Multi-source data comparison method for rural planning results

Abstract

This invention relates to the field of urban and rural planning informatization technology, and in particular to a method for comparing multi-source data of rural planning results. The method includes: acquiring multi-source spatial data and attribute data of the target rural area; constructing an entity graph with rural entities as nodes; and generating a planning state vector and an observation state vector for each rural entity; calculating the differences between the planning state vector and the observation state vector based on the entity graph; identifying the planning operation sequence and the actual operation sequence and establishing a correspondence to obtain the operation differences; calculating planning performance indicators and observation performance indicators based on the introduction of environmental and socio-economic external scenarios; constructing a third-order tensor corresponding to the actual operation, performance indicators, and external scenarios; analyzing the degree of influence of each actual operation on the performance indicators under different external scenarios; and outputting the comparison results of planning implementation deviations, thereby achieving quantitative evaluation of the effectiveness of rural planning implementation and providing support for rectification decisions.

Description

Technical Field

[0001] This invention relates to the field of urban and rural planning informatization technology, and in particular to a method for comparing multi-source data of rural planning results. Background Technology

[0002] Rural planning, as a crucial tool for guiding land use layout, infrastructure construction, and ecological protection, directly impacts agricultural production safety, flood control, ecological integrity, and the balanced supply of public services. In traditional practice, planning preparation and implementation supervision are often handled separately by different departments, with data sources scattered across natural resources, agriculture, water resources, housing and urban-rural development, and statistics systems. This lack of a unified data foundation and evaluation framework makes it difficult to promptly and systematically assess whether planning objectives have truly been implemented in specific spatial units. Furthermore, the planning implementation process is influenced by external factors such as climate fluctuations, economic cycles, and population flows. Relying solely on static map comparisons and simple land area statistics makes it difficult to identify the comprehensive impacts of key construction activities on agricultural production capacity, flood risk, ecological connectivity, and public service accessibility under different environmental conditions. This results in crude planning evaluations and insufficiently targeted rectification measures, hindering the formation of a closed-loop rural planning implementation and management mechanism. Summary of the Invention

[0003] To address the numerous problems existing in the prior art, this invention provides a method for comparing multi-source data of rural planning outcomes. This invention uses a rural entity map as a spatial framework, uniformly mapping multi-source spatial data and attribute data into planning state vectors and observation state vectors. Through difference calculations, it infers the planning operation sequence and the actual operation sequence. Furthermore, it calculates planning performance indicators and observation performance indicators under different external scenarios, constructing a third-order tensor of actual operations, performance indicators, and external scenarios to quantitatively characterize the impact of various actual operations on performance. This enables precise diagnosis and corrective guidance for deviations in rural planning implementation.

[0004] This specification provides one or more embodiments of a method for comparing multi-source data of rural planning results, including the following steps:

[0005] Acquire multi-source spatial data and attribute data of the target rural area, divide the target rural area into rural entities based on the multi-source spatial data and construct an entity map, and generate planning state vector and observation state vector for each rural entity according to the attribute data;

[0006] Based on the entity graph and the planned state vector and observed state vector, the difference between the planned state vector and observed state vector is calculated to determine the planned operation sequence and the actual operation sequence, establish the correspondence between the planned operation sequence and the actual operation sequence, and obtain the operation difference.

[0007] Based on entity graphs, planning state vectors, observation state vectors, operational differences, and external scenarios, planning performance indicators and observation performance indicators are calculated under each external scenario. Third-order tensors corresponding to actual operations, performance indicators, and external scenarios are constructed. The third-order tensors are analyzed to determine the degree of influence of actual operations on performance indicators under different external scenarios, and the planning implementation deviation comparison results are output.

[0008] According to one or more embodiments of the method described in this specification, the multi-source spatial data includes remote sensing image data, digital elevation data, and rural planning map data, and the attribute data includes land ownership data and rural planning text data. The target rural area is divided into spatial units based on the remote sensing image data and digital elevation data. Based on the rural planning map data and land ownership data, the spatial units are matched with the land boundaries to generate rural entities. The spatial adjacency relationships between rural entities and the connectivity relationships between rural entities and roads and waterways are recorded in the entity map.

[0009] According to the method described in one or more embodiments of this specification, when generating a planning state vector and an observation state vector for each rural entity, the planning state vector and the observation state vector include land use category information, functional use information, capacity index information and constraint information. The land use category information is used to represent one of cultivated land, forest land, water area land and construction land, and the functional use information is used to represent one or more of residential land, industrial land, public service land and ecological land.

[0010] According to the method described in one or more embodiments of this specification, when calculating the difference between the planned state vector and the observed state vector, a difference vector is constructed for each rural entity. The difference vector includes differences in land use category, functional use, capacity index, and connectivity. Based on the difference vector, the operations in the planned operation sequence and the actual operation sequence are divided into land use conversion operations, functional adjustment operations, capacity adjustment operations, and connectivity adjustment operations.

[0011] According to the method described in one or more embodiments of this specification, the multi-source spatial data and attribute data include remote sensing image time series data, land ownership change records, and construction approval records. When determining the planning operation sequence and the actual operation sequence, time sequence identifiers are assigned to the planning operations and the actual operations based on the remote sensing image time series data, land ownership change records, and construction approval records. Operations belonging to the same rural entity or spatially adjacent rural entities are sorted according to the time sequence identifiers and the spatial adjacency relationship between rural entities to form the planning operation sequence and the actual operation sequence.

[0012] According to one or more embodiments of the method described in this specification, the operational differences include a classification result for whether each planned operation in the planned operation sequence corresponds to an actual operation. The classification result includes operations executed according to the plan, operations not executed according to the plan, actual operations added beyond the plan, and actual operations opposite to the planned target direction.

[0013] According to the method described in one or more embodiments of this specification, the attribute data includes environmental time series data and socio-economic time series data. The external scenario is generated based on the environmental time series data and socio-economic time series data through scenario segmentation and scenario extraction. The external scenario includes at least an external scenario representing conventional meteorological conditions and an external scenario representing heavy rainfall conditions.

