Method for optimizing and adjusting soil monitoring points based on multi-level evaluation and dynamic updating

Abstract

The application provides a soil monitoring point position optimization adjustment method based on multi-level evaluation and dynamic updating, relates to the technical field of environmental protection and monitoring, and comprises the following steps: acquiring regional geographic information, attribute information of soil monitoring point positions in the region and geographic information of sub-regions in the region, judging whether soil monitoring point positions need to be newly added or adjusted, newly adding or adjusting the soil monitoring point positions based on preset soil monitoring point position layout specifications and the above information if the soil monitoring point positions need to be newly added or adjusted, and acquiring attribute information of the newly added or adjusted soil monitoring point positions. According to the application, flexible judgment steps can be respectively set for the addition and adjustment of soil monitoring point positions, so that the setting of soil monitoring point positions can be flexibly optimized based on actual monitoring requirements, monitoring convenience and the attributes of the soil monitoring point positions, which is beneficial to the dynamic updating and accurate layout of the soil environmental monitoring network and dynamically improves the scientificity and accuracy of monitoring.

Description

Technical Field

[0001] This invention relates to the field of environmental protection and monitoring technology, and in particular to a method for optimizing and adjusting soil monitoring sites based on multi-level assessment and dynamic updating. Background Technology

[0002] With the deepening of soil environmental monitoring, the scientific layout and dynamic optimization of soil monitoring sites have become crucial for improving monitoring efficiency and data representativeness. Currently, the setting of soil monitoring sites largely relies on fixed rules and static layouts, lacking a systematic integration and dynamic response mechanism for factors such as regional differences, site availability, historical data inheritance, and dynamic changes in risk.

[0003] The information disclosed in the background section of this application is intended only to enhance the understanding of the general background of this application and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0004] This invention provides a method for optimizing and adjusting soil monitoring sites based on multi-level assessment and dynamic updating.

[0005] According to a first aspect of the present invention, a method for optimizing and adjusting soil monitoring sites based on multi-level assessment and dynamic updating is provided, comprising: Obtain regional geographic information, attribute information of soil monitoring points within the region, and geographic information of sub-regions within the region; Based on the attribute information of soil monitoring points within the region and the geographical information of sub-regions, determine whether it is necessary to add new soil monitoring points; If it is necessary to add new soil monitoring points, the attribute information of the new soil monitoring points shall be determined based on the attribute information of the soil monitoring points, the geographical information of the sub-regions within the area, and the preset soil monitoring point layout specifications. Based on historical site adjustment application information and soil monitoring site attribute information, determine whether soil monitoring sites need to be adjusted; If it is necessary to adjust the soil monitoring points, the attribute information of the adjusted soil monitoring points shall be determined based on the regional geographic information, the attribute information of the soil monitoring points in the region, and the geographic information of the sub-regions in the region. Based on the attribute information of soil monitoring points, the attribute information of newly added soil monitoring points, and the attribute information of adjusted soil monitoring points, the soil monitoring point update results are obtained.

[0006] According to the present invention, determining whether new soil monitoring points need to be added based on the attribute information of soil monitoring points within the region and the geographical information of sub-regions includes: Based on the attribute information of soil monitoring points within the region and the geographical information of sub-regions, determine whether the number of basic soil monitoring points in each sub-region is less than the first preset number. If it exists, then determine that the number of basic soil monitoring points is less than the first preset number of sub-region attributes of the first sub-region; Based on the sub-region attributes, determine whether new basic soil monitoring points are needed, and the first number of new basic soil monitoring points; Obtain historical soil monitoring data from soil monitoring sites; Based on the historical soil monitoring data, determine whether new risk soil monitoring sites are needed, and the second number of new risk soil monitoring sites.

[0007] According to the present invention, if it is necessary to add new soil monitoring points, the attribute information of the new soil monitoring points is determined based on the attribute information of the soil monitoring points, the geographical information of the sub-regions within the area, and the preset soil monitoring point layout specifications, including: The first point attribute feature information corresponding to the attribute information of the soil monitoring point is obtained through a natural language processing model. The pre-defined soil monitoring point layout specifications are processed using a natural language processing model to obtain the feature information of the layout specifications. Based on the geographic information of the sub-regions within the region, obtain the geographic image information of the first sub-region, wherein the range of the geographic image information is greater than the bounding rectangle of the graphic of the first sub-region; Based on the attribute information of the soil monitoring points and the geographical information of the first sub-region, the first location of the basic soil monitoring points in the geographic image information is determined. Based on the first location annotation and the geographic image information, the point image information is obtained; Obtain the geographic range information of the geographic image information; By using the trained basic site planning model, the attribute features of the first site, the first quantity, the geographical range information, the site image information, and the layout specification features are processed to obtain the attribute information of the newly added soil monitoring sites.

[0008] According to the present invention, by processing the attribute feature information of the first point, the first quantity, the geographical range information, the point image information, and the layout specification feature information through a trained basic point planning model, the attribute information of the newly added soil monitoring points is obtained, including: Through the first mapping level of the basic point planning model, the attribute feature information of the basic soil monitoring points within the range of point image information is mapped to obtain the first point input feature information. Through the first mapping level of the basic point planning model, standard feature information is deployed for mapping processing to obtain the first planning input feature information. The geographic range information is mapped and processed through the second mapping level of the basic point planning model to obtain the geographic range input feature information. The point image information is segmented to obtain multiple point image blocks, and each point image block is processed by a convolutional neural network model. The feature information obtained by processing each point image block is added with a position code to obtain multiple first image feature information. The first image feature information is obtained by mapping the first image feature information through the third mapping level of the basic point planning model. The geographic range input feature information is combined with the first image input feature information to obtain geographic information input feature information; The first cross-attention mechanism of the basic point planning model is used to process the first point input feature information and the geographic information input feature information to obtain the first point output feature information corresponding to the first point input feature information. By using the second cross-attention mechanism of the basic point planning model, the output feature information of the first point and the planning specification feature information are processed to obtain the point planning output feature information corresponding to the first planning input feature information. By using the self-attention mechanism of the basic site planning model, the feature information output by the site planning is processed to obtain the compliance feature information of the site. The compliance feature information of multiple locations is input into the decoder of the basic location planning model to obtain multiple location planning feature vectors. Multiple location planning feature vectors are input into the first multilayer perceptron layer of the basic location planning model to obtain the compliance score of each location planning feature vector. Based on the first quantity and the compliance score, select the target location planning feature vector from multiple location planning feature vectors; Multiple site planning feature vectors are input into the second multilayer perceptual network layer of the basic site planning model to process the target site planning feature vectors and obtain the attribute information of the newly added soil monitoring sites.

