Soil hardening treatment method and device

Abstract

The invention relates to the technical field of soil hardening treatment, in particular to a soil hardening treatment method and device. Acquiring a soil remote sensing image, soil spectral data and soil radar reflection data corresponding to the target area; identifying the soil remote sensing image, the soil spectral data and the soil radar reflection data, and determining the soil hardening degree corresponding to the target area; and determining a soil hardening treatment measure corresponding to the target area according to the soil hardening degree corresponding to the target area. Therefore, the soil hardening degree corresponding to the target area does not need to be manually determined, the efficiency of determining the soil hardening degree corresponding to the target area is improved, and the labor cost is saved. According to the soil hardening degree corresponding to the target area, the soil hardening treatment measure corresponding to the target area is determined, it is guaranteed that the output soil hardening treatment measure corresponds to the soil hardening degree of the target area, and then the accuracy of the output soil hardening treatment measure is guaranteed. Therefore, the agricultural production benefit can be improved.

Description

Technical Field

[0001] The invention relates to the technical field of soil compaction treatment, and in particular to a soil compaction treatment method and device. Background Art

[0002] Soil compaction is a common problem in agricultural production. It destroys the stability of soil structure, significantly reduces air permeability and water permeability, and seriously affects the development of crop roots and nutrient absorption efficiency. Especially in areas with high clay content, the compaction problem is more prominent. Long-term accumulation can easily lead to soil degradation, directly threatening the sustainable use of farmland.

[0003] At present, the assessment of soil compaction mainly relies on manual field sampling and laboratory analysis. Although it can obtain local accurate data, it is inefficient and difficult to achieve large-scale dynamic monitoring.

[0004] Therefore, how to achieve large-scale dynamic monitoring to determine the soil compaction situation and deal with it has become an urgent problem to be solved. Summary of the invention

[0005] In view of this, the present invention provides a soil compaction treatment method and device to solve the problem of determining the degree of soil compaction and exploring an effective soil compaction treatment method.

[0006] In a first aspect, the present invention provides a method for treating soil compaction, the method comprising: Obtain soil remote sensing images, soil spectral data, and soil radar reflection data corresponding to the target area; Identify soil remote sensing images, soil spectral data, and soil radar reflection data to determine the degree of soil compaction in the target area; According to the degree of soil compaction in the target area, the soil compaction treatment measures corresponding to the target area are determined.

[0007] The soil compaction treatment method provided in the embodiment of the present application obtains the soil remote sensing image, soil spectral data and soil radar reflection data corresponding to the target area. Then, the soil remote sensing image, soil spectral data and soil radar reflection data are identified to determine the soil compaction degree corresponding to the target area, thereby ensuring the accuracy of determining the soil compaction degree corresponding to the target area. Thus, there is no need to manually determine the soil compaction degree corresponding to the target area, thereby improving the efficiency of determining the soil compaction degree corresponding to the target area and saving manpower costs. According to the soil compaction degree corresponding to the target area, the soil compaction treatment measures corresponding to the target area are determined, thereby ensuring that the output soil compaction treatment measures correspond to the soil compaction degree of the target area, thereby ensuring the accuracy of the output soil compaction treatment measures. Thus, the agricultural production efficiency can be improved.

[0008] In an optional implementation, identifying soil remote sensing images, soil spectral data, and soil radar reflection data to determine the degree of soil compaction corresponding to the target area includes: Identify the soil remote sensing image to determine the first soil texture feature corresponding to the target area; Identify the soil spectrum data and determine the first absorption peak characteristics corresponding to the target area; Identify soil radar reflection data and determine the first wave crest and trough characteristics corresponding to the target area; The first soil texture feature, the first absorption peak feature, and the first wave crest and trough feature are integrated to generate a target quantum bit feature; The target quantum bit characteristics are input into the preset soil compaction determination model, and the degree of soil compaction corresponding to the target area is output.

[0009] The soil compaction processing method provided in the embodiment of the present application identifies the soil remote sensing image and determines the first soil texture feature corresponding to the target area. By identifying and determining the first soil texture feature, the overall texture information of the soil surface, such as the roughness of the soil surface, the directionality and periodicity of the texture, etc., can be quickly obtained. This information helps to preliminarily judge the topography of the soil, the impact of vegetation coverage on the soil, etc. The soil spectrum data is identified to determine the first absorption peak feature corresponding to the target area. By identifying and determining the first absorption peak feature, the content and state of various minerals, organic matter, water and other components in the soil can be accurately analyzed. Different soil components have preset absorption peaks in the spectrum. The soil radar reflection data is identified to determine the first wave peak and trough feature corresponding to the target area. By identifying and determining the first wave peak and trough feature, the pore structure, stratification and whether there is a compaction layer in the deep soil layer can be clearly understood. This information is crucial for evaluating the aeration, water permeability and root growth environment of the soil. The first soil texture feature, the first absorption peak feature and the first wave peak and trough feature are fused to generate the target quantum bit feature. It is possible to achieve the complementarity and integration of different types of information. The target quantum bit features are input into the preset soil compaction determination model, and the soil compaction degree corresponding to the target area is output. The preset soil compaction determination model is based on the fused target quantum bit features for analysis and judgment, and can more accurately determine the degree of soil compaction in the target area. Since the target quantum bit features integrate key information from multi-source data, the model can comprehensively consider multiple factors such as the surface morphology, composition, and deep structure of the soil, avoiding the limitations of single feature judgment and improving the accuracy and reliability of the judgment.

[0010] In an optional embodiment, the first soil texture feature, the first absorption peak feature, and the first wave crest and wave valley feature are fused to generate a target quantum bit feature, including: Identifying the first soil texture feature, converting the geometric shape and spatial distribution information in the first soil texture feature into a state combination of quantum bits, and generating texture feature quantum bits; Identifying the first absorption peak feature, and determining the intensity and position information of the absorption peaks at different wavelengths in the first absorption peak feature; Quantize the intensity and position information of each absorption peak to generate absorption peak characteristic quantum bits; Identify the peak and trough characteristics of the first wave, use quantum bits to encode the amplitude, frequency and phase of the wave, and generate peak and trough characteristic quantum bits; The texture feature qubits, absorption peak feature qubits and peak and valley feature qubits are fused to generate the target qubit features.

[0011] The soil compaction treatment method provided in the embodiment of the present application identifies the first soil texture feature, converts the geometric shape and spatial distribution information in the first soil texture feature into a state combination of quantum bits, and generates texture feature quantum bits. Thus, the complex information of soil texture can be accurately represented in the form of quantum states. The superposition state characteristics of quantum bits make it possible to simultaneously represent the possibility of multiple geometric shapes and spatial distributions. In addition, the generated texture feature quantum bits can be directly applied to quantum computing environments. Quantum computing has the ability to parallelize computing and can simultaneously process the states of multiple quantum bits, greatly improving the processing speed and analysis efficiency of soil texture features. The first absorption peak feature is identified, the intensity and position information of the absorption peaks at different wavelengths in the first absorption peak feature are determined, and the intensity and position information of each absorption peak are quantized to generate absorption peak feature quantum bits. Thus, soil component information can be deeply mined, because different soil components have preset absorption peak characteristics in the spectrum. By quantizing, the intensity and position information of the absorption peak are converted into the state of quantum bits, which can more accurately represent the content and state of various minerals, organic matter and other components in the soil. The first wave peak and valley features are identified, and the amplitude, frequency and phase of the wave are encoded using quantum bits to generate peak and valley feature quantum bits. This can accurately represent the deep soil structure information contained in the peak and valley features in the soil radar reflection data. The encoding method of quantum bits can more accurately capture the subtle changes in the wave. The texture feature quantum bits, absorption peak feature quantum bits and peak and valley feature quantum bits are fused to generate the target quantum bit features. The deep fusion of multi-source soil information is achieved. Different types of feature quantum bits represent information on the surface texture, composition and deep structure of the soil. Through fusion, these scattered information can be organically combined to form a more comprehensive and in-depth description of soil characteristics. This deep fusion can tap the intrinsic connection between different levels and aspects of the soil.

[0012] In an optional implementation, the texture feature qubit, the absorption peak feature qubit, and the peak and valley feature qubit are fused to generate a target qubit feature, including: Preprocessing texture feature qubits, absorption peak feature qubits, and wave crest and trough feature qubits; The pre-processed texture feature qubits, absorption peak feature qubits and wave crest and trough feature qubits are input into the soil multimodal entanglement gate; The pulse sequence in the soil multimodal entanglement gate is used to control the interaction between the texture feature qubit, the absorption peak feature qubit, and the peak and trough feature qubit; the frequency, amplitude, and duration of the pulse sequence are obtained by multiple iterations of training the soil multimodal entanglement gate; Using the Sierpinski triangle topology in the soil multimodal entanglement gate, the pre-processed texture feature qubits, absorption peak feature qubits, and wave crest and trough feature qubits are transmitted and entangled between nodes to generate a multi-body entangled state. Post-process the multi-body entangled state to extract the target quantum bit characteristics.

