchers to identify the degradation trend in advance; the multi-objective optimization algorithm is used to generate optimal agricultural operation parameters to balance economic benefits and ecological restoration effects and improve the efficiency of agricultural restoration work; the blockchain technology ensures the non-tamperability and transparency of operation data, improves the data credibility and management transparency; the operation data is used to update the ecological geological map to form a closed-loop monitoring and early warning mechanism, which can continuously and effectively manage and restore the degradation.

[0042] The above is only the preferred specific embodiment of the application, but the protection scope of the application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the application, which should be covered within the protection scope of the application.

Claims

1. A real-time monitoring and early warning system for black soil degradation based on multi-source remote sensing and the Internet of Things, characterized in that, It includes a data perception layer, a data analysis layer, and an execution layer connected in sequence; The data perception layer is used to receive IoT data, satellite imagery data and UAV imagery data of the study area, and combine them with prior historical data to construct a multi-phase land use transfer matrix and geological profile, forming an ecological geological map. The data analysis layer is used to analyze the land use transfer matrix and geological profile based on the ecological geological map, using an enhanced regression tree model that integrates soil science mechanisms and a multi-objective optimization algorithm, to generate agricultural machinery operation instructions in order to achieve a balance between benefits and ecology. It is also used to issue generated agricultural machinery operation instructions and record the operation data uploaded by the execution layer based on blockchain technology; The execution layer is used to receive and parse agricultural machinery operation instructions, locate the corresponding degraded plots on the electronic map, download and execute the operation parameters; After the task is completed, the task trajectory and secondary sensor data are transmitted back in real time, and the task data is stored on the blockchain through one-way hashing.

2. The real-time monitoring and early warning system for black soil degradation based on multi-source remote sensing and the Internet of Things as described in claim 1, characterized in that, The data sensing layer includes a data acquisition unit and a geological analysis unit; The data acquisition unit is used to receive satellite image data, DEM data, UAV image data, and IoT data sent by IoT terminals arranged in a preset manner through FTP channels and MQTT message channels. The geological analysis unit is used to perform object-oriented classification on satellite and UAV imagery using a priori land use type system, and generate multi-period land use vector data. The land use transfer matrix is ​​obtained by overlaying the data period by period using preset tools; satellite imagery data and UAV imagery data are spatially registered with prior historical data; and horizontal geological labels and vertical profile labels are assigned to the study area according to the land use transfer matrix to form an ecological geological map.

3. The real-time monitoring and early warning system for black soil degradation based on multi-source remote sensing and the Internet of Things as described in claim 1, characterized in that, The data analysis layer includes a feature extraction module, a degradation prediction module, an optimization decision module, and an instruction generation module connected in sequence. The feature extraction module is used to extract land degradation-related features from the ecological geological map; The degradation prediction module is used to input land degradation-related features into a pre-trained augmented regression tree model that integrates soil science mechanisms, and output the change in soil organic carbon and the probability of land degradation over a preset time period to identify potential degradation risks. The optimization decision module is used to take the change in soil organic carbon and the probability of land degradation as inputs, and use the NSGA-II multi-objective optimization algorithm to perform Pareto optimization among economic benefits, ecological benefits and restoration suitability to generate the optimal set of agricultural machinery operation parameters. The instruction generation module is used to encapsulate the optimal agricultural machinery operation parameter set into JSON instructions, send them through the MQTT channel, and write the instruction hash value into the blockchain to achieve tamper-proof evidence storage.

4. The real-time monitoring and early warning system for black soil degradation based on multi-source remote sensing and the Internet of Things as described in claim 3, characterized in that, The enhanced regression tree model that integrates soil science mechanisms adopts a regression tree model with soil organic carbon as a continuous dependent variable, CART regression tree as the base learner, and embeds soil science physical constraints in the loss function.