[0014] According to the method described in one or more embodiments of this specification, when calculating planning performance indicators and observation performance indicators under each external scenario, the performance indicator value of each rural entity is calculated based on the planning state vector, the observation state vector and the parameters of the external scenario. The performance indicator value of the rural entity includes agricultural production capacity indicators, flood risk indicators, ecological connectivity indicators and public service accessibility indicators. The performance indicator value of the rural entity is weighted according to the area weight or population weight of the rural entity to obtain the regional performance indicator value.

[0015] According to the method described in one or more embodiments of this specification, when constructing a third-order tensor corresponding to actual operation, performance index and external scenario, for each actual operation, an observation state without the corresponding actual operation is constructed under each performance index and each external scenario. The change value is obtained by comparing the regional performance index value under the observation state without the corresponding actual operation with the regional performance index value under the observation state with the corresponding actual operation, and the change value is written into the corresponding element in the third-order tensor.

[0016] According to the method described in one or more embodiments of this specification, when analyzing a third-order tensor, the third-order tensor is decomposed into low-rank decomposition and sparse decomposition, decomposing the third-order tensor into a low-rank part representing a general influence pattern and a sparse part representing a local abnormal influence. The influence degree of each actual operation on each performance indicator under different external scenarios is calculated based on the change values ​​corresponding to each actual operation in the low-rank part and the sparse part, and rural planning rectification suggestions are generated based on actual operations whose influence degree exceeds a preset threshold.

[0017] Compared with the prior art, the advantages and beneficial effects of the present invention are as follows:

[0018] By constructing an entity map with rural entities as nodes and unifying multi-source spatial data and attribute data into planning state vectors and observation state vectors, the planning results and actual land use and facility configuration are precisely aligned on the same spatial unit, solving the problems of data dispersion and inconsistent comparison scales in existing technologies.

[0019] By performing difference calculations on the planned state vector and observed state vector on the entity map and then deriving the planned operation sequence and the actual operation sequence, the source analysis from the difference in results to the operation behavior is realized, which solves the problem that existing technologies can only statically compare land use changes and are difficult to locate specific implementation behaviors and time sequences.

[0020] By introducing environmental and socio-economic time series to construct external scenarios, and calculating planning performance indicators and observation performance indicators under multiple external scenarios, a unified quantitative assessment of agricultural production capacity, flood risk, ecological connectivity, and public service accessibility is achieved, solving the problem that existing technologies have a single evaluation method and are difficult to reflect planning resilience and risk exposure.

[0021] By constructing a third-order tensor corresponding to actual operations, performance indicators, and external scenarios, and performing low-rank and sparse decomposition on it, the system achieves automatic identification of key positive operations and high-risk operations, as well as threshold-based rectification suggestions. This solves the problem that existing technologies struggle to extract actionable governance measures from complex, multi-source data. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the execution flow of the method of the present invention. Detailed Implementation

[0023] The embodiments of the present disclosure will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the disclosure. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of the present disclosure for ease of explanation.

[0024] Rural planning outcomes typically include overall rural land use plans, village layout plans, land use structure adjustment schemes, and supporting control indicators, phased implementation arrangements, and management rules. These outcomes are not only reflected in drawings and texts but also in the implementation of these planning intentions on specific plots of land, roads, waterways, and public service facilities. Understanding rural planning outcomes essentially means simultaneously observing the "planning target form" and the "actual evolution trajectory" within a unified spatial unit. This allows us to trace whether a construction activity or functional change originates from the planning arrangement and to examine the comprehensive support capacity of the plan for agricultural production, safety and disaster prevention, ecological patterns, and public service provision under different scenarios. To address this need, this invention constructs a systematic comparison and analysis mechanism among multi-source data related to rural planning outcomes, transforming planning outcomes from static texts and maps into dynamic implementation results that are verifiable, quantifiable, and diagnostic.

[0025] like Figure 1 As shown, a method for comparing multi-source data of rural planning results includes the following steps:

[0026] Acquire multi-source spatial data and attribute data of the target rural area, divide the target rural area into rural entities based on the multi-source spatial data and construct an entity map, and generate planning state vector and observation state vector for each rural entity according to the attribute data;

[0027] In a feasible approach, the first step is to select a county or township as the target rural area. Multi-source spatial and attribute data covering this area are then obtained from planning departments, natural resources departments, and other relevant sources. The multi-source spatial data should include at least remote sensing imagery, digital elevation data, and rural planning maps. The attribute data should include at least land ownership data and rural planning text data. All data types are then unified to the same coordinate reference system, and reprojection, registration, cropping, and quality checks are performed. The processed data is then stored in a spatial database, ensuring a one-to-one correspondence between the same location across different data sources.

[0028] Based on multi-source spatial data, the target rural area is divided into rural entities. Object-oriented image segmentation methods can be used to segment remote sensing images, obtaining initial spatial units with relatively consistent land cover features. Digital elevation data and water system data are then overlaid, and areas with steep slopes or water buffer zones are appropriately segmented or merged to ensure that the spatial units maintain basic consistency in topography and land use. Subsequently, the spatial units are overlaid and analyzed with the planning plot boundaries in the rural planning map data and the parcel boundaries in the land ownership data. Through overlay, cutting, and merging operations, each spatial unit is matched as closely as possible with the planning plots and ownership parcels, ultimately resulting in rural entities as the basic analysis units, and a unique identifier is generated for each rural entity.

[0029] After generating village entities, an entity graph is constructed in the spatial database. The entity graph uses village entities as nodes. By determining whether the polygons of village entities share common boundaries or vertices, the spatial adjacency relationship between adjacent village entities is determined, and edges are added between them. Simultaneously, using road and river centerline data, village entities intersecting the same road or river are identified, and connecting edges are added between these entities. The node attributes of the entity graph record the geometric shape, area, and center point coordinates of the village entities, while the edge attributes record the adjacency and connectivity types, providing a topological foundation for subsequent entity graph-based operations.