[0009] According to the present invention, the training steps of the basic point planning model include: Acquire sample point location attribute feature information, sample range information, sample point location image information and sample quantity, as well as the layout specification feature information; Using the basic point planning model, the sample point attribute feature information, sample range information, sample point image information and the layout specification feature information are processed to obtain sample point input feature information, sample planning input feature information, sample geographical range input feature information and sample image input feature information. The sample image input matrix is ​​obtained by combining the sample geographic range input feature information and the sample image input feature information; The sample image input matrix is ​​processed through the third multi-layer perceptual network layer to obtain training feature information of multiple basic points; Based on the input feature information of the sample points and the training feature information of the base points, the mapping consistency loss function is obtained; The sample images corresponding to the image blocks at the edge positions of the sample point image information are combined with the input feature information to obtain the sample range matrix; The sample range matrix is ​​processed through the fourth layer of the multi-layer perceptron to obtain training range feature information. Based on the training range feature information and the sample geographical range input feature information, the range consistency loss function is obtained; By using the basic point planning model, the input feature information of sample points, the input feature information of sample planning, the input feature information of sample geographical range, and the input feature information of sample images are processed to obtain multiple training point planning feature vectors, and the training compliance score of each training point planning feature vector is determined. By using the number of samples and the training compliance score, the target training points are selected to plan feature vectors, which are then input into the second multilayer perceptron layer to obtain the attribute information of the training soil monitoring points. Based on the attribute information of the training soil monitoring points, obtain attribute compliance labeling; Based on the attribute compliance labeling and training compliance scores, a compliance loss function is obtained; Based on the compliance loss function, the scope consistency loss function, and the mapping consistency loss function, the loss function of the basic point planning model is obtained. The basic location planning model is trained using the loss function of the basic location planning model to obtain the trained basic location planning model.

[0010] According to the present invention, determining whether soil monitoring points need to be adjusted based on historical site adjustment application information and attribute information of soil monitoring points includes: Identify the monitoring points to be adjusted in the historical monitoring point adjustment application information; Based on the attribute information of the monitoring points to be adjusted, determine the geographical location of the monitoring points to be adjusted; Determine whether the geographical location is suitable for sampling; If sampling cannot be performed, it is determined that the soil monitoring points need to be adjusted.

[0011] According to the present invention, if it is necessary to adjust the soil monitoring points, the attribute information of the adjusted soil monitoring points is determined based on regional geographical information, attribute information of soil monitoring points within the region, geographical information of sub-regions within the region, and preset soil monitoring point layout specifications, including: Based on the historical data point adjustment application information, determine the adjustment location of the monitoring points to be adjusted; If the distance between the adjusted location and the geographical location of the monitoring point to be adjusted is less than or equal to a preset distance, the adjusted location will be used as the geographical location of the adjusted soil monitoring point, and the attribute information of the adjusted soil monitoring point will be obtained.

[0012] According to the present invention, if it is necessary to adjust the soil monitoring points, the attribute information of the adjusted soil monitoring points is determined based on regional geographic information, attribute information of soil monitoring points within the region, and geographic information of sub-regions within the region, and the method further includes: If the distance between the adjustment location and the geographical location of the monitoring point to be adjusted is greater than a preset distance, then the first descriptive information of the monitoring point to be adjusted is obtained based on the regional geographical information, the attribute information of the soil monitoring points in the region, and the geographical information of the sub-regions in the region. Grid-based point selection is carried out around the monitoring point to be adjusted, and the undetermined points are obtained where the distance between the geographical location of the monitoring point to be adjusted and the geographical location of the adjustment location is less than the distance between the geographical location of the adjustment location and the geographical location of the monitoring point to be adjusted. Obtain the second descriptive information for each undetermined point; The first descriptive feature information of the first descriptive information and the second descriptive feature information of the second descriptive information are obtained through a natural language processing model. Determine the feature similarity between the first descriptive feature information and the second descriptive feature information; Based on the feature similarity, the adjusted soil monitoring points are selected from the undetermined points, and the attribute information of the adjusted soil monitoring points is obtained.

[0013] According to a second aspect of the present invention, a soil monitoring site optimization and adjustment system based on multi-level assessment and dynamic updating is provided, comprising: The acquisition module is used to acquire regional geographic information, attribute information of soil monitoring points within the region, and geographic information of sub-regions within the region. A new judgment module has been added to determine whether new soil monitoring points need to be added based on the attribute information of soil monitoring points within the region and the geographical information of sub-regions. A new module is added to determine the attribute information of newly added soil monitoring points based on the attribute information of the soil monitoring points, the geographical information of the sub-regions within the area, and the preset soil monitoring point layout specifications. The adjustment judgment module is used to determine whether soil monitoring points need to be adjusted based on historical point adjustment application information and soil monitoring point attribute information. The adjustment module is used to determine the attribute information of the adjusted soil monitoring points based on the regional geographic information, the attribute information of the soil monitoring points within the region, and the geographic information of the sub-regions within the region if the soil monitoring points need to be adjusted. The results module is used to obtain the soil monitoring point update results based on the attribute information of the soil monitoring points, the attribute information of newly added soil monitoring points, and the attribute information of the adjusted soil monitoring points.

[0014] According to a third aspect of the present invention, a soil monitoring site optimization and adjustment device based on multi-level assessment and dynamic updating is provided, comprising: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to execute the soil monitoring site optimization and adjustment method based on multi-level assessment and dynamic updating.

[0015] By adopting the above technical solution, the present invention can achieve the following technical effects: According to this invention, flexible judgment steps can be set separately for the addition and adjustment of soil monitoring sites, thereby flexibly optimizing the setting of soil monitoring sites based on actual monitoring needs, monitoring convenience, and the inherent attributes of the soil monitoring sites. This facilitates the dynamic updating and precise deployment of the soil environmental monitoring network, dynamically improving the scientific nature and accuracy of monitoring. Furthermore, a basic site planning model can be used to integrate the characteristic information of existing sites, environmental information within the geographical area, and semantic information of the deployment specifications. A neural network model can then automatically select the location that best conforms to the site deployment specifications from among multiple existing soil monitoring sites, improving the accuracy of adding new soil monitoring sites, significantly reducing labor costs, and increasing site selection efficiency. Furthermore, the accuracy of the first and second mapping layers can be improved through the mapping consistency loss function, enabling them to obtain consistent results when processing information of different types but with the same content, thereby improving mapping accuracy. The correspondence between the geographical range in the sample image and the sample range information can be improved through the range consistency loss function, reducing the offset between the geographical location in the image and the actual geographical location. Furthermore, through the compliance loss function, the model can be trained in a deep and complex manner using a simple annotation method, reducing annotation costs and improving the accuracy and compliance of the attribute information of the training soil monitoring points.