[0013] The soil compaction processing method provided in the embodiment of the present application pre-processes the texture feature quantum bits, the absorption peak feature quantum bits and the peak and trough feature quantum bits, so that noise and interference can be effectively removed, the quality of the data can be improved, and it has a consistent format and range. Then, the pre-processed texture feature quantum bits, the absorption peak feature quantum bits and the peak and trough feature quantum bits are input into the soil multimodal entanglement gate. So that data from different soil characteristics can be processed and associated in the same system. This breaks the limitation of independent analysis of different data types in traditional methods, and can comprehensively explore the inherent connection between different aspects of soil characteristics, which provides the possibility for in-depth understanding of the comprehensive properties of soil. The pulse sequence in the soil multimodal entanglement gate is used to control the interaction between the texture feature quantum bits, the absorption peak feature quantum bits and the peak and trough feature quantum bits, so that the texture feature quantum bits, the absorption peak feature quantum bits and the peak and trough feature quantum bits can be changed in state and entangled in the expected manner, thereby improving the accuracy and repeatability of the entanglement operation. For example, by adjusting the parameters of the pulse sequence, the preset entanglement relationship between the texture feature qubit and the absorption peak feature qubit can be achieved, thereby better reflecting the association between soil texture and composition. Next, using the Sierpinski triangle topology in the soil multimodal entanglement gate, the pre-processed texture feature qubit, absorption peak feature qubit, and peak and trough feature qubit are transmitted and entangled between nodes to generate a multi-body entangled state. The Sierpinski triangle topology provides an efficient mechanism for the mutual transmission and entanglement of qubits. Its unique geometric structure enables qubits to be quickly transmitted between nodes and entangled through the connection between nodes. This topological structure can increase the interaction opportunities between qubits, improve the efficiency and success rate of entanglement, and ensure the accuracy of the generated multi-body entangled state. The multi-body entangled state is post-processed to extract the target qubit features from it. The extracted target qubit features integrate information from multiple aspects such as soil texture, composition, and deep structure, and can reflect the comprehensive properties of the soil more comprehensively and deeply. For example, by measuring and analyzing multi-body entangled states, key features related to soil compaction, soil fertility, etc. can be extracted, providing a scientific basis for soil management and agricultural production.

[0014] In an optional implementation, post-processing the multi-body entangled state to extract target quantum bit features therefrom includes: Based on the multi-body entangled state, the density matrix is ​​reconstructed; The quantum version of principal component analysis is introduced to perform principal component analysis on the density matrix. The main components are determined by finding the eigenvalues ​​and eigenvectors of the density matrix. Based on the main components, the target quantum bit characteristics are determined.

[0015] The soil compaction treatment method provided in the embodiment of the present application is based on the multi-body entangled state and reconstructs the density matrix. The multi-body entangled state contains rich quantum information. By reconstructing the density matrix, the state of the quantum system can be fully described. The density matrix can not only reflect the various possible states of quantum bits and their probability distribution, but also reflect important information such as the entanglement relationship between quantum bits. For the quantum bit system corresponding to the soil multimodal data, the reconstructed density matrix can integrate the complex associations between characteristic quantum bits such as texture, absorption peaks, peaks and valleys, and provide a complete framework for subsequent in-depth analysis, which helps to understand the comprehensive characteristics of the soil more accurately. The quantum version of principal component analysis is introduced to perform principal component analysis on the density matrix, and the main components are determined by finding the eigenvalues ​​and eigenvectors of the density matrix. Based on the main components, the target quantum bit characteristics are determined. When the quantum version of principal component analysis (Q-PCA) processes the density matrix, dimensionality reduction and information compression can be achieved. The multi-body entangled state corresponding to the soil multimodal data often has high dimensions and contains a large amount of information, some of which may be redundant or have little effect on the main features. By finding the eigenvalues ​​and eigenvectors of the density matrix through Q-PCA, the high-dimensional quantum system can be mapped to the low-dimensional principal component space, retaining the most important information and removing redundant information. This greatly reduces the complexity of data processing and improves computational efficiency without losing key soil characteristic information. Determining the main components of the density matrix can highlight the main features and patterns in the soil data. Among the multimodal characteristics of soil, different features have different degrees of influence on soil properties. Q-PCA can help us identify those characteristic quantum bits and their combinations that play a key role in the comprehensive properties of soil. Finally, determining the target quantum bit characteristics based on the main components can accurately find the key features that are of great significance to soil research. These key features are extracted from the multi-body entangled state, integrating information from different aspects of the soil, and can more accurately reflect the true nature and state of the soil. For example, when judging the degree of soil compaction, the target quantum bit characteristics may include the compactness information of the soil texture, the changes in the soil mineral composition reflected by the spectral absorption peaks, and the deep soil structure characteristics reflected by the radar peaks and troughs. Through these key features, the degree and cause of soil compaction can be more reliably evaluated.

[0016] In an optional embodiment, the target quantum bit feature is input into a preset soil compaction determination model, and the soil compaction degree corresponding to the target area is output, including: Inputting the target quantum bit characteristics into a preset soil compaction determination model; The multi-scale feature extraction module extracts the target quantum bit features based on quantum filters of different scales and outputs quantum bits of different scales; The quantum convolution module performs sliding convolution on quantum bits of different scales at different positions through quantum gate operations to generate convolution bit features; The quantum residual connection layer adds the convolution bit features and the target quantum bit features to output the final quantum bit; Based on the final quantum bit, the soil compaction process corresponding to the target area is output.

[0017] The soil compaction processing method provided in the embodiment of the present application inputs the target quantum bit feature into the preset soil compaction determination model. The multi-scale feature extraction module extracts the target quantum bit feature based on quantum filters of different scales and outputs quantum bits of different scales. Thus, the target quantum bit feature can be analyzed from multiple levels and angles, and the characteristic information of the soil can be fully captured. In addition, quantum bits of different scales are output so that the preset soil compaction determination model can adapt to the characteristic changes under different soil environments and conditions. The quantum convolution module performs sliding convolution on quantum bits of different scales at different positions through quantum gate operations, which can deeply explore the local associations and patterns between soil features and generate convolution bit features. The generated convolution bit feature is a further abstraction and refinement of the target quantum bit feature, which enhances the representation ability of the target quantum bit feature. The convolution operation can fuse and integrate the feature information of different scales and positions to form a more representative feature representation. These convolution bit features contain richer soil feature information, which can better reflect the complex nature of the soil, provide more powerful support for subsequent analysis and judgment, and help improve the accuracy of soil compaction degree judgment. The quantum residual connection layer adds the convolution bit features and the target quantum bit features and outputs the final quantum bit, which can retain the original target quantum bit feature information, so that the final quantum bit can be transmitted more directly, effectively preventing the problem of gradient disappearance and ensuring the stability and training effect of the model. At the same time, retaining the original information also helps the model to better utilize the comprehensive soil features obtained in the early stage of processing and improve the performance of the model. Based on the final quantum bit, the soil compaction process corresponding to the target area is output, which can more accurately reflect the actual compaction of the soil in the target area.

[0018] In an optional implementation, according to the soil compaction degree corresponding to the target area, the soil compaction treatment measures corresponding to the target area are determined, including: Identify soil remote sensing images, soil spectral data, and soil radar reflection data to determine the cause of soil compaction in the target area; According to the causes and degree of soil compaction, the corresponding soil compaction treatment measures for the target area are output.

[0019] The soil compaction treatment method provided in the embodiment of the present application identifies soil remote sensing images, soil spectral data and soil radar reflection data to determine the soil compaction cause corresponding to the target area. Soil remote sensing images can provide macroscopic information of the soil surface, such as land use type, topography, etc., to help determine whether soil compaction is caused by unreasonable land development or topographic factors; soil spectral data can reveal the chemical composition and mineral composition of the soil, and determine whether compaction is caused by the accumulation of certain chemical substances or changes in mineral composition; soil radar reflection data can detect the deep structure of the soil and determine whether soil pore changes, soil layer disturbance, etc. are the causes of compaction. This multi-source data comprehensive analysis method is like giving the soil a comprehensive physical examination, avoiding the one-sidedness of the diagnosis of a single data source, and greatly improving the accuracy of the diagnosis of the cause of soil compaction. Then, according to the cause of soil compaction and the degree of soil compaction, the soil compaction treatment measures corresponding to the target area are output, ensuring the accuracy of the output soil compaction treatment measures corresponding to the target area. Reasonable soil compaction treatment measures can improve soil structure, improve soil aeration and water permeability, and create a good soil environment for crop growth.

[0020] In an optional implementation, identifying soil remote sensing images, soil spectral data, and soil radar reflection data to determine the cause of soil compaction corresponding to the target area includes: Performing data preprocessing on the soil remote sensing image and extracting a second soil texture feature corresponding to the soil remote sensing image; Preprocessing the soil spectrum data, and extracting the second absorption peak feature corresponding to the absorption peak of the preset band; Preprocess the soil radar reflection data and extract the pore structure characteristics related to the soil pore structure; The second soil texture characteristics, the second absorption peak characteristics and the pore structure characteristics are input into the preset cause identification model, and the soil compaction causes corresponding to the target area are output.

[0021] The soil compaction processing method provided in the embodiment of the present application performs data preprocessing on the soil remote sensing image and extracts the second soil texture feature corresponding to the soil remote sensing image, thereby improving the data quality and accurately reflecting the soil surface condition. The soil spectral data is preprocessed, and the second absorption peak feature corresponding to the preset band absorption peak is extracted therefrom; thereby improving the data quality and accurately reflecting the soil surface condition. The soil radar reflection data is preprocessed, and the pore structure features related to the soil pore structure are extracted therefrom, so that the physical structure of the deep soil layer can be deeply understood, including pore size, connectivity, pore wall roughness, etc. This information is crucial for evaluating the air permeability, water permeability and root growth environment of the soil. The second soil texture feature, the second absorption peak feature and the pore structure feature are input into the preset cause identification model, and the soil compaction cause corresponding to the target area is output. It can comprehensively consider the information of multiple aspects such as the surface physical structure, chemical composition and deep pore structure of the soil. This multi-source data fusion analysis method avoids the limitations of single data type analysis and can diagnose the cause of soil compaction more comprehensively and accurately.

[0022] In an optional embodiment, the second soil texture feature, the second absorption peak feature and the pore structure feature are input into a preset cause identification model, and the soil compaction cause corresponding to the target area is output, including: Normalizing the second soil texture feature, the second absorption peak feature and the pore structure feature; The normalized second soil texture feature, the second absorption peak feature and the pore structure feature are input into a preset cause identification model; Multiple neurons in the hidden layer of the preset cause recognition model perform nonlinear transformation on the input data through activation functions and output transformation features; The output layer in the preset cause identification model outputs the cause of soil compaction based on the transformation features.