[0030] The organization of attribute data uses rural entities as the carrier. Planning information such as land use, building scale, and population capacity from rural planning map data is written into the planning attribute fields of the corresponding rural entities through spatial overlay. Content related to control requirements, ecological protection requirements, and infrastructure configuration requirements from rural planning text data is transformed into structured fields through clause parsing and rule matching, and then associated with the rural entities bound by those clauses. Simultaneously, the land parcel area, current land use type, existing building conditions, and rights holder information recorded in the land ownership data are written into the current status attribute fields of the corresponding rural entities. When necessary, the current status attributes can be updated based on field surveys or the latest remote sensing interpretation results.

[0031] Based on the above, a planning state vector and an observation state vector are generated for each rural entity. The planning state vector centrally represents the target state of the planning scheme for the rural entity, including fields such as planned land use category, planned functional use, planned building scale, planned population capacity, planned infrastructure level, and planned ecological control type. These fields are uniformly converted into numerical or categorical codes according to preset coding rules and combined to form the planning state vector. The observation state vector centrally represents the actual state during implementation, including fields such as current land use category, current functional use, existing building area, actual resident population, existing road and municipal facility level, and actual ecological damage or restoration status. These fields are also converted according to the same field order and coding rules as the planning state vector and combined to form the observation state vector, which is then associated with the unique identifier of the corresponding rural entity and nodes in the entity graph. Through the above processing, the target rural area is represented as a set of rural entities with topological relationships and their corresponding planning and observation state vectors, providing a complete and feasible data foundation for inferring operation sequences based on state differences and conducting performance analysis in subsequent steps.

[0032] Multi-source spatial data includes remote sensing image data, digital elevation data, and rural planning map data. Attribute data includes land ownership data and rural planning text data. Based on remote sensing image data and digital elevation data, spatial units are divided into target rural areas. Based on rural planning map data and land ownership data, spatial units are matched with land boundaries to generate rural entities. The spatial adjacency relationships between rural entities and the connectivity relationships between rural entities and roads and waterways are recorded in the entity map.

[0033] In one embodiment, a county or township is selected as the target rural area. Remote sensing imagery, digital elevation data, and rural planning map data covering the area are obtained from management departments such as natural resources and planning, along with corresponding land ownership data and rural planning text data. First, the remote sensing imagery and digital elevation data are unified in coordinates and geometrically corrected in the geographic information system software, ensuring that the remote sensing imagery, digital elevation data, rural planning map data, and land ownership data are in the same coordinate reference system, guaranteeing accurate overlay of spatial locations between different data sources.

[0034] After coordinate unification, spatial units are divided for the target rural area based on remote sensing image data and digital elevation data. Specifically, an object-oriented segmentation method can be used, based on surface reflectance characteristics and topographic elevation changes, to divide the remote sensing image data into several initial units with relatively consistent land feature and elevation characteristics. For areas such as roads, water bodies, and abrupt slope changes, boundaries are refined using digital elevation data to ensure clear topographical demarcation of spatial units. Alternatively, a regular grid division method can be used in areas with simpler terrain, employing smaller grids for areas requiring higher precision and larger grids for areas with less land feature variation, thus achieving a balance between accuracy and computational cost.

[0035] After spatial unit division, the spatial units are sequentially overlaid with the boundaries of planned land parcels in the rural planning map data. Spatial overlay operations are used to segment spatial units that cross planned land parcel boundaries, ensuring that the spatial unit boundaries match the planned land parcel boundaries as closely as possible. Subsequently, the segmented spatial units are overlaid a second time with the parcel boundaries in the land ownership data. Spatial units crossing parcel boundaries are further segmented or merged, ensuring that each spatial unit is spatially consistent with both the planned land parcel and its boundaries, or is a subdivision of those boundaries. After two rounds of overlay and segmentation, units that satisfy spatial consistency are defined as rural entities, and a unique identifier is generated for each rural entity. Key attributes such as planned land use and ownership information are written into the attribute fields of the rural entity.

[0036] Based on the polygonal geometry of rural entities, an entity graph is constructed in a spatial database. Specifically, by determining whether two rural entity polygons share a common boundary or a common vertex, their adjacency is determined, and edges representing spatial adjacency relationships are established between adjacent rural entities. Simultaneously, road centerline data and river centerline data are imported to identify sets of rural entities that spatially intersect with the same road or river. For rural entities located on continuous passages of the same road or river, edges representing road or river connectivity are added to the entity graph. Through this process, the entity graph not only reflects the spatial adjacency structure between rural entities but also clearly records the connectivity relationships between rural entities and roads and rivers, providing a fundamental topological framework and data support for subsequent accessibility analysis, flood discharge path analysis, and graph-based planning implementation comparison.

[0037] When generating planning state vectors and observation state vectors for each rural entity, the planning state vectors and observation state vectors include land use category information, functional use information, capacity index information, and constraint information. Land use category information is used to represent one of cultivated land, forest land, water area land, and construction land. Functional use information is used to represent one or more of residential land, industrial land, public service land, and ecological land.

[0038] In one embodiment, after constructing the village entities and entity maps, planning state vectors and observation state vectors are generated using the village entities as basic units. The system first establishes a record for each village entity in the database. The planning part and the observation part respectively include land use category field, functional use field, capacity index field, and constraint condition field, ensuring that the two types of state vectors have completely consistent field structure and order, which facilitates subsequent difference calculation.

[0039] Land use category information is derived from the planned land use nature and current land use type recorded in rural planning map data and land ownership data. The system pre-defines four main types: cultivated land, forest land, water area, and construction land. The land use code in the planning map is converted into one of the above main types and written into the planning state vector. The land use type obtained from the current status survey or remote sensing interpretation is converted into one of the same main types and written into the observation state vector, ensuring that the planned and observed land use categories are in the same classification system.