[0016] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Other features and aspects of the invention will become clearer from the following detailed description of exemplary embodiments with reference to the accompanying drawings. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained based on these drawings without creative effort. Figure 1 An exemplary flowchart of a method for optimizing and adjusting soil monitoring sites based on multi-level assessment and dynamic updating according to an embodiment of the present invention is shown. Figure 2 An exemplary block diagram of a soil monitoring site optimization and adjustment system based on multi-level assessment and dynamic updating according to an embodiment of the present invention is shown. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0020] Figure 1 An exemplary flowchart illustrates a method for optimizing and adjusting soil monitoring sites based on multi-level assessment and dynamic updating according to an embodiment of the present invention. The method includes: Step S1: Obtain regional geographic information, attribute information of soil monitoring points within the region, and geographic information of sub-regions within the region; Step S2: Based on the attribute information of the soil monitoring points in the region and the geographical information of the sub-region, determine whether it is necessary to add new soil monitoring points. Step S3: If it is necessary to add new soil monitoring points, determine the attribute information of the new soil monitoring points based on the attribute information of the soil monitoring points, the geographical information of the sub-regions within the area, and the preset soil monitoring point layout specifications. Step S4: Based on the historical site adjustment application information and the attribute information of the soil monitoring sites, determine whether it is necessary to adjust the soil monitoring sites. Step S5: If it is necessary to adjust the soil monitoring points, determine the attribute information of the adjusted soil monitoring points based on the regional geographic information, the attribute information of the soil monitoring points in the region, and the geographic information of the sub-regions in the region. Step S6: Based on the attribute information of the soil monitoring points, the attribute information of the newly added soil monitoring points, and the attribute information of the adjusted soil monitoring points, obtain the soil monitoring point update results.

[0021] The soil monitoring point optimization and adjustment method based on multi-level evaluation and dynamic updating according to embodiments of the present invention can set flexible judgment steps for the addition and adjustment of soil monitoring points, thereby flexibly optimizing the setting of soil monitoring points based on actual monitoring needs, monitoring convenience and the inherent attributes of soil monitoring points. This is conducive to the dynamic updating and precise deployment of the soil environmental monitoring network, and dynamically improves the scientificity and accuracy of monitoring.

[0022] Example 1: According to an embodiment of the present invention, in step S1, the region may include multiple sub-regions, and multiple soil monitoring points may be set up in each sub-region. The regional geographic information may include the overall geographic range, geographic boundaries, and geographic division of sub-regions. The geographic information of the sub-regions may include the geographic range and geographic boundaries of the sub-regions. The attribute information of the soil monitoring points includes the location information of the soil monitoring points, the sub-region to which they belong, the type of soil being monitored (e.g., podzolic soil, acidic sulfate soil, etc.), the type of land in which they are located (e.g., cultivated land, orchard, forest land, grassland, wetland), and the type of soil monitoring points (e.g., whether they are basic soil monitoring points or risk soil monitoring points). The present invention does not limit the specific information types contained in the attribute information.

[0023] Example 2: According to one embodiment of the present invention, the optimization of soil monitoring sites includes three types of operations: retention, addition, and adjustment. If an existing soil monitoring site already exists within a 30-meter radius of a newly added soil monitoring site, the existing soil monitoring site will be retained, and there is no need to add a new soil monitoring site. Furthermore, existing soil monitoring sites can be evaluated by experts. If the evaluation determines that the soil monitoring site is not redundant, it can be retained.

[0024] According to an embodiment of the present invention, in step S2, for the operation of adding soil monitoring points, it can be determined whether it is necessary to add soil monitoring points based on the attribute information of soil monitoring points in the region and the geographical information of sub-regions. If so, the addition can be carried out based on relevant deployment specifications.

[0025] According to one embodiment of the present invention, determining whether new soil monitoring points need to be added based on the attribute information of soil monitoring points within a region and the geographical information of sub-regions includes: determining whether the number of basic soil monitoring points in each sub-region is less than a first preset number based on the attribute information of soil monitoring points within a region and the geographical information of sub-regions; if so, determining that the number of basic soil monitoring points is less than the first preset number of sub-region attributes of a first sub-region; determining whether new basic soil monitoring points need to be added and a first number of new basic soil monitoring points based on the sub-region attributes; acquiring historical soil monitoring data of soil monitoring points; and determining whether new risk soil monitoring points need to be added and a second number of new risk soil monitoring points based on the historical soil monitoring data.

[0026] According to one embodiment of the present invention, the attribute information of soil monitoring points includes the location information of the soil monitoring points and the sub-region to which they belong. Therefore, by combining the geographical information of the sub-region, it can be determined whether the number of basic soil monitoring points in the sub-region is less than a first preset number, for example, 30. If there is a first sub-region where the number of basic soil monitoring points is less than the first preset number, the sub-region attributes of the first sub-region can be obtained, for example, the city type of the sub-region can be obtained to determine whether the first sub-region is an oasis city, an island city, or a small-area city (for example, a city with an area less than or equal to 3000 km²). If the attributes of the first sub-region do not belong to the above types, the number of basic soil monitoring points in the first sub-region is less than the first preset number, for example, it is supplemented to 30, thereby determining the first number of basic soil monitoring points that need to be added, that is, the first preset number is used to subtract the number of existing basic soil monitoring points in the first sub-region to obtain the first number. If the number of basic soil monitoring points in the sub-region is greater than or equal to the first preset number, no new ones are needed. If the first sub-region falls into one of the above categories, a decision on whether to add monitoring points can be made based on a statistical assessment of historical soil monitoring data. For example, if historical monitoring data indicates pollution in the first sub-region, the number of basic soil monitoring points can be increased, perhaps to 30. Alternatively, the need for additional basic soil monitoring points can be determined based on land type or soil type. For instance, if no basic soil monitoring points exist in an area containing a certain soil type, 1-3 additional basic soil monitoring points can be added within that area.

[0027] According to one embodiment of the present invention, it can also be determined whether new risk soil monitoring sites are needed. The need for new risk soil monitoring sites, and a second number of new risk soil monitoring sites, can be determined based on historical soil monitoring data. Based on historical soil monitoring data, a portion of the basic soil monitoring sites can be selected and upgraded to risk soil monitoring sites. For example, basic soil monitoring sites where the heavy metal content in historical soil monitoring data exceeds a threshold can be upgraded to risk soil monitoring sites. Furthermore, if both basic soil monitoring sites and risk soil monitoring sites exist around the enterprise, they are all upgraded to risk soil monitoring sites. Further, the number of the aforementioned newly added risk monitoring sites can be counted to obtain a second number.

[0028] Example 3: According to one embodiment of the present invention, in step S3, if it is necessary to add new soil monitoring points, manual selection can be performed based on preset soil monitoring point layout specifications, such as the "Technical Regulations for the Layout of Soil Environmental Monitoring Network Points." Alternatively, the points can be laid out using a trained basic point planning model based on the attribute information of the soil monitoring points, the geographical information of sub-regions within the area, and preset soil monitoring point layout specifications. On the other hand, risk soil monitoring points are usually upgraded from basic soil monitoring points; therefore, the basic point planning model is used here to lay out the first number of basic soil monitoring points.

[0029] According to one embodiment of the present invention, if it is necessary to add new soil monitoring points, the attribute information of the new soil monitoring points is determined based on the attribute information of the soil monitoring points, the geographical information of the sub-regions within the region, and the preset soil monitoring point layout specifications. This includes: obtaining first point attribute feature information corresponding to the attribute information of the soil monitoring points through a natural language processing model; processing the preset soil monitoring point layout specifications through a natural language processing model to obtain layout specification feature information; acquiring geographical image information of the first sub-region based on the geographical information of the sub-regions within the region, wherein the range of the geographical image information is larger than the bounding rectangle of the first sub-region's shape; determining a first location marker of the basic soil monitoring points in the geographical image information based on the attribute information of the soil monitoring points and the geographical information of the first sub-region; obtaining point image information based on the first location marker and the geographical image information; acquiring the geographical range information of the geographical image information; and processing the first point attribute feature information, the first quantity, the geographical range information, the point image information, and the layout specification feature information through a trained basic point planning model to obtain the attribute information of the new soil monitoring points.