[0023] The soil compaction processing method provided in the embodiment of the present application performs normalization processing on the second soil texture feature, the second absorption peak feature and the pore structure feature, thereby eliminating the influence of the data dimension and improving the generalization ability of the preset cause recognition model. The normalized second soil texture feature, the second absorption peak feature and the pore structure feature are input into the preset cause recognition model. Multiple neurons in the hidden layer of the preset cause recognition model perform nonlinear transformation on the input data through the activation function and output the transformation feature. The nonlinear transformation effect of the activation function enables the preset cause recognition model to learn more complex relationships and patterns in the input data. The cause of soil compaction is often caused by the interaction of multiple factors, and there are complex nonlinear relationships between these factors. By performing nonlinear transformation on the input data through the activation function, the model can capture these nonlinear relationships, thereby more accurately describing the connection between the cause of soil compaction and soil characteristics. For example, the activation function can transform the complex interaction between soil texture features, absorption peak features and pore structure features into a form that the model can understand and process, thereby improving the model's ability to recognize the cause of soil compaction. The output layer in the preset cause recognition model outputs the cause of soil compaction based on the transformation feature. The output layer analyzes and judges based on the transformation features output by the hidden layer, and can accurately output the cause of soil compaction. After the previous normalization processing and nonlinear transformation of the hidden layer, the input data has been converted into highly representative transformation features. Based on these transformation features, combined with the training results and algorithms of the model, the output layer can accurately identify the main and secondary causes of soil compaction.

[0024] In a second aspect, the present invention provides a soil compaction treatment device, the device comprising: An acquisition module is used to acquire soil remote sensing images, soil spectral data, and soil radar reflection data corresponding to the target area; A determination module is used to identify soil remote sensing images, soil spectral data, and soil radar reflection data to determine the degree of soil compaction corresponding to the target area; The output module is used to determine the soil compaction treatment measures corresponding to the target area according to the soil compaction degree corresponding to the target area.

[0025] The soil compaction treatment device provided in the embodiment of the present application obtains the soil remote sensing image, soil spectral data and soil radar reflection data corresponding to the target area. Then, the soil remote sensing image, soil spectral data and soil radar reflection data are identified to determine the soil compaction degree corresponding to the target area, thereby ensuring the accuracy of determining the soil compaction degree corresponding to the target area. Thus, there is no need to manually determine the soil compaction degree corresponding to the target area, thereby improving the efficiency of determining the soil compaction degree corresponding to the target area and saving manpower costs. According to the soil compaction degree corresponding to the target area, the soil compaction treatment measures corresponding to the target area are determined, thereby ensuring that the output soil compaction treatment measures correspond to the soil compaction degree of the target area, thereby ensuring the accuracy of the output soil compaction treatment measures. Thus, the agricultural production efficiency can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0027] Figure 1 is a schematic flow chart of a soil compaction treatment method according to an embodiment of the present invention; Figure 2 is a schematic flow chart of another soil compaction treatment method according to an embodiment of the present invention; Figure 3 is a schematic diagram of generating an entangled state according to an embodiment of the present invention; Figure 4 is a schematic flow chart of another soil compaction treatment method according to an embodiment of the present invention; Figure 5 4 is a structural block diagram of a soil compaction treatment device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0029] It should be noted that the method for treating soil compaction provided in the embodiment of the present application, its execution subject may be a soil compaction treatment device, and the soil compaction treatment device may be implemented as part or all of a computer device through software, hardware, or a combination of software and hardware, wherein the computer device may be a server or a terminal, wherein the server in the embodiment of the present application may be a single server or a server cluster composed of multiple servers, and the terminal in the embodiment of the present application may be a smart phone, a personal computer, a tablet computer, a wearable device, an intelligent robot, and other intelligent hardware devices. In the following method embodiments, the execution subject is an electronic device as an example for explanation.

[0030] According to an embodiment of the present invention, an embodiment of a soil compaction treatment method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0031] In this embodiment, a soil compaction treatment method is provided, which can be used for the above-mentioned electronic equipment. Figure 1 : is a flow chart of a soil compaction treatment method according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps: Step S101, obtaining soil remote sensing images, soil spectrum data and soil radar reflection data corresponding to the target area.

[0032] Specifically, an aircraft or a UAV can carry a remote sensing device to collect data on the target area and obtain a remote sensing image of the soil corresponding to the target area. The electronic device can receive a remote sensing image of the soil sent by an aircraft or a UAV. The electronic device can also receive a spectrometer to collect data on the target area and obtain soil spectral data corresponding to the target area. The electronic device can also receive a radar device to collect data on the target area and obtain soil radar reflection data corresponding to the target area.

[0033] Step S102, identifying the soil remote sensing image, soil spectral data and soil radar reflection data to determine the soil compaction degree corresponding to the target area.

[0034] Specifically, the electronic device can input soil remote sensing images, soil spectral data, and soil radar reflection data into a preset soil compaction determination model to determine the degree of soil compaction corresponding to the target area.

[0035] This step will be described in detail below.

[0036] Step S103, determining soil compaction treatment measures corresponding to the target area according to the soil compaction degree corresponding to the target area.

[0037] Specifically, if the soil compaction degree corresponding to the target area is primary compaction, the soil compaction treatment measures corresponding to the target area are determined to be at least one of optimized farming methods, reasonable fertilization and soil conditioning, and biological improvement measures.

[0038] Among them, optimizing the tillage method can be to adopt less tillage or no tillage technology to reduce excessive disturbance to the soil structure. When less tillage is used, the tillage depth is controlled at 10-15 cm to avoid destroying the aggregate structure of the soil surface. At the same time, agricultural machinery with special plows, such as deep loosening plows, can be used to break the shallow compaction layer without turning the soil, thereby enhancing soil aeration and water permeability. For example, before sowing in spring, deep loosening operations are carried out on slightly compacted farmland, and deep loosening strips are set at intervals of 30-40 cm to improve the physical structure of the soil. Among them, reasonable fertilization and soil conditioning are to adjust the fertilization strategy, increase the application of organic fertilizers, and reduce the use of chemical fertilizers. Organic fertilizers can be selected from decomposed farmyard manure, compost or commercial organic fertilizers, with an application rate of 1000-1500 kg per mu. Through the decomposition of microorganisms, the cohesion between soil particles is improved and the formation of soil aggregates is promoted. At the same time, apply an appropriate amount of soil conditioner, such as gypsum, to adjust the soil pH, reduce the content of sodium ions in the soil, reduce the dispersion of soil particles, and enhance the stability of soil structure. Among them, biological improvement measures include planting green manure crops, such as astragalus and clover. After the autumn harvest, the green manure seeds are evenly spread in the farmland and pressed into the soil in the spring of the following year. During the decomposition process, green manure can increase the organic matter content of the soil, improve the structure of the soil microbial community, and promote the formation of soil aggregate structure, thereby alleviating soil compaction. In addition, beneficial microbial agents, such as Bacillus subtilis and phosphate- and potassium-solubilizing bacteria, can be introduced to improve the soil nutrient cycle and enhance soil fertility and structural stability through the metabolic activities of microorganisms. If the soil compaction degree corresponding to the target area is secondary compaction, the soil compaction treatment measures corresponding to the target area are determined to be at least one of deep plowing combined with soil conditioners, optimization of irrigation and drainage management, and adjustment of crop rotation and intercropping systems.

[0039] Among them, deep plowing combined with soil conditioner is to carry out deep plowing operation, the depth reaches 25-30 cm, and break the compaction of deeper soil layers. During the deep plowing process, soil conditioner, such as polyacrylamide (PAM), is evenly mixed in, with an amount of 1-2 kg per mu. PAM can promote the agglomeration of soil particles through adsorption and bridging, form larger soil aggregates, and improve soil structure. At the same time, in conjunction with deep plowing, the soil is fertilized in layers, and organic fertilizers and chemical fertilizers are applied at different depths to meet the nutrient needs of crops at different growth stages and improve fertilizer utilization. Irrigation and drainage management optimization is to improve the farmland irrigation and drainage system to avoid excessive soil moisture or dryness. Use water-saving irrigation methods such as drip irrigation and sprinkler irrigation to control the amount and frequency of irrigation water, keep soil moisture within an appropriate range, and generally maintain the relative soil moisture content at 60%-80%. At the same time, strengthen the construction of drainage facilities to ensure that field water is removed in time during the rainy season to prevent the soil from becoming more compacted due to long-term water accumulation. For example, drainage pipes are set up in the fields, with the spacing determined according to the soil texture and topography, generally 5-10 meters, to effectively drain excess water. Crop rotation and intercropping system adjustment is to implement crop rotation and intercropping system.

[0040] Among them, the secondary compaction is more serious than the primary compaction.

[0041] This step will be described in detail below.

[0042] The soil compaction treatment method provided in the embodiment of the present application obtains the soil remote sensing image, soil spectral data and soil radar reflection data corresponding to the target area. Then, the soil remote sensing image, soil spectral data and soil radar reflection data are identified to determine the soil compaction degree corresponding to the target area, thereby ensuring the accuracy of determining the soil compaction degree corresponding to the target area. Thus, there is no need to manually determine the soil compaction degree corresponding to the target area, thereby improving the efficiency of determining the soil compaction degree corresponding to the target area and saving manpower costs. According to the soil compaction degree corresponding to the target area, the soil compaction treatment measures corresponding to the target area are determined, thereby ensuring that the output soil compaction treatment measures correspond to the soil compaction degree of the target area, thereby ensuring the accuracy of the output soil compaction treatment measures. Thus, the agricultural production efficiency can be improved.

[0043] In this embodiment, a soil compaction treatment method is provided, which can be used for the above-mentioned electronic equipment. Figure 2 : is a flow chart of a soil compaction treatment method according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps: Step S201, obtaining soil remote sensing images, soil spectrum data and soil radar reflection data corresponding to the target area.