[0040] Functional use information is used to reflect the specific functions undertaken by rural entities. The system extracts land use requirements such as residential land, industrial land, public service land, and ecological land from planning texts and planning map appendices, and records the use items falling within the scope of the rural entity in the planning state vector in a one-to-many manner; it also extracts the distribution of actual residential buildings, production facilities, public service facilities, and ecological green spaces from current status surveys, industrial layout ledgers, and remote sensing interpretation results, and writes the corresponding uses into the observation state vector, so that a rural entity can have multiple functional use tags at the same time.

[0041] Capacity indicators are used to describe constraints at the intensity and scale levels. Capacity indicators in the planning state vector may include planned building area, planned population capacity, planned industrial land scale, and planned public service facility configuration scale, etc. The system reads values ​​from the planning map appendices and the regulatory detailed plan and associates them with the corresponding rural entities. Capacity indicators in the observation state vector are obtained by summarizing building census data, population statistics, and facility lists, such as the total built-up building area, actual resident population, built-up industrial plant area, and the service capacity of existing schools and health clinics within the rural entity.

[0042] The constraint information is used to characterize whether the rural entity is under a specific control line or management requirement. Based on thematic maps such as the boundaries of basic farmland protection zones, ecological protection red lines, flood channels, and geological hazard zones, the system writes the control zone categories intersecting with the rural entity into the planning state vector. Simultaneously, based on current monitoring records, it sets corresponding constraint markers in the observation state vector to indicate whether the rural entity illegally occupies basic farmland, damages ecological protection zones, or obstructs flood channels.

[0043] In terms of coding methods, the system uses fixed enumeration codes or multi-value tags for land use categories and functional uses, numerical fields with uniform units for capacity indicators, and Boolean fields or multi-value tags for constraints. Through these methods, the planning state vector and observed state vector of each rural entity are transformed into a clearly structured and uniformly typed set of attributes, providing a directly accessible basic data representation for subsequent comparisons of differences based on the same field, inferences about planning operations, and assessments of planning implementation deviations.

[0044] Based on the entity graph and the planned state vector and observed state vector, the difference between the planned state vector and observed state vector is calculated to determine the planned operation sequence and the actual operation sequence, establish the correspondence between the planned operation sequence and the actual operation sequence, and obtain the operation difference.

[0045] In one embodiment, after obtaining the entity map and the planning state vector and observed state vector corresponding to each village entity, the system first compares the two item by item along the attribute dimension. For four types of fields—land use category information, functional use information, capacity index information, and constraint condition information—the system generates a difference record for each village entity. The record indicates the planning value, observed value, whether there has been a change, the direction of the change, and the magnitude of the change for each field. For example, when the land use category in the planning state vector is cultivated land while the land use category in the observed state vector is construction land, this change is recorded as a category change of "cultivated land converted to construction land."

[0046] After generating the difference records, the system converts the differences into standardized operation units based on pre-configured rules. Changes in land use categories can be mapped to land conversion operations such as "new construction land," "returning farmland to forest," and "restoring water areas." Changes in functional uses can be mapped to function adjustment operations such as "adding public service functions," "introducing industrial functions," and "canceling residential functions." Changes in capacity indicators can be mapped to capacity adjustment operations such as "expanding construction scale" and "reducing population capacity." Changes in constraints can be mapped to constraint-related operations such as "breaking through basic farmland control" and "improving ecological protection measures." Through this rule mapping, each difference record is converted into one or more operation units with type identifiers and target village entity identifiers.

[0047] When determining the planning operation sequence, the system prioritizes using the difference between the planning state vector and the planning baseline state to generate planning-side operation units. The planning baseline state can be the current state at the time of planning or the approved state of the previous round of planning. Planning operation units of the same type on the same rural entity or spatially adjacent rural entities are aggregated to form spatially continuous and content-homogeneous planning operation areas. These planning operation areas are then sorted according to the implementation stages recorded in the planning text or the internally set priority order to obtain the planning operation sequence. Each element in the sequence contains at least the operation type, the set of rural entities involved, and the expected implementation stage information.

[0048] When determining the actual operation sequence, the system uses the difference between the planned state vector and the observed state vector as a basis, extracts the actual changes from the difference records, and forms actual operation units according to the same rules as the planning side. For spatially continuous changes of the same type, they are aggregated through adjacency relationships in the entity map, merging similar changes occurring in adjacent rural entities at the same time into one actual operation area. If supporting remote sensing image time series data, land ownership change records, or construction approval records are available, time sequence identifiers can be assigned to the actual operation areas, arranging them in chronological order to form the actual operation sequence.

[0049] When establishing the correspondence between planned operation sequences and actual operation sequences, the system uses operation type, the scope of rural entities involved, and spatial location as matching conditions. It matches planned operation areas with highly overlapping spatial scopes and similar operation types with actual operation areas, establishing a one-to-one or one-to-many correspondence. Successfully matched parts are marked as "planned operations with corresponding actual operations," planned operations without matching actual operations are marked as "planned operations not executed," actual operations without matching planning operations are marked as "actual operations exceeding the plan," and operations whose direction is opposite to the planning expectation are marked as "actual operations opposite to the planning goal." Through these steps, the system obtains structured operational difference results based on the entity map and state vectors, providing clear operational-level input for subsequent performance analysis and deviation diagnosis.

[0050] When calculating the difference between the planned state vector and the observed state vector, a difference vector is constructed for each rural entity. The difference vector includes differences in land use category, functional use, capacity index, and connectivity. Based on the difference vector, the operations in the planned operation sequence and the actual operation sequence are divided into land use conversion operation, function adjustment operation, capacity adjustment operation, and connectivity adjustment operation.