[0030] According to one embodiment of the present invention, the natural language processing model can be an LSTM-based model, and the present invention is not limited thereto. Through the natural language processing model, the attribute information of soil monitoring points can be processed. For example, the descriptive text of the attribute information can be processed to obtain the attribute feature information of the first monitoring point. Similarly, the preset soil monitoring point layout specifications can be processed to obtain layout specification feature information; for example, each specification text can obtain corresponding layout specification feature information. Both the first monitoring point attribute feature information and the layout specification feature information are in vector form.

[0031] According to one embodiment of the present invention, the geographic image information of the first sub-region may include image information of an electronic map of the first sub-region. In the electronic map, a portion including the first sub-region is selected, and the image of that portion is cropped to obtain geographic image information. The range of the geographic image information is slightly larger than the range of the first sub-region; that is, the geographic image information includes the complete first sub-region and geographic information of some other adjacent regions. Further, based on the location information in the attribute information of the soil monitoring points, annotations can be made in the geographic image information to obtain a first location annotation for each soil monitoring point. That is, the location of the soil monitoring point is marked in the geographic image information. In the example, the geographic range of the geographic image information can be obtained, for example, the latitude and longitude range (i.e., geographic range information). Then, based on the location information of the soil monitoring points, the latitude and longitude of the soil monitoring points are determined. Then, based on the above latitude and longitude range, the location of the latitude and longitude of the soil monitoring points in the geographic image information is determined, thereby annotating and obtaining the first location annotation. Further, the annotated geographic image information (i.e., geographic image information with the first location annotation) can be used as the point image information.

[0032] According to an embodiment of the present invention, the attribute information of the first point location, the first quantity, the geographical range information, the point location image information, and the layout specification feature information are processed through a trained basic point location planning model to obtain attribute information of newly added soil monitoring points. This includes: mapping the first point location attribute feature information of the basic soil monitoring points within the point location image information range through a first mapping layer of the basic point location planning model to obtain first point location input feature information; mapping the layout specification feature information through a first mapping layer of the basic point location planning model to obtain first planning input feature information; mapping the geographical range information through a second mapping layer of the basic point location planning model to obtain geographical range input feature information; segmenting the point location image information to obtain multiple point location image blocks, processing each point location image block through a convolutional neural network model, and adding positional encoding to the feature information obtained from processing each point location image block to obtain multiple first image feature information; mapping the first image feature information through a third mapping layer of the basic point location planning model to obtain first image input feature information; and mapping the geographical range input feature information with the first image input feature information. The system combines input feature information to obtain geographic information input feature information; through the first cross-attention mechanism of the basic point planning model, the first point input feature information and geographic information input feature information are processed to obtain the first point output feature information corresponding to the first point input feature information; through the second cross-attention mechanism of the basic point planning model, the first point output feature information and the first planning input feature information are processed to obtain the point planning output feature information corresponding to the first planning input feature information; through the self-attention mechanism of the basic point planning model, the point planning output feature information is processed to obtain point compliance feature information; multiple point compliance feature information is input into the decoder of the basic point planning model to obtain multiple point planning feature vectors; multiple point planning feature vectors are input into the first multilayer perception network layer of the basic point planning model to obtain the compliance score of each point planning feature vector; based on the first quantity and the compliance score, a target point planning feature vector is selected from the multiple point planning feature vectors; multiple point planning feature vectors are input into the second multilayer perception network layer of the basic point planning model to process the target point planning feature vector and obtain the attribute information of the newly added soil monitoring point.

[0033] According to one embodiment of the present invention, the first mapping layer includes multiple fully connected layers, pooling layers, and activation layers, which can be used to map the attribute feature information of the first point to obtain the input feature information of the first point. The input feature information of the first point can be a higher-dimensional vector, which can more comprehensively express the features of the attribute feature information of the first point. Similarly, the first mapping layer can map the layout specification feature information to obtain the first planning input feature information.

[0034] According to one embodiment of the present invention, the second mapping layer has a similar structure to the first mapping layer, but different parameters. It can perform mapping processing on geographic range information to obtain geographic range input feature information. The geographic range information can be vector information composed of longitude and latitude ranges. For example, the two endpoints of the longitude interval and the two endpoints of the latitude interval can be combined into a vector, and this vector can be input into the second mapping layer for processing to obtain geographic range input feature information.

[0035] According to one embodiment of the present invention, the point image information can be segmented to obtain multiple point image blocks. Each point image block is then processed by a convolutional neural network, and a positional encoding is added to obtain first image feature information. The first image feature information may include feature information of the content within a local point image block. This feature information can be used to reflect features such as soil type, land type, and location name at various locations within the point image block. If the image block contains soil monitoring points, the first image feature information may also include feature information of the environment in which the soil monitoring points are located within the point image block, such as their location within the point image block, soil type, land type, and location name. Further, the first image feature information can be input into a third mapping layer for processing to obtain first image input feature information. Both the first image feature information and the first image input feature information are vector-based feature information. The first image input feature information can more comprehensively express the content features of the point image block in a higher dimension. The third mapping layer has a similar structure to the first mapping layer, but with different parameters.

[0036] According to an embodiment of the present invention, both the geographic range input feature information and the first image input feature information are features describing the geographic information of the first sub-region. That is, they can be used to represent the geographic range of the first sub-region and the specific content of the geographic information within that geographic range. After combination, geographic information input feature information can be obtained. For example, the vector-form geographic range input feature information and the first image input feature information can be combined into a matrix-form geographic information input feature information.

[0037] According to an embodiment of the present invention, the first point input feature information and the geographic information input feature information can be fused through a first cross-attention mechanism. The geographic information input feature information is fused into the first point input feature information, so that the output feature information of each first point not only carries the attribute information of the point itself, but also includes the environmental information, geographical location information, and overall environmental information and geographical range information of the area where the point is located. That is, the output feature information of the first point can represent its own features and the correlation features between itself and the surrounding environment.

[0038] A second cross-attention mechanism can be used to fuse the output feature information of the first monitoring point with the feature information of the planning specifications. That is, the feature information carried by the output feature information of the first monitoring point is integrated into the first planning input feature information. This results in the obtained point planning output feature information including not only the semantic information of each point layout specification, but also various information from existing soil monitoring points and various information about the geographical environment. Therefore, based on the point planning output feature information, locations conforming to the point layout specifications can be found in a geographical environment with multiple existing soil monitoring points to deploy new soil monitoring points—that is, to add new soil monitoring points. Furthermore, since the number of point layout specifications is fixed, the number of output point planning feature information points is also fixed, facilitating subsequent processing.