[0044] For details about this step, please refer to the above description of step S101, which will not be elaborated here.

[0045] Step S202, identifying the soil remote sensing image, soil spectral data and soil radar reflection data to determine the soil compaction degree corresponding to the target area.

[0046] Specifically, the above step S202 may include the following steps: Step S2021, identifying the soil remote sensing image to determine the first soil texture feature corresponding to the target area.

[0047] Specifically, when the electronic device recognizes the soil remote sensing image, it first uses the convolutional neural network (CNN) in deep learning to perform preliminary feature extraction on the original soil remote sensing image. The extracted features are then input into the trained generative adversarial network (GAN) model, and the generator enhances and supplements these features to generate a richer and more accurate representation of soil texture features. The discriminator evaluates the generated features to ensure their authenticity and effectiveness. In this way, more refined first soil texture features can be obtained, such as the ability to identify detailed information such as the arrangement of tiny particles in the soil and the directionality of the texture, which helps to gain a deeper understanding of the physical structure of the soil surface. Among them, the generative adversarial network consists of a generator and a discriminator. The generator learns the potential pattern of texture in the soil remote sensing image and tries to generate realistic soil texture images, while the discriminator is responsible for distinguishing the generated images from the real soil remote sensing images. During the training process, the two confront each other and evolve together, and the generator is gradually able to generate high-quality soil texture images, thereby enhancing the expression of soil texture features.

[0048] Step S2022, identifying the soil spectrum data to determine the first absorption peak characteristics corresponding to the target area.

[0049] Specifically, electronic equipment can use time series analysis algorithms, such as the autoregressive integrated moving average (ARIMA) model, to identify soil spectral data and determine the change trend of absorption peak characteristics over time. At the same time, combined with geographic information system (GIS) technology, spatial interpolation and analysis of soil spectral data at different spatial locations are performed to draw a spatial distribution map of absorption peak characteristics, thereby determining the first absorption peak characteristics corresponding to the target area.

[0050] For example, through the analysis of spatiotemporal sequence spectra, it was found that over time, the organic matter content in the soil of a certain area gradually decreased, resulting in a weakening of the absorption peak intensity at the preset wavelength; spatially, the absorption peak characteristics of the soil near the river are significantly different from those of the soil far from the river due to moisture conditions and sedimentation. This analysis method provides a new perspective for studying the soil evolution process and regional soil characteristic differences.

[0051] Step S2023, identifying the soil radar reflection data to determine the first wave crest and wave trough features corresponding to the target area.

[0052] Specifically, the electronic device can identify the soil radar reflection data based on the deep learning model to determine the first peak and valley characteristics corresponding to the target area.

[0053] Among them, the deep learning model is a hybrid model of recurrent neural network (RNN) and convolutional neural network (CNN). CNN is good at extracting local features of data and can identify local characteristic patterns of peaks and valleys in soil radar reflection data; RNN can process time series data and analyze the changing trends of peaks and valleys at different measurement times. By learning a large amount of data, the deep learning model can automatically extract complex and accurate peak and valley features.

[0054] Step S2024, fusing the first soil texture feature, the first absorption peak feature, and the first wave crest and wave valley feature to generate a target quantum bit feature.

[0055] Specifically, the above step S2024 may include the following steps: Step a1, identifying the first soil texture feature, converting the geometric shape and spatial distribution information in the first soil texture feature into a state combination of quantum bits, and generating texture feature quantum bits.

[0056] Specifically, when identifying the first soil texture feature, the electronic device can use advanced artificial intelligence algorithms, especially convolutional neural networks (CNNs) based on deep learning. CNNs can automatically learn complex patterns in soil texture images and accurately identify geometric shapes such as round, elliptical, irregular polygonal shapes of particles, as well as their spatial distribution information in soil samples.

[0057] Next, electronic devices can use quantum simulation technology to convert the identified geometric shape and spatial distribution information of the soil into a combination of quantum bit states. Specifically, quantum simulation is a method of using quantum systems to simulate the behavior of other quantum or classical systems. In this process, we map the geometric shape and spatial distribution information of the soil texture to the state of the quantum system. For example, different geometric shapes of particles in the soil can correspond to different spin states of quantum bits; and the spatial distribution information of the particles can be represented by the interaction between quantum bits.

[0058] Finally, based on quantum entanglement and quantum gate operations, the electronic device can efficiently encode soil texture feature information into quantum bits. Specifically, the electronic device converts the geometric shape and spatial distribution parameters of each soil texture feature into a series of quantum gate operation instructions, and generates the corresponding texture feature quantum bits by performing these operations on the quantum bits.

[0059] This method of converting soil texture features into quantum bits has many potential advantages. On the one hand, quantum bits have the characteristics of quantum superposition and quantum entanglement, and can store and process a large amount of information at the same time, which allows us to conduct a more comprehensive and in-depth analysis of soil texture features. On the other hand, algorithms based on quantum computing have significant advantages in processing complex data patterns and optimization problems, which can help us more quickly explore the relationship between soil texture and soil properties and crop growth, and provide stronger support for precision agriculture, soil protection and other fields.

[0060] Step a2, identifying the first absorption peak feature, and determining the intensity and position information of the absorption peaks at different wavelengths in the first absorption peak feature.

[0061] Specifically, the electronic device can input the first absorption peak feature into a trained RNN or LSTM model. The model can analyze whether the absorption peak position moves and how the intensity changes over time, thereby outputting the intensity and position information of the absorption peaks at different wavelengths in the first absorption peak feature.

[0062] Step a3, quantize the intensity and position information of each absorption peak to generate absorption peak characteristic quantum bits.

[0063] Specifically, first, the electronic device can determine the number of quantum bits to be encoded based on the absorption peak intensity and position information. Then, according to the selected quantum error correction code rules, the probability amplitude corresponding to the intensity of the absorption peak and the phase information corresponding to the position are encoded into multiple quantum bits. In the actual storage and transmission of these absorption peak characteristic quantum bits, even if some quantum bits are interfered by environmental noise, the original accurate information can be restored through the error correction operation of the quantum error correction code. This is of great significance for long-term monitoring of soil spectral data and ensuring the integrity and accuracy of absorption peak information when transmitting data in complex environments.

[0064] Step a4, identifying the first wave peak and trough characteristics, using quantum bits to encode the amplitude, frequency and phase of the wave, and generating peak and trough characteristic quantum bits.

[0065] Specifically, the electronic device can input the first wave peak and trough characteristics into the quantum neural network. The quantum neurons in the quantum neural network process the first wave peak and trough characteristics through quantum gate operations, extract key features such as the amplitude, frequency and phase of the peaks and troughs, and encode these features into the state of the quantum bit. The trained quantum neural network can quickly and accurately convert radar data into peak and trough characteristic quantum bits. For example, when analyzing large-area soil radar data, this method can greatly shorten the data processing time while improving the quality of encoding, providing an efficient means for quickly understanding soil conditions.

[0066] Step a5, fusing the texture feature quantum bits, absorption peak feature quantum bits, and peak and valley feature quantum bits to generate target quantum bit features.

[0067] Specifically, Figure 3 As shown, the above step a5 may include the following steps: Step a51, pre-processing the texture feature quantum bits, absorption peak feature quantum bits, and peak and valley feature quantum bits.

[0068] Specifically, for texture feature qubits, the electronic device uses the quantum Fourier transform in the quantum image transformation algorithm to convert the spatial domain information of the texture feature qubits into the frequency domain, highlighting the periodicity and directional characteristics of the texture feature qubits. For example, for the striped texture on the soil surface, it will appear as a peak of the preset frequency in the frequency domain. Then, the texture feature qubits are multi-scale decomposed using quantum wavelet transform to obtain texture details at different scales from macro to micro. Large-scale wavelet transform can identify texture trends over large areas, while small-scale wavelet transform focuses on subtle texture changes, such as the local arrangement details of soil particles. These transformed texture feature qubits are adjusted to the information dimension that matches the other two types of feature qubits for subsequent efficient entanglement fusion. For the absorption peak feature qubits, the absorption peak feature qubits carry key information about the soil composition and properties in the soil spectrum. After accessing the soil multimodal entanglement gate, they are preprocessed with the help of quantum chemical simulation algorithms. According to the soil chemical components corresponding to different absorption peaks, such as organic matter, iron oxides, clay minerals, etc., their chemical reaction processes under different environmental conditions are simulated. Specifically, by adjusting the phase and probability amplitude of the absorption peak characteristic quantum bits, the intensity, position, and peak shape of the absorption peak are re-encoded. For example, for the absorption peak representing the organic matter content, when the organic matter content in the soil increases, the probability amplitude of the corresponding quantum bit increases, and the phase is also preset and adjusted according to the chemical reaction kinetics. In this way, after preprocessing, the absorption peak characteristic quantum bits can be more accurately entangled with other characteristic quantum bits, providing richer chemical information for mining soil characteristic associations. For the peak and valley characteristic quantum bits, the peak and valley characteristic quantum bits reflect the information about the deep structure of the soil in the soil radar reflection data. After accessing the soil multimodal entanglement gate, the quantum Fourier transform is used to re-encode its frequency and phase information. When the radar wave propagates in the soil, the soil structure at different depths will cause the frequency and phase of the reflected wave to change, and these change information can be clearly extracted through the quantum Fourier transform. Combined with the physical properties of the soil, such as soil compactness and pore structure, the peak and valley characteristic quantum bits are further adjusted. For example, when there is a compaction layer in the soil, the peak amplitude of the radar reflected wave will increase and the trough depth will become shallower. The state of the peak and trough characteristic quantum bits will be adjusted accordingly, so that they can better interact with the texture and absorption peak characteristic quantum bits in the subsequent entanglement process, thereby fully reflecting the relationship between the physical structure and other characteristics of the soil.

[0069] Step a52, input the preprocessed texture feature quantum bits, absorption peak feature quantum bits and peak and valley feature quantum bits into the soil multimodal entanglement gate.