[0051] This embodiment, after obtaining the rural entity map and the planning and observation state vectors for each rural entity, constructs a difference vector for each rural entity using unified coding and comparison rules. The system establishes a difference record for each rural entity in the database, mapping the land use category, functional use, capacity index, and connectivity fields in the planning and observation state vectors one-to-one, and comparing them according to the order of identical fields. Land use categories and functional uses use discrete codes to represent different types; capacity indices use numerical fields to represent size; and connectivity uses Boolean or count fields to indicate whether road connectivity, river connectivity, and changes in the number of connectivity points exist.

[0052] At the implementation level, the difference vector can be abstractly represented using the following computational expression:

[0053] in, For the first The difference vector of each rural entity For the first The observed state vector of each rural entity. For the first The planning status vector of each rural entity. The difference vector is expanded into four fields in the system: land use category difference, functional use difference, capacity index difference, and connectivity difference, and written into the difference record table in a structured form.

[0054] Differences in land use categories are obtained by comparing the planned land use category code with the observed land use category code. For example, if the land is planned as arable land but observed as construction land, the difference is recorded as "conversion of arable land to construction land." Differences in function and use are determined by set comparison to identify newly added or missing use labels. For example, if public service land use is added or ecological land use is missing in the observed state, the difference record is marked as an increase or decrease in function, respectively. Differences in capacity indicators are calculated by the difference between planned building area, planned population capacity, and actual built building area and actual population, used to characterize the degree to which construction intensity or population size deviates from the planning target. Differences in connectivity are determined based on changes in road and river connectivity edges in the entity map. For example, if a new road connectivity edge appears or an existing river connectivity edge is missing in the observed state, the difference is recorded as an increase in road connectivity, a decrease in road connectivity, or a blockage of flood passage.

[0055] After constructing the difference vector, the system classifies the operations in the planned and actual operation sequences into four standardized categories based on the values ​​of each field in the difference vector. Operations with non-zero land use category differences and cross-category changes are categorized as land use conversion operations, representing fundamental changes in land use such as converting cultivated land to construction land or forest land to water area land. Operations with added or deleted use tags in functional use differences are categorized as function adjustment operations, representing changes such as adding public service functions, introducing industrial functions, or canceling residential functions in rural entities. Operations with capacity index differences exceeding a preset threshold are categorized as capacity adjustment operations, representing situations where construction scale expands or contracts, or population capacity exceeds or falls below planned values. Operations with significant changes in road or river connectivity structures in connectivity differences are categorized as connectivity adjustment operations, representing the impact of constructing new roads, closing roads, or altering flood channels on accessibility and water system connectivity between rural entities. Through the above difference vector construction and operation type classification, this invention transforms multidimensional state changes into a clearly structured and uniformly categorized set of operations, providing directly accessible basic data for subsequent operation sequence comparison and planning implementation deviation analysis.

[0056] Multi-source spatial and attribute data include remote sensing image time series data, land ownership change records, and construction approval records. When determining the planning operation sequence and the actual operation sequence, time sequence identifiers are assigned to the planning operations and the actual operations based on the remote sensing image time series data, land ownership change records, and construction approval records. Operations belonging to the same rural entity or spatially adjacent rural entities are sorted according to the time sequence identifiers and the spatial adjacency relationship between rural entities to form the planning operation sequence and the actual operation sequence.

[0057] In one embodiment, multi-source spatial data and attribute data are accompanied by complete time information upon being entered into the database. Remote sensing image time-series data records the acquisition date of each image period; land ownership change records include the registration date and effective date of the ownership transfer; and construction approval records include the project initiation date, approval date, and commencement and completion dates. The system establishes a time index table outside the spatial database, associating the aforementioned time fields with village entity identifiers to form a time-searchable evidence set.

[0058] For planning-related operations, the system extracts the planned implementation stages of various planning measures from rural planning documents and implementation plans, such as near-term, medium-term, and long-term, or specific annual arrangements, and maps this time information to the corresponding planning operations. If the planning documents do not provide a specific year, a relative time period can be set by combining the planning approval time and common implementation cycles, and stored in the planning time index table using the rural entity number and planning operation type as key values, thereby attaching a planned time sequence identifier to each planning operation.

[0059] For actual operations, the system uses time series of remote sensing images to detect changes in the status of rural entities. It compares the land use classification results and building outline extraction results of the same rural entity in two adjacent image periods. Once a change occurs from undeveloped to construction land status, the system marks the interval of the change as the initial time window of the actual construction operation. This time window is then narrowed by combining the registration time of the first change of land use or ownership type to construction use in the land ownership change record, and the approval and commencement time of the corresponding project in the construction approval record. Finally, a representative time point is selected as the time sequence identifier for the actual operation.

[0060] After obtaining the temporal sequence identifiers of planned and actual operations, the system uses village entities as the basic unit and sorts the operations on each village entity from morning to evening, forming a single-entity time series. Then, utilizing the spatial adjacency relationships in the entity graph, it aggregates spatially consecutive operations within the same time period, treating similar operations occurring on adjacent village entities as a spatially continuous operation region. This sorting and aggregation process is performed on both the planning and actual sides, resulting in a planned operation sequence and an actual operation sequence. Both sequences preserve the temporal sequence and reflect the spatial expansion process on the village entity graph, providing a clear serialized input for subsequent comparisons of planning execution and operational differences based on both time and space dimensions.

[0061] Operational differences include the classification results of whether each planned operation in the planned operation sequence has a corresponding actual operation. The classification results include operations executed according to the plan, operations not executed according to the plan, actual operations added beyond the plan, and actual operations opposite to the planned objective.

[0062] In one embodiment, after the planned operation sequence and the actual operation sequence have been formed according to the time sequence and the spatial relationships of rural entities, the system reserves an operation difference field in the database for each planned operation to record the matching status between the planned operation and the actual operation. Each planned operation includes at least the operation type, the set of rural entities involved, and the planned implementation stage, and each actual operation includes at least the operation type, the set of rural entities involved, and the time interval of occurrence. The correspondence between the two is given by the aforementioned matching process.