[0039] According to one embodiment of the present invention, the output feature information of the site planning may include a site layout specification, as well as various information of each soil monitoring site and various information of the geographical environment. Therefore, multiple site planning output feature information can be processed by a self-attention mechanism to integrate the information carried by each site planning output feature information to obtain site compliance feature information. The site compliance feature information carries semantic feature information of various site layout specifications, as well as various information of each soil monitoring site and various information of the geographical environment. This can be used as a constraint condition for selecting the location of a new soil monitoring site. Under this constraint condition, the optimal location (i.e., in the geographical environment of multiple existing soil monitoring sites, the location that best conforms to the site layout specification) can be selected to set up the new soil monitoring site.

[0040] According to one embodiment of the present invention, the decoder can be a transformer-based decoder. It can utilize multiple site compliance feature information and the decoder's key and weight matrices to obtain a key-value matrix and a weight matrix, respectively. It then uses the semantic vector of the start symbol and the query matrix to obtain a query vector. Furthermore, it uses the key matrix, weight matrix, and query vector to perform autoregression to generate multiple site planning feature vectors (more than the first number). Each site planning feature vector may contain the location information of a newly added soil monitoring site and the feature information of the aforementioned constraints. Therefore, the site planning feature vector is the feature vector of the location information of the newly added soil monitoring site selected by the basic site planning model that meets the constraints. Further, the compliance score of each of the multiple site planning feature vectors can be calculated through a first multilayer perceptron layer. Since the site planning feature vector contains the location information of the newly added soil monitoring site and the feature information of the aforementioned constraints, the compliance score can represent the probability that the location information of the newly added soil monitoring site meets the constraints. The higher the probability, the higher the compliance score, indicating that in a geographical environment with multiple existing soil monitoring sites, the selected location has a higher compliance with the site layout specifications. The system can select the top number of target site planning feature vectors with the highest compliance scores and input them into the second multilayer perceptron layer. This second layer processes the target site planning feature vectors to obtain the attribute information of the newly added soil monitoring sites. The second layer then converts the target site planning feature vectors into textual attribute information. Both the first and second layers of the multilayer perceptron include fully connected layers and activation layers. The newly obtained soil monitoring sites can be used as recommended sites, and after manual verification, they can be used as new basic soil monitoring sites.

[0041] In this way, the characteristic information of existing monitoring points, environmental information within the geographical area, and semantic information of the deployment specifications can be integrated through the basic site planning model. Then, the neural network model can automatically select the location that best meets the site deployment specifications in the geographical environment of multiple existing soil monitoring points based on the above information, thereby improving the accuracy of adding new soil monitoring points, significantly reducing labor costs, and improving site selection efficiency.

[0042] According to an embodiment of the present invention, before using the above-mentioned basic point planning model, the basic point planning model can be trained. The training steps of the basic point planning model include: acquiring sample point attribute feature information, sample range information, sample point image information and sample quantity, as well as the layout specification feature information; processing the sample point attribute feature information, sample range information, sample point image information and the layout specification feature information through the basic point planning model to obtain sample point input feature information, sample planning input feature information, sample geographical range input feature information and sample image input feature information; combining the sample geographical range input feature information and the sample image input feature information to obtain a sample image input matrix; processing the sample image input matrix through a third multilayer perceptron layer to obtain multiple basic point training feature information; obtaining a mapping consistency loss function based on the sample point input feature information and the basic point training feature information; and combining the sample image input feature information corresponding to the image blocks at the edge positions of the sample point image information to obtain a sample range matrix. The process involves: processing the sample range matrix through the fourth multilayer perceptron layer to obtain training range feature information; obtaining a range consistency loss function based on the training range feature information and the sample geographical range input feature information; processing the sample point input feature information, sample planning input feature information, sample geographical range input feature information, and sample image input feature information through the basic point planning model to obtain multiple training point planning feature vectors and determining the training compliance score for each training point planning feature vector; selecting the target training point planning feature vector based on the number of samples and the training compliance score, and inputting it into the second multilayer perceptron layer to obtain the attribute information of the training soil monitoring points; obtaining attribute compliance labels based on the attribute information of the training soil monitoring points; obtaining a compliance loss function based on the attribute compliance labels and the training compliance score; obtaining the loss function of the basic point planning model based on the compliance loss function, the range consistency loss function, and the mapping consistency loss function; and training the basic point planning model using the loss function of the basic point planning model to obtain the trained basic point planning model.

[0043] According to one embodiment of the present invention, the sample point attribute feature information, sample range information, sample point image information, and sample quantity, as well as the layout specification feature information, are obtained from the training samples (e.g., including sample area geographic information for training, attribute information of sample soil monitoring points, and geographic information of sub-regions in the sample area). These methods are similar to the methods for obtaining the first point attribute feature information, geographic range information, point image information, and first quantity as described above. The methods for obtaining the sample point input feature information, sample planning input feature information, sample geographic range input feature information, and sample image input feature information are similar to the methods for obtaining the first point input feature information, first planning input feature information, geographic range input feature information, and first image feature information as described above, and will not be repeated here.

[0044] According to one embodiment of the present invention, the sample geographical range input feature information and the sample image input feature information may contain feature information of various soil monitoring points within the geographical range, which can be compared with the sample point input feature information. After mapping processing by the first mapping layer and the second mapping layer, the two are theoretically consistent. Therefore, a mapping consistency loss function can be determined based on the difference between the two, so that the information of the mapped soil monitoring points is consistent. This allows the first mapping layer and the second mapping layer to obtain consistent results when processing information of different types but consistent content, thereby improving mapping accuracy. The third multilayer perceptron layer (including multiple fully connected layers and activation layers) can be used to extract feature information about soil monitoring points from the sample image input matrix to obtain basic point training feature information. The similarity (e.g., cosine similarity) between the basic point training feature information and the sample point input feature information can be used, and the difference between 1 and this similarity is used as the mapping consistency loss function. During training, the mapping consistency loss function can be reduced by gradient descent, thereby improving the accuracy of the first mapping layer and the second mapping layer.

[0045] According to one embodiment of the present invention, the image blocks at the edge positions of the sample point image information are the top row image blocks, the bottom row image blocks, the leftmost column image blocks, and the rightmost column image blocks. The sample image input feature information corresponding to these image blocks is combined to obtain a sample range matrix, which is then processed through a fourth multilayer perceptron layer (including multiple fully connected layers and activation layers) to obtain training range feature information. The similarity (e.g., cosine similarity) between the training range feature information and the sample geographic range input feature information is calculated, and the difference between 1 and the similarity is used as the range consistency loss function. During training, the range consistency loss function can be reduced by gradient descent, thereby improving the accuracy of the model in defining the geographic range, that is, improving the correspondence between the geographic range in the sample image and the sample range information, and reducing the offset between the geographic location in the image and the actual geographic location.