[0070] Specifically, the electronic device inputs the preprocessed texture feature quantum bits, absorption peak feature quantum bits, and peak and trough feature quantum bits into the soil multimodal entanglement gate.

[0071] Step a53, using the pulse sequence in the soil multimodal entanglement gate to control the interaction between the texture characteristic quantum bits, the absorption peak characteristic quantum bits, and the peak and trough characteristic quantum bits.

[0072] The frequency, amplitude and duration of the pulse sequence are obtained through multiple iterative training of the soil multimodal entanglement gate.

[0073] Specifically, when constructing a soil multimodal entanglement gate, the electronic device can start a multi-objective optimization algorithm. The multi-objective optimization algorithm continuously evolves in the parameter space to generate a series of pulse sequence schemes. For example, one scheme performs well in entanglement effect, but has high energy consumption; another scheme has good entanglement stability, but the generated target quantum bit characteristics are slightly less interpretable. Then, according to the experimental conditions and research focus, such as when the energy consumption requirements are not high but the entanglement effect and interpretability requirements are high, the corresponding pulse sequence parameters are selected from the non-dominated solutions to achieve comprehensive optimization of the performance of the soil multimodal entanglement gate, and better serve soil science research.

[0074] Finally, the pulse sequence in the soil multimodal entanglement gate is used to control the interaction between texture characteristic qubits, absorption peak characteristic qubits, and peak and trough characteristic qubits.

[0075] Step a54, using the Sierpinski triangle topological structure in the soil multimodal entanglement gate, the preprocessed texture feature quantum bits, absorption peak feature quantum bits, and peak and trough feature quantum bits are transmitted and entangled between nodes to generate a multi-body entangled state.

[0076] Specifically, the pre-processed texture feature qubits, absorption peak feature qubits, and wave crest and trough feature qubits are placed at different nodes of the Sierpinski triangle topological structure. For example, the texture feature qubits can be placed at the vertex nodes of the triangle, the absorption peak feature qubits can be placed at the midpoint nodes of the edges, and the wave crest and trough feature qubits can be placed at other suitable node positions so that they can interact with each other through the connections between the nodes.

[0077] Due to the connectivity characteristics of the Sierpinski triangle topology, the preprocessed texture feature qubits, absorption peak feature qubits, and peak and valley feature qubits can be transferred between nodes along the edges of the triangle. During the transfer process, the qubits on each node will carry their own feature information and interact with the qubits on other nodes. For example, when a texture feature qubit is transferred from a vertex node to an adjacent edge midpoint node, it will meet the absorption peak feature qubit on that node, and the quantum states may be exchanged or superimposed between the two, thus starting the entanglement process.

[0078] In order to achieve a more complex entangled state, the qubits will be transferred between the nodes of the Sierpinski triangle for multiple rounds. Each round of transfer will cause the qubit to interact with more different types of qubits, further enriching the entanglement relationship between them. For example, after several rounds of transfer, the texture feature qubit may be entangled with multiple absorption peak feature qubits and wave crest and trough feature qubits, forming a preliminary state of multi-body entanglement.

[0079] In the process of quantum bits passing between each other, the entanglement between them is achieved through quantum interactions. For example, when two quantum bits meet at the same node, they may become entangled through some quantum gate operations (such as CNOT gates, Hadamard gates, etc.). These quantum gate operations will change their quantum states according to the states of the quantum bits and the rules of interaction, so that a non-classical correlation is formed between them, that is, a multi-body entangled state is generated.

[0080] Step a55, post-process the multi-body entangled state to extract the target quantum bit characteristics.

[0081] Specifically, the above step a55 may include the following steps: Step a551, reconstruct the density matrix based on the multi-body entangled state.

[0082] Specifically, the multi-body entangled state composed of multiple quantum bits is represented by |Ψ〉. Then, according to the definition of the density matrix, ρ=|Ψ〉〈Ψ| is calculated, and the expression of the density matrix is ​​obtained by performing an outer product operation on the multi-body entangled state. This process involves operations such as the tensor product of quantum states.

[0083] Finally, the obtained density matrix is ​​normalized to ensure that its trace is 1, that is, , and obtain the final reconstructed density matrix ρ norm .

[0084] Step a552, introduce the quantum version of principal component analysis, perform principal component analysis on the density matrix, and determine the main components by finding the eigenvalues ​​and eigenvectors of the density matrix.

[0085] Specifically, for a given density matrix ρ, its eigenvalue equation is ρ|v i 〉=λ i ∣v i 〉, where λ i is the eigenvalue, |v i 〉 is the corresponding eigenvector.

[0086] Electronic devices can solve eigenvalues ​​and eigenvectors through numerical calculation methods or by using quantum algorithms. In quantum computing, some algorithms such as quantum phase estimation (QPE) can be used to accurately calculate eigenvalues ​​and eigenvectors. Specifically, the QPE algorithm uses quantum gate operations and the evolution of quantum states to encode the eigenvalue information of the density matrix in the phase of the quantum bit, and then extracts the eigenvalues ​​and eigenvectors through measurement and post-processing.

[0087] Then, the eigenvectors are usually sorted according to the size of the eigenvalues, and the eigenvectors with larger eigenvalues ​​are selected as the main components. For example, a threshold can be set to select the components corresponding to the eigenvectors with eigenvalues ​​greater than the threshold as the main components; or the first k eigenvectors with the largest eigenvalues ​​are selected, where k is a parameter determined according to specific problems and requirements. These main components can capture most of the information in the density matrix, thereby achieving effective dimensionality reduction and feature extraction of quantum systems.

[0088] Step a553, determine the target quantum bit characteristics based on the main components.

[0089] Specifically, the electronic device can extract information related to the target from the eigenvector corresponding to the main component. This may involve analyzing certain preset parameters of the quantum state, such as the phase and amplitude of the quantum bit. For example, if we are concerned about the fertility characteristics of the soil, we may need to extract quantum information related to the nutrient content of the soil from the eigenvector. Then, the extracted information is quantized and converted into specific eigenvalues. Next, the electronic device performs correlation analysis on the extracted quantitative features to determine the degree of correlation between them. Some features may be highly correlated, which means that they may reflect the same physical phenomenon or soil characteristics. By removing features with high correlation, redundant information can be reduced and the effectiveness of the features can be improved. Finally, the electronic device selects appropriate features for combination based on the results of the correlation analysis. Linear or nonlinear combinations can be used to construct features that can more comprehensively and accurately describe the target quantum bit features. For example, multiple related features can be combined into a new feature by weighted summation, and the weight can be determined based on the size of the eigenvalue.

[0090] Step S2025, input the target quantum bit characteristics into the preset soil compaction determination model, and output the soil compaction degree corresponding to the target area.

[0091] Specifically, the above step S2026 may include the following steps: Step b1, input the target quantum bit characteristics into the preset soil compaction determination model.

[0092] Specifically, the electronic device inputs the target quantum bit characteristics into a preset soil compaction determination model.

[0093] Step b2: The multi-scale feature extraction module extracts the target quantum bit features based on quantum filters of different scales and outputs quantum bits of different scales.

[0094] Specifically, the multi-scale feature extraction module in the preset soil compaction determination model includes quantum filters of different scales. When the target quantum bit feature is input into the small-scale quantum filter, the small-scale quantum filter performs a series of quantum gate operations on the target quantum bit feature according to its own parameter settings. These operations may include single-bit quantum gate operations (such as Hadamard gate, Pauli gate, etc.) and multi-bit quantum gate operations (such as CNOT gate, Toffoli gate, etc.). Through these operations, the state between quantum bits changes, and the filter extracts the local detail information related to the small scale in the target quantum bit feature. For example, when analyzing soil characteristics, small-scale features may reflect microscopic information such as the distribution and interaction of tiny particles in the soil.

[0095] For the mesoscale quantum filter, the operation process is similar to that of the small scale, but the parameter settings are different. The mesoscale filter processes the quantum bit features within a moderate range, which can capture certain local details while taking into account some overall structural information. Through the preset quantum gate operation sequence and quantum state evolution, the mesoscale-related features in the target quantum bit features are extracted. For example, in soil analysis, the mesoscale features may be related to the structure and properties of soil aggregates.

[0096] The large-scale quantum filter focuses on the overall macroscopic information of the target quantum bit characteristics. It processes quantum bits through more complex quantum gate operations and stronger quantum entanglement to extract information that reflects the overall trend and structure of the characteristics. For example, in soil research, large-scale characteristics may be related to macroscopic information such as soil topography and soil texture distribution over a large area.

[0097] Step b3: The quantum convolution module performs sliding convolution on quantum bits of different scales at different positions through quantum gate operations to generate convolution bit features.

[0098] Specifically, the electronic device places the quantum convolution kernel at the starting position of the input quantum bits of different scales. Then, the sequence of quantum gate operations in the convolution kernel is applied to the region of quantum bits of different scales covered by the convolution kernel. This will cause the state of the quantum bits in the region to change, and through the action of the quantum gate, the characteristic information of the local region is extracted. For example, if the convolution kernel covers several quantum bits in a superposition state, after the quantum gate operation, the superposition state of these quantum bits may change, generating a new quantum state, which contains characteristic information related to the original quantum bit state.

[0099] After completing the operation at the initial position, the quantum convolution kernel is slid on the different scales of the input quantum bits according to a certain step size. The step size determines the distance the convolution kernel moves each time, which is similar to the sliding step size in classical convolution. After sliding to the new position, the same sequence of quantum gate operations is applied to the quantum bit area covered by the convolution kernel again. This process is repeated until the convolution kernel traverses the entire input quantum bits of different scales. In each slide and operation, the convolution kernel extracts local feature information at different positions.

[0100] As the convolution kernel slides and operates, a series of transformed qubit states are generated at different locations. These qubit states contain the characteristic information of the input qubits of different scales in each local area. These qubit states generated at different locations are integrated to form a new set of qubits, namely the convolution bit feature.