[0063] The system first performs a preliminary screening based on operation type and spatial scope. For each planning operation, it searches the actual operation sequence for operations with the same or equivalent operation type, involving a high degree of overlap between the rural entity set and the spatial scope of the planning operation, and requires that the actual operation occurs within a preset time window near the planned implementation stage of the planning operation. Actual operations that meet the above conditions are marked as candidate corresponding operations. If a planning operation has at least one candidate corresponding operation, the system further checks whether the direction of change of the candidate corresponding operation in key fields such as land use category, functional use, capacity index, and constraints is consistent with the planning objective. If they are consistent, the planning operation is classified as an operation executed according to the plan, and the identifier of the corresponding actual operation is recorded in the operation difference field.

[0064] For planned operations that do not have any corresponding candidate operations in the planned operation sequence, or whose key field changes significantly deviate from the planning objectives and do not meet the execution conditions according to the threshold, the system classifies them as operations not executed according to the plan. Operations not executed according to the plan will be recorded as locations where the plan is not fully implemented or is lagging behind, providing a direct basis for the subsequent formulation of rectification plans.

[0065] For actual operations in the actual operation sequence that are not referenced as candidate corresponding operations by any planning operation, the system identifies them as actual operations added beyond the planning scope. These operations typically include situations where construction land was actually added without being planned in the planning map, or where industrial facilities were introduced into areas where no industrial function was defined in the planning. The system adds a record for this type of actual operation in the operation difference table, marks it as an actual operation added beyond the planning scope, and associates it with the set of affected rural entities.

[0066] When the planned objective direction is opposite to the actual change direction, the system identifies this by comparing the changing directions of the planned state vector and the observed state vector. For example, if the plan specifies that a rural entity should be maintained as arable land or restored as ecological land, but the observed state vector shows that the rural entity has actually been converted to construction land, or if the plan requires a reduction in construction intensity while the actual capacity index has significantly increased, the system will mark the corresponding planning operation as an actual operation opposite to the planned objective direction. Through the above discrimination rules, this invention subdivides operational differences into four categories: operations executed according to the plan, operations not executed according to the plan, actual operations added beyond the plan, and actual operations opposite to the planned objective direction. This provides clear and directly usable classification results for subsequent assessment of planning implementation deviations and identification of high-risk operations under different external scenarios.

[0067] Based on entity graphs, planning state vectors, observation state vectors, operational differences, and external scenarios, planning performance indicators and observation performance indicators are calculated under each external scenario. Third-order tensors corresponding to actual operations, performance indicators, and external scenarios are constructed. The third-order tensors are analyzed to determine the degree of influence of actual operations on performance indicators under different external scenarios, and the planning implementation deviation comparison results are output.

[0068] After constructing the rural entity map, generating planning and observation state vectors, and identifying operational differences, the system introduces external scenarios to quantitatively evaluate the planning implementation effect. External scenarios can be pre-defined based on environmental and socio-economic time-series data, such as scenarios with normal weather conditions, heavy rainfall, and population growth. Each external scenario provides a set of scenario parameters to drive the evaluation model. Using rural entities as the basic unit, and combining the spatial adjacency and connectivity relationships recorded in the entity map, the system calculates performance indicators under both the planning and observation states for each external scenario.

[0069] The performance indicator system can include multiple indicators such as agricultural production capacity, flood risk, ecological connectivity, and public service accessibility. For the planning state, the system uses the planning state vector as input, substituting land use categories, functional uses, capacity indicators, and constraints into common models in fields such as agricultural output estimation models, runoff and waterlogging models, ecological corridor analysis models, and transportation accessibility models. Combined with external scenario parameters, it obtains village-level entity-level performance indicator values, which are then aggregated into regional planning performance indicators according to area weight or population weight. For the observation state, the system uses the same model structure and external scenario parameters, using the observation state vector as input, to obtain regional observation performance indicators, thus ensuring the comparability of planning performance indicators and observation performance indicators in terms of models and scenarios.

[0070] To characterize the impact of each actual operation under different external scenarios and performance indicators, the system constructs virtual revocation scenarios based on the observed states. For each actual operation, the system first identifies the set of rural entities involved and the specific changes to land use categories, functional uses, capacity indicators, and connectivity relationships. Then, it copies the observed state vector to the database, restoring the changed parts related to the actual operation to the planned state or the neutral state before the operation was implemented, forming the observed state for the virtual revocation of the actual operation. The system repeats the above performance calculation process for the observed states after virtual revocation under each external scenario to obtain the regional performance indicators under the virtual revocation state.

[0071] Based on this, the system constructs a three-dimensional array corresponding to actual operations, performance indicators, and external scenarios, denoted as a third-order tensor, and uses the following calculation expression to define the meaning and relationship of each element:

[0072] in, For the first The first practical operation in The first performance indicator and the first Impact value under each external scenario For the observation state, the first The performance indicator in the first Regional performance indicator values ​​under various external scenarios To virtually cancel the first in the observation state After the first actual operation The performance indicator in the first The regional performance indicator value under an external scenario. A positive impact value indicates that the actual operation improved the corresponding performance indicator compared to the reversal state, while a negative impact value indicates that the actual operation weakened the corresponding performance indicator.

[0073] After establishing the third-order tensor, the system can perform statistical analysis on the impact of each actual operation across all external scenarios and all performance indicators. For example, it can average, find extreme values, or calculate the frequency of occurrence of the same performance indicator under different external scenarios. This helps identify key operations that continuously improve performance in multiple scenarios and high-risk operations that cause significant negative impacts in some scenarios. Simultaneously, comparing the impact values ​​of different performance indicators under the same external scenario allows the system to determine whether certain operations improve one type of performance indicator at the expense of others. Finally, based on the difference curve between planned and observed performance indicators and the impact characteristics of each actual operation in the third-order tensor, the system generates comparison results of planning implementation deviations and a list of key operations, providing quantitative evidence for planning evaluation and corrective action decisions.