[0046] According to one embodiment of the present invention, the acquisition method of the training compliance score is similar to that of the aforementioned compliance score, and the acquisition method of the attribute information of the training soil monitoring points is similar to that of the aforementioned newly added soil monitoring points, and will not be repeated here. Attribute compliance labels can be determined manually, that is, by expert labeling, each training soil monitoring point is judged to be compliant; compliant points are labeled with 1, otherwise 0. This simple labeling method allows for deep and complex training of the model, reducing labeling costs and improving training quality. A cross-entropy loss function can be constructed using attribute compliance labels and training compliance scores to obtain a compliance loss function. This compliance loss function can then be reduced during training using gradient descent, improving the accuracy and compliance of the attribute information of the training soil monitoring points, and helping to more accurately identify compliant new points.

[0047] According to one embodiment of the present invention, the compliance loss function, the range consistency loss function, and the mapping consistency loss function can be weighted and summed to obtain the loss function of the basic site planning model. The basic site planning model can then be trained using the gradient descent method. After multiple training iterations, the trained basic site planning model can be obtained to determine the attribute information of newly added soil monitoring sites.

[0048] In this way, the accuracy of the first and second mapping layers can be improved through the mapping consistency loss function, so that the first and second mapping layers can obtain consistent results when processing information of different types but consistent content, thereby improving mapping accuracy. Furthermore, the correspondence between the geographical range in the sample image and the sample range information can be improved through the range consistency loss function, reducing the offset between the geographical location in the image and the actual geographical location. Moreover, through the compliance loss function, the model can be trained in a deep and complex manner using a simple annotation method, reducing annotation costs and improving the accuracy and compliance of the attribute information of the training soil monitoring points.

[0049] Example 4: According to an embodiment of the present invention, in step S4, it can be determined whether the soil monitoring points need to be adjusted based on historical point adjustment application information and attribute information of soil monitoring points. This step includes: determining the monitoring points to be adjusted involved in the historical point adjustment application information; determining the geographical location of the monitoring points to be adjusted according to the attribute information of the monitoring points to be adjusted; determining whether the geographical location can be sampled; if sampling cannot be performed, determining that the soil monitoring points need to be adjusted.

[0050] According to one embodiment of the present invention, the historical monitoring point adjustment application information may include the monitoring point to be adjusted, and the adjustment method includes adjusting its location. Based on the attribute information of the monitoring point to be adjusted, the geographical location of the monitoring point can be determined, and it can be determined whether the geographical location cannot be sampled. For example, the geographical location may be stored in a restricted area, protected area, construction land, bedrock exposed area, permanent water body, or aquaculture water body that will not change within a preset time period (e.g., within 5 years). If sampling cannot be performed, it is determined that the soil monitoring point needs to be adjusted.

[0051] Example 5: According to an embodiment of the present invention, in step S5, if it is necessary to adjust the soil monitoring points, the attribute information of the adjusted soil monitoring points is determined based on regional geographic information, attribute information of soil monitoring points within the region, geographic information of sub-regions within the region, and preset soil monitoring point layout specifications. This includes: determining the adjustment location of the monitoring point to be adjusted based on historical point adjustment application information; if the distance between the adjustment location and the geographical location of the monitoring point to be adjusted is less than or equal to a preset distance, the adjustment location is taken as the geographical location of the adjusted soil monitoring point, and the attribute information of the adjusted soil monitoring point is obtained.

[0052] According to one embodiment of the present invention, the adjustment location of the monitoring point to be adjusted can be queried in the historical point adjustment application information, that is, the planned adjusted location, and the distance between the location and the geographical location (original location) of the monitoring point to be adjusted can be calculated. If the distance is less than or equal to a preset distance (e.g., 1km), the adjustment can be made directly, that is, the adjusted location is used as the geographical location of the adjusted soil monitoring point, and the attribute information of the adjusted soil monitoring point is obtained, that is, the location information, the sub-region to which the adjusted soil monitoring point belongs, the type of soil being monitored, the type of land in the location, and the type of soil monitoring point are determined.

[0053] On the other hand, if it is necessary to adjust the soil monitoring points, the attribute information of the adjusted soil monitoring points is determined based on the regional geographic information, the attribute information of the soil monitoring points within the region, and the geographic information of the sub-regions within the region. This also includes: if the distance between the adjusted location and the geographical location of the monitoring point to be adjusted is greater than a preset distance, then, based on the regional geographic information, the attribute information of the soil monitoring points within the region, and the geographic information of the sub-regions within the region, the first descriptive information of the monitoring point to be adjusted is obtained; grid-based point selection is performed around the monitoring point to be adjusted, and pending points are obtained whose distance from the geographical location of the monitoring point to be adjusted is less than the distance between the adjusted location and the geographical location of the monitoring point to be adjusted; second descriptive information of each pending point is obtained; first descriptive feature information of the first descriptive information and second descriptive feature information of the second descriptive information are obtained through a natural language processing model; the feature similarity between the first descriptive feature information and the second descriptive feature information is determined; based on the feature similarity, the adjusted soil monitoring point is selected from the pending points, and the attribute information of the adjusted soil monitoring point is obtained.

[0054] According to one embodiment of the present invention, if the distance between the adjustment location and the geographical location of the monitoring point to be adjusted is greater than a preset distance, the regional geographical information, the attribute information of the soil monitoring points in the region, and the geographical information of the sub-regions in the region are described by a fixed language pattern to obtain the language description information of the monitoring point to be adjusted. For example, it can describe the various types of land plots in the region where it is located (e.g., the proportion of each type of land plot), the regions of various soil types in the region where it is located (e.g., the proportion of each type of region), the number and location information of soil monitoring points in each region, the number and location information of other monitoring points (including soil monitoring points belonging to the same sub-region and soil monitoring points in adjacent sub-regions) whose distance from the monitoring point to be adjusted is less than a distance threshold, and the type of soil and land monitored, etc. This language description information is the first description information.

[0055] According to one embodiment of the present invention, gridded points can be selected around the monitoring point to be adjusted, and undetermined points whose geographical distance from the monitoring point to be adjusted is less than the geographical distance between the adjustment location and the monitoring point to be adjusted can be obtained. That is, a circular area is set with the monitoring point to be adjusted as the center and the geographical distance between the adjustment location and the monitoring point to be adjusted as the radius, and a grid is set within the circular area. For example, the side length of the grid is 200m, and the grid vertices are all undetermined points. Points whose distance is less than the adjustment location, whose attribute information is closer to that of the monitoring point to be adjusted (i.e., can still be monitored in a similar environment), and whose sampling can be performed (e.g., not belonging to a restricted area) can be selected from these undetermined points as the attribute information of the adjusted soil monitoring point. In the example, the second description information of each undetermined point can be obtained in a manner similar to the first description information. The first and second description information can be processed by the natural language processing model described above to obtain the first and second description feature information. Then, the feature similarity (e.g., cosine similarity) of the first and second description feature information is determined. The undetermined point with the highest feature similarity and which can be sampled is then selected as the adjusted soil monitoring point, and its attribute information is obtained.

[0056] According to one embodiment of the present invention, in step S6, the attribute information of the soil monitoring points, the attribute information of the newly added soil monitoring points, and the attribute information of the adjusted soil monitoring points are summarized to obtain the soil monitoring point update results, and written materials are formed for subsequent manual review, such as expert review.