[0101] Step b4: the quantum residual connection layer adds the convolution bit features and the target quantum bit features to output the final quantum bit.

[0102] Specifically, the quantum residual connection layer receives two inputs, namely the convolution bit features generated by the quantum convolution module and the target quantum bit features previously determined. Both features exist in the form of quantum states and contain different aspects of soil information in the target area. The target quantum bit features are obtained through a series of processing such as principal component analysis of the density matrix, reflecting the key characteristics of the soil; while the convolution bit features are local feature information at different scales extracted through quantum convolution operations.

[0103] In quantum computing, the "addition" operation is not a simple numerical addition like in classical computing, but is based on the superposition principle of quantum states. For a quantum bit, its state can be represented by the Dirac symbol |ψ〉. For example, the state of a quantum bit can be , where α and β are complex numbers and satisfy When two qubit features are added together, their quantum states are actually superimposed. Assuming that the quantum state corresponding to the convolution bit feature is |Φ1〉, and the quantum state corresponding to the target qubit feature is |Φ2〉, then the quantum state after addition |Φ〉=|Φ1〉+|Φ2〉. In this process, the superposition and entanglement relationship between qubits will change accordingly, thus generating a new quantum state that contains the combined information of both.

[0104] After the above addition operation, the quantum bits corresponding to the new quantum state are the final quantum bits. These final quantum bits integrate the information of the convolution bit features and the target quantum bit features, not only retaining the key information in the original target quantum bit features, but also incorporating the local features and multi-scale information extracted by the quantum convolution operation, making the final quantum bits more comprehensive and rich in their description of soil characteristics.

[0105] Step b5, based on the final quantum bit, output the soil compaction process corresponding to the target area.

[0106] Specifically, after obtaining the final qubit, it is necessary to establish a mapping relationship between the final qubit and the degree of soil compaction. This is usually achieved by training a quantum machine learning model or a classical machine learning model. First, a large number of samples of the target area with known soil compaction degrees are collected, and the final qubit corresponding to each sample is obtained according to the previous process. Then, the model is trained using the final qubits of these samples as input and the degree of soil compaction as output. During the training process, the model will learn the intrinsic connection between the quantum state of the final qubit and the degree of soil compaction.

[0107] The trained model can be used to predict the degree of soil compaction in a new target area. When the final quantum bit of a new target area is input, the model predicts the degree of soil compaction based on the mapping relationship learned during training. For example, if the model is a neural network-based model, the quantum state information of the final quantum bit will be processed through the various layers of the neural network, and finally a numerical value or classification result representing the degree of soil compaction will be output. The numerical result can be a continuous scalar, such as between 0 and 1 to represent the severity of the degree of soil compaction; the classification result can classify the degree of soil compaction into different levels, such as "slightly compacted", "moderately compacted", "severely compacted", etc.

[0108] Step S203, determining soil compaction treatment measures corresponding to the target area according to the soil compaction degree corresponding to the target area.

[0109] For details about this step, please refer to the above description of step S103, which will not be elaborated here.

[0110] The soil compaction treatment method provided in the embodiment of the present application identifies the soil remote sensing image and determines the first soil texture feature corresponding to the target area. By identifying and determining the first soil texture feature, the overall texture information of the soil surface, such as the roughness of the soil surface, the directionality and periodicity of the texture, etc., can be quickly obtained. This information is helpful for preliminarily judging the topography of the soil, the impact of vegetation coverage on the soil, etc. In addition, remote sensing technology has the characteristics of wide coverage and fast data acquisition speed. By identifying the soil remote sensing image, a large area of ​​the target area can be monitored in a short time, and changes in soil texture features can be discovered in time. Compared with the traditional field measurement method, the work efficiency is greatly improved and the manpower and time costs are reduced. The soil spectral data is identified to determine the first absorption peak feature corresponding to the target area. By identifying and determining the first absorption peak feature, the content and state of various minerals, organic matter, water and other components in the soil can be accurately analyzed. Different soil components have preset absorption peaks on the spectrum. In addition, the soil spectral data can sensitively reflect the subtle changes in soil components. By monitoring the changes in the absorption peak features, changes in soil components can be detected before problems such as soil compaction are clearly manifested, thereby achieving early warning. The soil radar reflection data is identified to determine the first wave crest and trough characteristics corresponding to the target area. The soil radar reflection data can penetrate the soil surface and obtain the structural information of the deep soil layer. By identifying and determining the first wave crest and trough characteristics, the pore structure, stratification, and whether there is a compaction layer in the deep soil layer can be clearly understood. This information is crucial for evaluating the aeration, water permeability, and root growth environment of the soil. The first soil texture feature is identified, and the geometric shape and spatial distribution information in the first soil texture feature are converted into a state combination of quantum bits to generate texture feature quantum bits. Thus, the complex information of soil texture can be accurately represented in the form of quantum states. The superposition state characteristics of quantum bits make it possible to simultaneously represent the possibility of multiple geometric shapes and spatial distributions. In addition, the generated texture feature quantum bits can be directly applied to quantum computing environments. Quantum computing has the ability of parallel computing and can process the states of multiple quantum bits at the same time, greatly improving the processing speed and analysis efficiency of soil texture features. The first absorption peak feature is identified, the intensity and position information of the absorption peaks at different wavelengths in the first absorption peak feature are determined, the intensity and position information of each absorption peak are quantized, and the absorption peak feature quantum bits are generated. This enables in-depth mining of soil composition information, because different soil components have preset absorption peak characteristics in the spectrum. By quantizing, the intensity and position information of the absorption peak is converted into the state of quantum bits, which can more accurately represent the content and state of various minerals, organic matter and other components in the soil. The first wave peak and valley characteristics are identified, and the amplitude, frequency and phase of the wave are encoded using quantum bits to generate peak and valley characteristic quantum bits.Thus, the deep soil structure information contained in the peak and trough features in the soil radar reflection data can be accurately represented. The encoding method of quantum bits can more accurately capture the subtle changes of waves. The texture feature quantum bits, absorption peak feature quantum bits, and peak and trough feature quantum bits are fused to generate target quantum bit features. The deep fusion of multi-source soil information is achieved. Different types of feature quantum bits represent information on the surface texture, composition, and deep structure of the soil. Through fusion, these scattered information can be organically combined to form a more comprehensive and in-depth description of soil characteristics. This deep fusion can dig out the intrinsic connection between different levels and aspects of the soil. The target quantum bit feature is input into the preset soil compaction determination model, and the soil compaction degree corresponding to the target area is output. The preset soil compaction determination model analyzes and judges based on the fused target quantum bit features, and can more accurately determine the soil compaction degree of the target area. Since the target quantum bit feature integrates the key information of multi-source data, the model can comprehensively consider multiple factors such as the surface morphology, composition, and deep structure of the soil, avoiding the limitations of single feature judgment and improving the accuracy and reliability of judgment.

[0111] In this embodiment, a soil compaction treatment method is provided, which can be used for the above-mentioned electronic equipment. Figure 4 : is a flow chart of a soil compaction treatment method according to an embodiment of the present invention. Figure 4 As shown, the process includes the following steps: Step S301, obtaining soil remote sensing images, soil spectrum data and soil radar reflection data corresponding to the target area.

[0112] For details about this step, please refer to the above description of step S201, which will not be elaborated here.

[0113] Step S302, identifying the soil remote sensing image, soil spectral data and soil radar reflection data to determine the soil compaction degree corresponding to the target area.

[0114] For details about this step, please refer to the above description of step S202, which will not be repeated here. Step S303, determining soil compaction treatment measures corresponding to the target area according to the soil compaction degree corresponding to the target area.

[0115] Specifically, the above step S303 may include the following steps: Step S3031, identifying the soil remote sensing image, soil spectral data and soil radar reflection data to determine the cause of soil compaction corresponding to the target area.

[0116] Specifically, the above step S3031 may include the following steps: Step c1, performing data preprocessing on the soil remote sensing image, and extracting a second soil texture feature corresponding to the soil remote sensing image.

[0117] Specifically, the electronic device can first perform radiation correction and geometric correction on the soil remote sensing image. Radiation correction is used to eliminate the image radiation difference caused by factors such as lighting conditions and atmospheric scattering, so that the remote sensing images obtained at different times and locations are comparable. Geometric correction corrects the image geometric deformation caused by sensor posture, terrain undulation, etc., to ensure the accuracy of information such as soil texture patterns in the image. Then, the soil remote sensing image is enhanced, and edge enhancement, contrast stretching and other methods are used to highlight the texture features of the soil surface, so as to facilitate the subsequent identification of texture patterns related to soil compaction. Texture feature extraction is performed on the pre-processed soil remote sensing image to obtain the second soil texture feature. Specifically, the electronic device uses gray level co-occurrence matrix, wavelet texture analysis and other methods to extract the roughness, directionality, contrast and other features of the soil surface texture. Identify the compact texture pattern formed by the aggregation of fine particles, and analyze its distribution range and intensity in the soil remote sensing image. For example, in the soil compaction area, the texture roughness is reduced, showing a more uniform and compact texture feature. By quantifying the changes in these texture features, the arrangement state of the particles on the soil surface and the possibility of soil compaction can be judged.

[0118] Step c2, performing data preprocessing on the soil spectrum data, and extracting the second absorption peak feature corresponding to the absorption peak of the preset band.

[0119] Specifically, since the absorption peak of the preset band is closely related to the mineral composition of the soil, the original soil spectral data needs to be denoised first. Specifically, electronic equipment can use algorithms such as wavelet transform to remove abnormal signals caused by instrument noise and environmental interference to ensure the accuracy of soil spectral data. After that, the soil spectral data is normalized to unify the soil spectral data obtained under different measurement conditions to the same scale range for subsequent analysis of the characteristic changes of the absorption peak of the preset band (such as the near-infrared band 1400-1900nm). In the pre-processed soil spectral data, the characteristics of the absorption peak of the preset band are extracted. Including parameters such as the position, intensity, and half-height width of the absorption peak. By accurately measuring and analyzing these parameters, the content and state changes of clay minerals such as kaolinite and montmorillonite in the soil can be judged. For example, when the absorption peak intensity increases and the position moves toward the long-wave direction, it may indicate that the soil water content increases and the clay minerals swell, which is related to the changes in clay minerals during soil compaction.