[0074] The attribute data includes environmental time series data and socio-economic time series data. The external scenario is generated based on the environmental time series data and socio-economic time series data through scenario segmentation and scenario extraction. The external scenario includes at least an external scenario representing normal meteorological conditions and an external scenario representing heavy rainfall conditions.

[0075] In one embodiment, the attribute data includes not only land ownership data and planning text data, but also environmental and socio-economic time-series data covering the target rural area. The environmental time-series data can be sourced from meteorological, hydrological, and other monitoring systems, and at least includes daily or monthly precipitation, temperature, and river water levels. The socio-economic time-series data can be sourced from statistical yearbooks or local databases, and at least includes indicators that change over time, such as population size, industrial output, and urban and rural construction investment. When the data is entered into the database, the system associates the aforementioned time-series data with the administrative unit or watershed unit where the rural entity is located, enabling each rural entity to query its corresponding environmental and socio-economic status for a given year or time period.

[0076] In the environmental time series data preprocessing stage, the system first fills in and corrects missing and outlier values. For example, it uses interpolation of nearby time periods for short-term missing data and manually verifies or removes obviously abnormal extreme values ​​in conjunction with the original monitoring records. Subsequently, based on multi-year historical data, it calculates the multi-year average and distribution characteristics of indicators such as precipitation and temperature for each year or season. The actual observed values ​​for each year or time period are compared with the multi-year average to determine the relative position of that time period in terms of meteorological conditions. Socioeconomic time series data are organized by year, with standardized statistical standards, eliminating the impact of changes in statistical standards due to administrative division adjustments, and eliminating price factors by using fixed prices or indices to ensure the comparability of data from different years.

[0077] After preprocessing, the system performs scenario segmentation based on environmental and socio-economic time-series data. For precipitation in the environmental time-series data, multi-year precipitation distribution is used as a reference. Years or periods with precipitation close to the multi-year average are classified as candidate sets for conventional meteorological conditions, while years or periods with precipitation significantly higher than the multi-year average and reaching a preset high precipitation threshold are classified as candidate sets for heavy rainfall conditions. For socio-economic time-series data, indicators such as population growth rate and construction investment intensity can be used to distinguish between years of relatively stable economic development and years of rapid expansion, which can be used to construct external scenarios under different development intensities when needed.

[0078] During the scenario extraction phase, the system selects several representative years or time periods from the aforementioned candidate set as external scenario samples. For external scenarios with normal meteorological conditions, years with precipitation and temperature close to the multi-year average and socioeconomic indicators at a moderate level can be prioritized as representatives; for external scenarios with heavy rainfall conditions, years with regional rainstorms or continuous heavy rainfall events and reliable monitoring records can be selected as representatives. The system generates a set of scenario parameters for each external scenario, including summary indicators such as the average precipitation of the corresponding year or time period, the duration of extreme precipitation events, the highest river level, population size, and industrial structure, and stores these scenario parameters along with the external scenario identifier in the scenario database.

[0079] Through the above processing, this invention transforms long-term environmental and socio-economic time-series data into a limited number of clearly defined external scenarios, each of which can be directly invoked in subsequent performance calculations. At least two external scenarios are included: one for routine meteorological conditions and one for heavy rainfall conditions. These are used to characterize the implementation effectiveness of the plan under normal conditions and its safety and resilience under extreme rainfall pressure, respectively. This ensures that the comparison of rural planning results not only focuses on static differences but also enables targeted analysis of implementation deviations under different external constraints.

[0080] When calculating planning performance indicators and observation performance indicators under each external scenario, the performance indicator values ​​of each rural entity are calculated based on the planning state vector, observation state vector and parameters of the external scenario. The performance indicator values ​​of rural entities include agricultural production capacity indicators, flood risk indicators, ecological connectivity indicators and public service accessibility indicators. The performance indicator values ​​of rural entities are weighted according to the area weight or population weight of the rural entities to obtain the regional performance indicator values.

[0081] In one embodiment, after selecting any external scenario from the external scenario library, the system uses the planned state vector and the observed state vector as inputs to calculate the performance index value of each rural entity under that external scenario. For the agricultural production capacity index, the system uses an agricultural yield estimation model based on the rural entity's land use category, crop type, arable land quality, irrigation conditions, and meteorological parameters in the external scenario to obtain the theoretical yield under the planned state and the actual achievable yield under the observed state, thus characterizing the strength of agricultural production capacity. For the flood risk index, the system combines the rural entity's elevation location on the entity map, its connectivity with the river, and rainfall and river water level parameters in the external scenario to perform a simplified runoff and water accumulation analysis, calculating the probability and potential loss of each rural entity under that scenario, thus forming the flood risk index value.

[0082] The ecological connectivity index is based on the spatial adjacency relationships and ecological land distribution in the entity map. Rural entities marked as ecological land are considered ecological nodes. The analysis examines whether ecological patches remain continuous through woodlands, water bodies, or other ecological corridors under a given external scenario. This allows for the calculation of the connectivity degree of each rural entity within the ecological network, reflecting the level of ecological connectivity. The public service accessibility index, on the other hand, utilizes road connectivity relationships and the spatial location of public service facilities, combined with population distribution and traffic condition parameters in the external scenario, to estimate the travel time or travel impedance from rural entities to the nearest school, health service point, or comprehensive service center, converting this into a public service accessibility index value.