[0057] The soil monitoring site optimization and adjustment method based on multi-level evaluation and dynamic updating according to embodiments of the present invention can set flexible judgment steps for both the addition and adjustment of soil monitoring sites. This allows for flexible optimization of soil monitoring site settings based on actual monitoring needs, monitoring convenience, and the inherent attributes of the soil monitoring sites. This facilitates the dynamic updating and precise deployment of the soil environmental monitoring network, dynamically improving the scientific rigor and accuracy of monitoring. Furthermore, a basic site planning model can integrate the characteristic information of existing sites, environmental information within the geographical area, and semantic information of deployment specifications. A neural network model can then automatically select the location that best conforms to the site deployment specifications from among multiple existing soil monitoring sites, improving the accuracy of newly added soil monitoring sites, significantly reducing labor costs, and increasing site selection efficiency. Furthermore, the accuracy of the first and second mapping layers can be improved through the mapping consistency loss function, enabling them to obtain consistent results when processing information of different types but with the same content, thereby improving mapping accuracy. The correspondence between the geographical range in the sample image and the sample range information can be improved through the range consistency loss function, reducing the offset between the geographical location in the image and the actual geographical location. Furthermore, through the compliance loss function, the model can be trained in a deep and complex manner using a simple annotation method, reducing annotation costs and improving the accuracy and compliance of the attribute information of the training soil monitoring points.

[0058] Example 6: Figure 2 An exemplary block diagram of a soil monitoring site optimization and adjustment system based on multi-level assessment and dynamic updating according to an embodiment of the present invention is shown, the system comprising: The acquisition module is used to acquire regional geographic information, attribute information of soil monitoring points within the region, and geographic information of sub-regions within the region. A new judgment module has been added to determine whether new soil monitoring points need to be added based on the attribute information of soil monitoring points within the region and the geographical information of sub-regions. A new module is added to determine the attribute information of newly added soil monitoring points based on the attribute information of the soil monitoring points, the geographical information of the sub-regions within the area, and the preset soil monitoring point layout specifications. The adjustment judgment module is used to determine whether soil monitoring points need to be adjusted based on historical point adjustment application information and soil monitoring point attribute information. The adjustment module is used to determine the attribute information of the adjusted soil monitoring points based on the regional geographic information, the attribute information of the soil monitoring points within the region, and the geographic information of the sub-regions within the region if the soil monitoring points need to be adjusted. The results module is used to obtain the soil monitoring point update results based on the attribute information of the soil monitoring points, the attribute information of newly added soil monitoring points, and the attribute information of the adjusted soil monitoring points.

[0059] According to an embodiment of the present invention, a soil monitoring site optimization and adjustment device based on multi-level assessment and dynamic update is provided, comprising: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to execute the soil monitoring site optimization and adjustment method based on multi-level assessment and dynamic update.

[0060] According to an embodiment of the present invention, a computer-readable storage medium is provided, on which computer program instructions are stored, wherein the computer program instructions, when executed by a processor, implement the method for optimizing and adjusting soil monitoring points based on multi-level evaluation and dynamic updating.

[0061] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0062] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are merely examples and do not limit the present invention. The objectives of the present invention have been fully and effectively achieved. The functions and structural principles of the present invention have been demonstrated and explained in the embodiments, and any modifications or variations of the embodiments of the present invention may be made without departing from the stated principles.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A soil monitoring point optimization adjustment method based on multi-level evaluation and dynamic update, characterized in that, include: Obtain regional geographic information, attribute information of soil monitoring points within the region, and geographic information of sub-regions within the region; Based on the attribute information of soil monitoring points within the region and the geographical information of sub-regions, determine whether it is necessary to add new soil monitoring points; If it is necessary to add new soil monitoring points, the attribute information of the new soil monitoring points shall be determined based on the attribute information of the soil monitoring points, the geographical information of the sub-regions within the area, and the preset soil monitoring point layout specifications. Based on historical site adjustment application information and soil monitoring site attribute information, determine whether soil monitoring sites need to be adjusted; If it is necessary to adjust the soil monitoring points, the attribute information of the adjusted soil monitoring points shall be determined based on the regional geographic information, the attribute information of the soil monitoring points in the region, and the geographic information of the sub-regions in the region. Based on the attribute information of soil monitoring points, the attribute information of newly added soil monitoring points, and the attribute information of adjusted soil monitoring points, the soil monitoring point update results are obtained.

2. The method for optimizing and adjusting soil monitoring sites based on multi-level assessment and dynamic updating according to claim 1, characterized in that, Based on the attribute information of soil monitoring sites within the region and the geographical information of sub-regions, determine whether it is necessary to add new soil monitoring sites, including: Based on the attribute information of soil monitoring points within the region and the geographical information of sub-regions, determine whether the number of basic soil monitoring points in each sub-region is less than the first preset number. If it exists, then determine that the number of basic soil monitoring points is less than the first preset number of sub-region attributes of the first sub-region; Based on the sub-region attributes, determine whether new basic soil monitoring points are needed, and the first number of new basic soil monitoring points; Obtain historical soil monitoring data from soil monitoring sites; Based on the historical soil monitoring data, determine whether new risk soil monitoring sites are needed, and the second number of new risk soil monitoring sites.

3. The method for optimizing and adjusting soil monitoring sites based on multi-level evaluation and dynamic updating according to claim 2, characterized in that, If new soil monitoring sites need to be added, the attribute information of the new soil monitoring sites shall be determined based on the attribute information of the soil monitoring sites, the geographical information of the sub-regions within the area, and the preset soil monitoring site layout specifications, including: The first point attribute feature information corresponding to the attribute information of the soil monitoring point is obtained through a natural language processing model. The pre-defined soil monitoring point layout specifications are processed using a natural language processing model to obtain the feature information of the layout specifications. Based on the geographic information of the sub-regions within the region, obtain the geographic image information of the first sub-region, wherein the range of the geographic image information is greater than the bounding rectangle of the graphic of the first sub-region; Based on the attribute information of the soil monitoring points and the geographical information of the first sub-region, the first location of the basic soil monitoring points in the geographic image information is determined. Based on the first location annotation and the geographic image information, the point image information is obtained; Obtain the geographic range information of the geographic image information; By using the trained basic site planning model, the attribute features of the first site, the first quantity, the geographical range information, the site image information, and the layout specification features are processed to obtain the attribute information of the newly added soil monitoring sites.