[0120] Among them, the preset band can be a band of 1400-1900nm, or a band of 1900nm-2100nm. The embodiment of the present application does not specifically limit the preset band.

[0121] Step c3, preprocessing the soil radar reflection data and extracting pore structure features related to the soil pore structure.

[0122] Specifically, the electronic device can use a filtering algorithm to remove noise from the soil radar reflection data. Common filtering methods include mean filtering, median filtering, Gaussian filtering, etc. Then, the electronic device corrects the soil radar reflection data, including time correction and amplitude correction. Time correction is used to ensure the accuracy of the radar wave propagation time, and amplitude correction is to make the data at different locations comparable. Normalization is to map the amplitude value of the data to a preset range, such as [0, 1] or [-1, 1], to facilitate subsequent analysis and processing. In addition, during the data acquisition process, there may be missing data or uneven sampling. At this time, an interpolation algorithm can be used to fill in the missing data points to make the soil radar reflection data more continuous and smooth. Common interpolation methods include linear interpolation, spline interpolation, etc.

[0123] Next, the electronic device can identify different reflection interfaces by analyzing the reflection wave characteristics in the soil radar reflection data, such as the intensity, phase and arrival time of the reflection wave. For example, when the radar wave encounters pores or voids in the soil, a strong reflection wave will be generated. By detecting the position and characteristics of these strong reflection waves, the approximate position and shape of the pores can be determined. The waveform of the radar reflection wave is analyzed to extract characteristic parameters related to the pore structure. For example, parameters such as the width, height, and slope of the waveform may be related to the size, shape, and distribution of the pores. By calculating these parameters, the characteristics of the soil pore structure can be quantitatively described. In addition, the electronic device converts the soil radar reflection data from the time domain to the frequency domain for spectrum analysis. Different pore structures will produce different responses to different frequency components of the radar wave. By analyzing the characteristics such as the peak position, amplitude, and width in the spectrum graph, the information about the soil pore structure can be further understood. For example, the attenuation of high-frequency components may be related to the size and connectivity of the soil pores.

[0124] The processed soil radar reflection data is converted into an image form, and the soil pore area in the image is separated from other areas using an image segmentation algorithm. Then, feature extraction is performed on the segmented pore area, such as calculating the geometric features such as the area, perimeter, and shape factor of the pores, as well as the topological features such as the connectivity and distribution density of the pores, so as to obtain the pore structure features related to the soil pore structure. These features can more intuitively reflect the characteristics of the soil pore structure.

[0125] Step c4, inputting the second soil texture feature, the second absorption peak feature and the pore structure feature into a preset cause identification model, and outputting the soil compaction cause corresponding to the target area.

[0126] Specifically, the above step c4 may include the following steps: Step c41, normalizing the second soil texture feature, the second absorption peak feature and the pore structure feature.

[0127] Specifically, the electronic device normalizes the second soil texture feature, the second absorption peak feature, and the pore structure feature.

[0128] Step c42, inputting the normalized second soil texture characteristics, second absorption peak characteristics and pore structure characteristics into a preset cause identification model.

[0129] Specifically, the electronic device can input the normalized second soil texture features, the second absorption peak features, and the pore structure features into a preset cause identification model.

[0130] Step c43, multiple neurons in the hidden layer of the preset cause recognition model perform nonlinear transformation on the input data through the activation function and output the transformation features.

[0131] Specifically, after the normalized second soil texture feature, the second absorption peak feature, and the pore structure feature enter the input layer of the preset cause identification model, matrix multiplication is performed with the preset weights between the layers of the preset cause identification model. In a model based on deep learning, the hidden layer usually contains multiple neurons, and each neuron performs a nonlinear transformation on the input data through an activation function. For example, in a multi-layer perceptron structure, after the input data vector is multiplied by the weight matrix from the input layer to the first hidden layer, it passes through the ReLU (Rectified Linear Unit) activation function to change the negative value to 0 and retain the positive value, which can increase the nonlinear expression ability of the model and enable the model to learn more complex feature relationships. In a convolutional neural network, the input data vector (which can be regarded as feature data in the form of an image at this time) will pass through the convolution layer, and the convolution kernel slides on the data to perform a convolution operation to extract local features, and then passes through the pooling layer for dimensionality reduction processing to reduce the amount of data and improve the operation efficiency, and then the processed data is passed to the next hidden layer.

[0132] Between the multiple hidden layers of the preset cause identification model, the data continues to undergo similar matrix multiplication and nonlinear transformation operations. As the data is transferred between the hidden layers, the preset cause identification model gradually extracts more advanced and abstract features. For example, in a model that combines decision trees with deep learning, the hidden layer may first extract comprehensive soil features through the deep learning part, and then input these features into the decision tree structure. The decision tree classifies and judges the features according to the preset decision rules, and further explores the causal relationship and logical connection between the features. In the recurrent neural network structure, the hidden layer will also consider time series information (if the data has a time dimension). Through the recurrent connection, the hidden layer state at the previous moment is combined with the current input data for calculation to better capture the change law of soil characteristics over time, which is of great significance for analyzing the evolution of soil compaction causes on a time scale. Step c44, the output layer in the preset cause identification model outputs the cause of soil compaction based on the transformation features.

[0133] Specifically, after multiple hidden layer operations, the data reaches the output layer of the preset cause identification model. The number of neurons in the output layer usually corresponds to the number of soil compaction cause categories that need to be identified. For example, if the preset model divides the causes of soil compaction into five categories: over-cultivation, unreasonable fertilization, clay mineral changes, soil erosion, etc., then the output layer has five neurons. The output layer obtains the output value of each neuron by performing matrix multiplication operations with the weights of the last hidden layer. These output values ​​represent the probability that the input data belongs to different categories of soil compaction causes. For example, a neuron output value of 0.7 means that there is a 70% probability that the soil sample corresponding to the input soil feature vector is soil compaction due to the cause corresponding to the neuron. The result of the output layer of the preset cause identification model is a probability vector, in which each element corresponds to the probability of a soil compaction cause. The output value is normalized by the Softmax function to ensure that the sum of all probabilities is 1. Then, the category with the largest probability value is selected as the output of the soil compaction cause corresponding to the target area. For example, after Softmax processing, the output probability vector is [0.1, 0.05, 0.7, 0.1, 0.05], where the probability of the third element corresponding to the cause of clay mineral change is the largest, then the model outputs that the cause of soil compaction in the target area is clay mineral change. At the same time, the model can also output the confidence interval of each category probability to reflect the credibility of the model's judgment result. For example, for the cause of clay mineral change, the confidence interval is [0.65, 0.75], indicating that the model has a high degree of confidence that the cause of soil compaction in this area is clay mineral change. In addition, the model can also provide an analysis report on the impact of input features on output results. For example, by calculating the feature importance score, it can point out the relative importance of roughness in soil texture features, intensity in absorption peak features, and pore size in pore structure features in judging the result of soil compaction caused by clay mineral changes, providing a reference basis for further research and verification of the causes of soil compaction.

[0134] Step S3032: output soil compaction treatment measures corresponding to the target area according to the cause and degree of soil compaction.

[0135] Specifically, the electronic device can determine the corresponding initial treatment measures according to the soil compaction cause, and then adjust the initial treatment measures according to the degree of soil compaction, and output the soil compaction treatment measures corresponding to the target area.

[0136] For example, for compaction caused by excessive tillage, initial treatment measures may include adopting no-till or minimum tillage technology to reduce soil disturbance; promoting a crop rotation system to select crops with different root characteristics for rotation. For compaction caused by unreasonable fertilization, initial treatment measures may include conducting soil nutrient testing and formulating precise fertilization plans based on the test results; and promoting the use of slow-release fertilizers. For compaction caused by changes in clay minerals, initial treatment measures may include regulating soil moisture to avoid excessive dryness or wetness of the soil; and applying soil conditioners. For compaction caused by soil erosion, initial treatment measures may include implementing soil and water conservation measures, such as building terraces, planting vegetation for slope protection, etc.; improving imported soil, and for areas with severe erosion and poor soil, transporting fertile imported soil to mix with the original soil.

[0137] For mild compaction, on the basis of the above reasons for compaction, strengthen the use of soil microorganisms. Apply microbial agents containing beneficial microorganisms, such as Bacillus subtilis and lactic acid bacteria, to promote the decomposition and transformation of organic matter in the soil and increase the number of soil aggregates. Apply 1-2 kg of microbial agents per mu, mix with organic fertilizer and apply to the soil. Shallow tillage and loosening of the soil can also be used in combination with returning straw to the field. After harvesting, perform shallow tillage of about 10 cm, chop the straw and spread it evenly on the field, and then plow it into the soil. During the decomposition process, the straw can increase soil organic matter, improve soil structure, and alleviate mild compaction. For moderately compacted soil caused by unreasonable fertilization, in addition to precise fertilization, soil washing can be carried out. In the rainy season or through artificial irrigation, a large amount of water can be used to wash the soil and take away some of the accumulated salt. At the same time, salt-tolerant plants such as Suaeda salsa and Salicornia can be planted to reduce the salt content of the soil.