5. The real-time monitoring and early warning system for black soil degradation based on multi-source remote sensing and the Internet of Things as described in claim 4, characterized in that, The soil science and physical constraint term is the difference between the average theoretical value calculated based on the vertical distribution theory of soil organic carbon and the measured average soil organic carbon value of the previous period, which is used to obtain the change in soil organic carbon within the predicted time period. ; The formula for calculating the theoretical value of vertical distribution of soil organic carbon is as follows: in, The depth of the soil layer. Input intensity factors for vegetation. The coefficient of organic carbon decay with depth; The mineral bonding coefficient, The volume fraction of soil clay particles at depth d. This represents the background value of inert organic carbon from bedrock weathering.

6. The real-time monitoring and early warning system for black soil degradation based on multi-source remote sensing and the Internet of Things as described in claim 4, characterized in that, The loss function for the embedded soil physical constraints is expressed by the following formula: In the formula, The total loss function of the model, To enhance the original mean squared error loss of the regression tree, For the penalty weighting coefficient, To predict changes in soil organic carbon over a given period of time. To predict the time step, This represents the threshold for soil organic carbon loss.

7. The real-time monitoring and early warning system for black soil degradation based on multi-source remote sensing and the Internet of Things as described in claim 6, characterized in that, The method for calculating the land degradation probability is as follows: The change in soil organic carbon within the predicted time period... Substituting the values ​​into the degradation probability mapping formula for solution, the degradation probability mapping formula is as follows: In the formula, The probability of degradation. To predict changes in soil organic carbon over a given period of time. For translation parameters, For scale parameters, is the base of the natural logarithm.

8. The real-time monitoring and early warning system for black soil degradation based on multi-source remote sensing and the Internet of Things as described in claim 3, characterized in that, The optimization decision-making module includes a target setting unit and a decision-making unit; The target setting unit is used to calculate the economic benefits based on the average market price of crops in the current season, the ecological benefits based on the carbon sink trading guidance price, and the restoration suitability parameters based on the average hardness of the land. The decision-making unit is used to use economic benefits, ecological benefits, and restoration suitability parameters as fitness functions of NSGA-II, to perform Pareto optimization on three-dimensional decision variables including deep tillage depth, fertilizer application rate, and irrigation amount, to obtain the Pareto front, and to select the optimal set of agricultural machinery operation parameters at the plot level from the Pareto front.

9. The real-time monitoring and early warning system for black soil degradation based on multi-source remote sensing and the Internet of Things as described in claim 8, characterized in that, The process of selecting the optimal set of agricultural machinery operation parameters at the plot level from the Pareto front includes: Sort the parameters by crowding distance in descending order, and select the individual with the largest crowding distance as the optimal set of agricultural machinery operation parameters; if the crowding distances are the same, sort them by economic benefit value in descending order, and select the individual with the highest value as the optimal set of agricultural machinery operation parameters.

10. A method for real-time monitoring and early warning of black soil degradation based on multi-source remote sensing and the Internet of Things, implemented using the real-time monitoring and early warning system for black soil degradation based on multi-source remote sensing and the Internet of Things as described in any one of claims 1-9, characterized in that, include: The system receives IoT data, satellite imagery data, and UAV imagery data from the study area, and combines them with prior historical data to construct a multi-phase land use transfer matrix and a plot-level geological profile, forming a plot-level ecological geological map. Based on the plot-level ecological geological map, an enhanced regression tree model with soil organic carbon as a continuous dependent variable and embedded soil science and physics constraints is invoked to output the change in soil organic carbon and the probability of land degradation over a preset time period. The NSGA-II multi-objective optimization algorithm is used to perform Pareto optimization among economic benefits, ecological benefits and restoration suitability, generate the optimal set of agricultural machinery operation parameters at the plot level, and issue the agricultural machinery operation instructions. The system parses and executes the agricultural machinery operation instructions, locates the corresponding degraded plots on the electronic map, downloads and executes the operation parameters, and transmits the operation trajectory and secondary sensor data back in real time after the operation is completed. The operation data is then written into the blockchain via one-way hashing. The returned data will be used to update the plot-level ecological geological map, enabling real-time closed-loop monitoring and early warning of black soil degradation.

Images