4. The method for optimizing and adjusting soil monitoring sites based on multi-level assessment and dynamic updating according to claim 3, characterized in that, Using the trained basic site planning model, the attribute feature information of the first site, the first quantity, the geographical range information, the site image information, and the layout specification feature information are processed to obtain the attribute information of the newly added soil monitoring sites, including: Through the first mapping level of the basic point planning model, the attribute feature information of the basic soil monitoring points within the range of point image information is mapped to obtain the first point input feature information. Through the first mapping level of the basic point planning model, standard feature information is deployed for mapping processing to obtain the first planning input feature information. The geographic range information is mapped and processed through the second mapping level of the basic point planning model to obtain the geographic range input feature information. The point image information is segmented to obtain multiple point image blocks, and each point image block is processed by a convolutional neural network model. The feature information obtained by processing each point image block is added with a position code to obtain multiple first image feature information. The first image feature information is obtained by mapping the first image feature information through the third mapping level of the basic point planning model. The geographic range input feature information is combined with the first image input feature information to obtain geographic information input feature information; The first cross-attention mechanism of the basic point planning model is used to process the first point input feature information and the geographic information input feature information to obtain the first point output feature information corresponding to the first point input feature information. By using the second cross-attention mechanism of the basic point planning model, the output feature information of the first point and the planning specification feature information are processed to obtain the point planning output feature information corresponding to the first planning input feature information. By using the self-attention mechanism of the basic site planning model, the feature information output by the site planning is processed to obtain the compliance feature information of the site. The compliance feature information of multiple locations is input into the decoder of the basic location planning model to obtain multiple location planning feature vectors. Multiple location planning feature vectors are input into the first multilayer perceptron layer of the basic location planning model to obtain the compliance score of each location planning feature vector. Based on the first quantity and the compliance score, select the target location planning feature vector from multiple location planning feature vectors; Multiple site planning feature vectors are input into the second multilayer perceptual network layer of the basic site planning model to process the target site planning feature vectors and obtain the attribute information of the newly added soil monitoring sites.

5. The method for optimizing and adjusting soil monitoring sites based on multi-level evaluation and dynamic updating according to claim 4, characterized in that, The training steps for the basic point planning model include: Acquire sample point location attribute feature information, sample range information, sample point location image information and sample quantity, as well as the layout specification feature information; Using the basic point planning model, the sample point attribute feature information, sample range information, sample point image information and the layout specification feature information are processed to obtain sample point input feature information, sample planning input feature information, sample geographical range input feature information and sample image input feature information. The sample image input matrix is ​​obtained by combining the sample geographic range input feature information and the sample image input feature information; The sample image input matrix is ​​processed through the third multi-layer perceptual network layer to obtain training feature information of multiple basic points; Based on the input feature information of the sample points and the training feature information of the base points, the mapping consistency loss function is obtained; The sample images corresponding to the image blocks at the edge positions of the sample point image information are combined with the input feature information to obtain the sample range matrix; The sample range matrix is ​​processed through the fourth layer of the multi-layer perceptron to obtain training range feature information. Based on the training range feature information and the sample geographical range input feature information, the range consistency loss function is obtained; By using the basic point planning model, the input feature information of sample points, the input feature information of sample planning, the input feature information of sample geographical range, and the input feature information of sample images are processed to obtain multiple training point planning feature vectors, and the training compliance score of each training point planning feature vector is determined. By using the number of samples and the training compliance score, the target training points are selected to plan feature vectors, which are then input into the second multilayer perceptron layer to obtain the attribute information of the training soil monitoring points. Based on the attribute information of the training soil monitoring points, obtain attribute compliance labeling; Based on the attribute compliance labeling and training compliance scores, a compliance loss function is obtained; Based on the compliance loss function, the scope consistency loss function, and the mapping consistency loss function, the loss function of the basic point planning model is obtained. The basic location planning model is trained using the loss function of the basic location planning model to obtain the trained basic location planning model.

6. The method for optimizing and adjusting soil monitoring sites based on multi-level assessment and dynamic updating according to claim 1, characterized in that, Based on historical site adjustment application information and soil monitoring site attribute information, determine whether soil monitoring site adjustments are necessary, including: Identify the monitoring points to be adjusted in the historical monitoring point adjustment application information; Based on the attribute information of the monitoring points to be adjusted, determine the geographical location of the monitoring points to be adjusted; Determine whether the geographical location is suitable for sampling; If sampling cannot be performed, it is determined that the soil monitoring points need to be adjusted.

7. The method for optimizing and adjusting soil monitoring sites based on multi-level assessment and dynamic updating according to claim 6, characterized in that, If it is necessary to adjust the soil monitoring points, the attribute information of the adjusted soil monitoring points shall be determined based on the regional geographic information, the attribute information of the soil monitoring points within the region, the geographic information of the sub-regions within the region, and the preset soil monitoring point layout specifications, including: Based on the historical data point adjustment application information, determine the adjustment location of the monitoring points to be adjusted; If the distance between the adjusted location and the geographical location of the monitoring point to be adjusted is less than or equal to a preset distance, the adjusted location will be used as the geographical location of the adjusted soil monitoring point, and the attribute information of the adjusted soil monitoring point will be obtained.

8. The method for optimizing and adjusting soil monitoring sites based on multi-level assessment and dynamic updating according to claim 6, characterized in that, If it is necessary to adjust the soil monitoring sites, the attribute information of the adjusted soil monitoring sites shall be determined based on the regional geographic information, the attribute information of the soil monitoring sites within the region, and the geographic information of the sub-regions within the region. This also includes: If the distance between the adjustment location and the geographical location of the monitoring point to be adjusted is greater than a preset distance, then the first descriptive information of the monitoring point to be adjusted is obtained based on the regional geographical information, the attribute information of the soil monitoring points in the region, and the geographical information of the sub-regions in the region. Grid-based point selection is carried out around the monitoring point to be adjusted, and the undetermined points are obtained where the distance between the geographical location of the monitoring point to be adjusted and the geographical location of the adjustment location is less than the distance between the geographical location of the adjustment location and the geographical location of the monitoring point to be adjusted. Obtain the second descriptive information for each undetermined point; The first descriptive feature information of the first descriptive information and the second descriptive feature information of the second descriptive information are obtained through a natural language processing model. Determine the feature similarity between the first descriptive feature information and the second descriptive feature information; Based on the feature similarity, the adjusted soil monitoring points are selected from the undetermined points, and the attribute information of the adjusted soil monitoring points is obtained.

9. A soil monitoring site optimization and adjustment system based on multi-level assessment and dynamic updating, characterized in that, include: The acquisition module is used to acquire regional geographic information, attribute information of soil monitoring points within the region, and geographic information of sub-regions within the region. A new judgment module has been added to determine whether new soil monitoring points need to be added based on the attribute information of soil monitoring points within the region and the geographical information of sub-regions. A new module is added to determine the attribute information of newly added soil monitoring points based on the attribute information of the soil monitoring points, the geographical information of the sub-regions within the area, and the preset soil monitoring point layout specifications. The adjustment judgment module is used to determine whether soil monitoring points need to be adjusted based on historical point adjustment application information and soil monitoring point attribute information. The adjustment module is used to determine the attribute information of the adjusted soil monitoring points based on the regional geographic information, the attribute information of the soil monitoring points within the region, and the geographic information of the sub-regions within the region if the soil monitoring points need to be adjusted. The results module is used to obtain the soil monitoring point update results based on the attribute information of the soil monitoring points, the attribute information of newly added soil monitoring points, and the attribute information of the adjusted soil monitoring points.

10. A soil monitoring point optimization and adjustment device based on multi-level assessment and dynamic updating, characterized in that, include: processor; A memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to perform the method as described in any one of claims 1-8.

Images