[0138] For moderate compaction caused by changes in clay minerals, deep tillage combined with soil conditioners can be used. The deep tillage depth reaches 25-30 cm to break up the compaction layer, and 2-3 kg of soil conditioners such as polyacrylamide (PAM) are applied per mu to promote soil particle aggregation and improve soil ventilation and water permeability. For severe compaction, if it is caused by excessive tillage or soil erosion, land leveling and backfilling can be carried out. Use bulldozers and other machinery to level the land, and then backfill a large amount of high-quality soil to improve the soil texture and structure. This method is often used in soil remediation of urban construction wasteland. For severe compaction caused by changes in clay minerals or unreasonable fertilization, physical, chemical and biological remediation methods are combined. Physically, high-pressure water jets are used to break up the compaction layer; chemically, chemical improvers are used to quickly adjust the soil pH and salinity; biologically, pioneer plants such as sea buckthorn and Lespedeza, which are resistant to barrenness, are planted to gradually restore the soil ecosystem. For example, in severely compacted saline-alkali land, high-pressure water jets are first used to break up the compaction layer, then chemical improvers are applied, and finally sea buckthorn and other plants are planted for comprehensive remediation.

[0139] The soil compaction processing method provided in the embodiment of the present application performs data preprocessing on the soil remote sensing image and extracts the second soil texture feature corresponding to the soil remote sensing image, thereby improving the data quality and accurately reflecting the soil surface condition. The soil spectral data is preprocessed, and the second absorption peak feature corresponding to the preset band absorption peak is extracted therefrom; thereby improving the data quality and accurately reflecting the soil surface condition. The soil radar reflection data is preprocessed, and the pore structure features related to the soil pore structure are extracted therefrom, so as to have a deep understanding of the physical structure of the deep soil layer, including pore size, connectivity, pore wall roughness, etc. This information is essential for evaluating the aeration, water permeability and root growth environment of the soil. The second soil texture feature, the second absorption peak feature and the pore structure feature are normalized, thereby eliminating the data dimension effect and improving the generalization ability of the preset cause recognition model. The normalized second soil texture feature, the second absorption peak feature and the pore structure feature are input into the preset cause recognition model. The multiple neurons in the hidden layer of the preset cause recognition model perform nonlinear transformation on the input data through the activation function and output the transformation feature. The nonlinear transformation of the activation function enables the preset cause recognition model to learn more complex relationships and patterns in the input data. The causes of soil compaction are often caused by the interaction of multiple factors, and there are complex nonlinear relationships between these factors. By performing nonlinear transformation of the input data with the activation function, the model can capture these nonlinear relationships, thereby more accurately describing the connection between the causes of soil compaction and soil characteristics. For example, the activation function can transform the complex interactions between soil texture characteristics, absorption peak characteristics, and pore structure characteristics into a form that the model can understand and process, thereby improving the model's ability to identify the causes of soil compaction. In the preset cause recognition model, the output layer outputs the causes of soil compaction based on the transformation features. The output layer analyzes and judges based on the transformation features output by the hidden layer, and can accurately output the causes of soil compaction. After the previous normalization processing and the nonlinear transformation of the hidden layer, the input data has been converted into highly representative transformation features. Based on these transformation features, the output layer, combined with the training results and algorithms of the model, can accurately identify the main and secondary causes of soil compaction.

[0140] Then, according to the cause and degree of soil compaction, the soil compaction treatment measures corresponding to the target area are output, ensuring the accuracy of the output soil compaction treatment measures corresponding to the target area. Reasonable soil compaction treatment measures can improve soil structure, increase soil aeration and water permeability, and create a good soil environment for crop growth.

[0141] In this embodiment, a soil compaction treatment device is also provided, which is used to implement the above-mentioned embodiments and preferred embodiments, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0142] This embodiment provides a soil compaction treatment device, such as Figure 5 As shown, including: An acquisition module 401 is used to acquire soil remote sensing images, soil spectrum data, and soil radar reflection data corresponding to a target area; A determination module 402 is used to identify the soil remote sensing image, soil spectrum data and soil radar reflection data to determine the soil compaction degree corresponding to the target area; The output module 403 is used to determine the soil compaction treatment measures corresponding to the target area according to the soil compaction degree corresponding to the target area.

[0143] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0144] The soil compaction treatment device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0145] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for treating soil compaction, characterized in that: The method comprises: Obtain soil remote sensing images, soil spectral data, and soil radar reflection data corresponding to the target area; Identify the soil remote sensing image, the soil spectrum data, and the soil radar reflection data to determine the degree of soil compaction corresponding to the target area; According to the soil compaction degree corresponding to the target area, the soil compaction treatment measures corresponding to the target area are determined.

2. The method according to claim 1, characterized in that The identifying the soil remote sensing image, the soil spectral data and the soil radar reflection data to determine the soil compaction degree corresponding to the target area includes: Identifying the soil remote sensing image to determine a first soil texture feature corresponding to the target area; Identifying the soil spectrum data to determine the first absorption peak characteristics corresponding to the target area; Identify the soil radar reflection data to determine the first wave crest and wave trough characteristics corresponding to the target area; The first soil texture feature, the first absorption peak feature, and the first wave crest and wave valley feature are integrated to generate a target quantum bit feature; The target quantum bit characteristics are input into a preset soil compaction determination model to output the soil compaction degree corresponding to the target area.

3. The method according to claim 2, characterized in that The step of fusing the first soil texture feature, the first absorption peak feature, and the first wave crest and wave trough feature to generate a target quantum bit feature includes: Identifying the first soil texture feature, converting the geometric shape and spatial distribution information in the first soil texture feature into a state combination of quantum bits, and generating texture feature quantum bits; Identifying the first absorption peak feature, and determining the intensity and position information of the absorption peaks at different wavelengths in the first absorption peak feature; quantizing the intensity and position information of each absorption peak to generate absorption peak characteristic quantum bits; Identify the first wave peak and trough characteristics, encode the amplitude, frequency and phase of the wave using quantum bits, and generate peak and trough characteristic quantum bits; The texture feature quantum bits, the absorption peak feature quantum bits, and the peak and valley feature quantum bits are merged to generate the target quantum bit features.

4. The method according to claim 3, characterized in that The step of fusing the texture feature qubit, the absorption peak feature qubit, and the wave crest and trough feature qubit to generate the target qubit feature includes: Preprocessing the texture feature quantum bits, the absorption peak feature quantum bits, and the peak and valley feature quantum bits; Inputting the preprocessed texture feature qubits, the absorption peak feature qubits and the wave crest and trough feature qubits into a soil multimodal entanglement gate; The pulse sequence in the soil multimodal entanglement gate is used to control the interaction between the texture feature quantum bit, the absorption peak feature quantum bit and the wave crest and trough feature quantum bit; the frequency, amplitude and duration of the pulse sequence are obtained by multiple iterations of training the soil multimodal entanglement gate; By using the Sierpinski triangle topological structure in the soil multimodal entanglement gate, the pre-processed texture feature quantum bits, the absorption peak feature quantum bits, and the wave crest and trough feature quantum bits are mutually transmitted and entangled between nodes to generate a multi-body entangled state; The multi-body entangled state is post-processed to extract the target quantum bit characteristics therefrom.

5. The method according to claim 4, characterized in that The post-processing of the multi-body entangled state to extract the target quantum bit feature comprises: Based on the multi-body entangled state, reconstructing a density matrix; A quantum version of principal component analysis is introduced to perform principal component analysis on the density matrix, and the main components are determined by finding the eigenvalues ​​and eigenvectors of the density matrix; Based on the main components, the target quantum bit characteristics are determined.

6. The method according to claim 2, characterized in that The step of inputting the target quantum bit feature into a preset soil compaction determination model and outputting the soil compaction degree corresponding to the target area includes: Inputting the target quantum bit characteristics into a preset soil compaction determination model; The multi-scale feature extraction module extracts the target quantum bit features based on quantum filters of different scales and outputs quantum bits of different scales; The quantum convolution module performs sliding convolution on the quantum bits of different scales at different positions through quantum gate operations to generate convolution bit features; The quantum residual connection layer adds the convolution bit feature and the target quantum bit feature to output the final quantum bit; Based on the final quantum bit, the degree of soil compaction corresponding to the target area is output.

7. The method according to claim 1, characterized in that Determining soil compaction treatment measures corresponding to the target area according to the soil compaction degree corresponding to the target area includes: Identify the soil remote sensing image, the soil spectrum data, and the soil radar reflection data to determine the cause of soil compaction corresponding to the target area; According to the cause of soil compaction and the degree of soil compaction, soil compaction treatment measures corresponding to the target area are output.

8. The method according to claim 7, characterized in that The identifying the soil remote sensing image, the soil spectral data and the soil radar reflection data to determine the soil compaction cause corresponding to the target area includes: Performing data preprocessing on the soil remote sensing image, and extracting a second soil texture feature corresponding to the soil remote sensing image; Performing data preprocessing on the soil spectrum data, and extracting a second absorption peak feature corresponding to an absorption peak in a preset band; Preprocessing the soil radar reflection data and extracting pore structure characteristics related to the soil pore structure; The second soil texture feature, the second absorption peak feature and the pore structure feature are input into a preset cause identification model, and the soil compaction cause corresponding to the target area is output.

9. The method according to claim 8, characterized in that The step of inputting the second soil texture feature, the second absorption peak feature, and the pore structure feature into a preset cause identification model and outputting the soil compaction cause corresponding to the target area includes: normalizing the second soil texture feature, the second absorption peak feature, and the pore structure feature; Inputting the normalized second soil texture feature, the second absorption peak feature and the pore structure feature into a preset cause identification model; The plurality of neurons in the hidden layer of the preset cause recognition model perform nonlinear transformation on the input data through an activation function and output transformation features; The output layer in the preset cause identification model outputs the cause of soil compaction based on the transformation features.

10. A soil compaction treatment device, characterized in that: The device comprises: An acquisition module is used to acquire soil remote sensing images, soil spectral data, and soil radar reflection data corresponding to the target area; A determination module, used to identify the soil remote sensing image, the soil spectral data and the soil radar reflection data, and determine the soil compaction degree corresponding to the target area; The output module is used to determine the soil compaction treatment measures corresponding to the target area according to the soil compaction degree corresponding to the target area.

Images