[0083] The above four types of rural entity performance indicators are calculated separately under planning and observation conditions, but the same model structure and parameter settings are used in the same external scenario to ensure comparability. To obtain regional-scale performance evaluation results, the system weights and summarizes the rural entity performance indicator values ​​according to area weight or population weight, which can be expressed by the following formula:

[0084] in, In the first Under the first external scenario Regional performance indicator values ​​for each performance indicator. For the first The rural entities in the first The first performance indicator and the first Performance indicator values ​​for rural entities under various external scenarios For the first The weights of each rural entity are normalized by area or population. A weighted calculation is performed separately for the planning state and the observed state to obtain the performance index values ​​for the planning area and the observed area under the same external scenario. By comparing the differences, the implementation deviations of rural planning in agricultural production, safety and disaster prevention, ecological connectivity, and public service provision under different external scenarios such as normal weather conditions or heavy rainfall can be intuitively identified. This provides preliminary data for subsequent analysis of the impact of key operations on performance using third-order tensors.

[0085] When constructing the third-order tensor corresponding to actual operations, performance indicators, and external scenarios, for each actual operation, an observation state without the corresponding actual operation is constructed under each performance indicator and each external scenario. The change value is obtained by comparing the regional performance indicator value under the observation state without the corresponding actual operation with the regional performance indicator value under the observation state with the corresponding actual operation, and the change value is written into the corresponding element in the third-order tensor.

[0086] In one embodiment, after obtaining the list of actual operations, regional performance index values ​​of various performance indicators, and external scenarios, the system constructs a third-order tensor corresponding to the actual operations, performance indicators, and external scenarios, with the actual operations as the main thread. The system establishes a record for each actual operation, which includes the actual operation identifier, operation type, set of affected rural entities, and time interval of the operation. It also associates the aforementioned operation difference classification results to ensure that it can accurately identify which rural entities and which field-level changes the actual operation caused in the observed state.

[0087] For any given operation, an observation state "excluding the operation" is constructed for each performance indicator and each external scenario. Specifically, based on the already constructed set of observation state vectors, a copy of the observation state is generated in memory. Relevant fields of rural entities within the scope of the operation's influence are reverted according to a preset method: for fields such as land use category and functional purpose that have changed due to the operation, their values ​​are restored to the corresponding values ​​in the planning state vector; for fields such as capacity indicators formed by multiple operations, time series records can be used to trace back to the value in the period preceding the operation; for connectivity fields, road or river connectivity edges newly added by the operation are removed according to the entity graph structure. The observation states obtained after the above reversion process, excluding the corresponding operation, only change for rural entities affected by the operation; the remaining rural entities retain their original observation states.

[0088] After obtaining the observational state without corresponding actual operations, the system invokes the aforementioned performance calculation process under each external scenario. Using the same evaluation model and scenario parameters, it recalculates the regional-scale agricultural productivity indicators, flood risk indicators, ecological connectivity indicators, and public service accessibility indicators to obtain the corresponding regional performance indicator values. These regional performance indicator values ​​are then compared with those calculated under the same external scenario in the original observational state. The difference between the two values ​​represents the change in performance indicators and external scenario for that actual operation. A positive change indicates that the actual operation improved the corresponding performance compared to the unimplemented scenario, while a negative change indicates that the actual operation had adverse effects under that condition.

[0089] The system, following a pre-defined indexing order, sequentially writes the changes in each actual operation under each performance indicator and each external scenario into the corresponding element positions of a third-order tensor. This ensures that the three dimensions of the third-order tensor correspond to the actual operation, performance indicator, and external scenario, respectively. Through this construction method, each element in the third-order tensor has a clear physical meaning: "the marginal impact of a certain actual operation on a certain performance indicator under a certain external scenario." This provides a well-structured and directly accessible data foundation for subsequent identification of key operations, cross-scenario comparative analysis, and multi-dimensional diagnosis of planning implementation deviations.

[0090] When analyzing the third-order tensor, low-rank decomposition and sparse decomposition are performed on the third-order tensor. The third-order tensor is decomposed into a low-rank part representing the general influence pattern and a sparse part representing the local abnormal influence. Based on the change values ​​corresponding to each actual operation in the low-rank part and sparse part, the influence degree of each actual operation on each performance indicator under different external scenarios is calculated. Based on the actual operations whose influence degree exceeds the preset threshold, suggestions for rural planning rectification are generated.

[0091] In one embodiment, after constructing a third-order tensor with actual operation, performance indicators, and external scenarios as three dimensions in the aforementioned steps, the system performs low-rank decomposition and sparse decomposition on this third-order tensor in the data analysis module to characterize the impact of actual operation on performance from both the overall pattern and local anomaly perspectives. The elements of the third-order tensor are defined as the impact value of a certain actual operation under a certain performance indicator and a certain external scenario. Before entering the decomposition, the system can standardize different performance indicator dimensions to ensure that the values ​​of each performance indicator are within a comparable range, avoiding the dominance of a single indicator with an excessively large dimension in the decomposition results.

[0092] In this embodiment, a low-rank plus sparse decomposition is adopted, representing the third-order tensor as the sum of the low-rank and sparse parts, which can be characterized by the following formula:

[0093] in, For the first The first practical operation in The first performance indicator and the first Impact value under each external scenario For the first The first practical operation in The first performance indicator and the first The low-rank impact values ​​under various external scenarios represent relatively stable impact patterns that are common across multiple external scenarios. For the first The first practical operation in The first performance indicator and the first The sparse component impact value under each external scenario represents the amplified or anomalous impact on a specific external scenario or performance indicator. In practical implementation, existing low-rank plus sparse tensor decomposition algorithm libraries can be used to solve the problem through iterative optimization. and Furthermore, rank constraints and sparsity constraints are set during the solution process, and those skilled in the art can adjust the parameters based on the data scale and computing resources.

[0094] After obtaining the low-rank and sparse components, the system analyzes actual operations, calculating the degree of impact from two perspectives: "general impact" and "local anomaly." For the low-rank component, the system statistically summarizes the low-rank impact values ​​of each actual operation across various performance indicators and all external scenarios. For example, it takes the average or the maximum absolute value of the low-rank impact values ​​under different external scenarios to characterize the overall impact strength of the actual operation on a certain performance indicator under normal conditions. For the sparse component, the system focuses on the peak value of the sparse component impact value under extreme external scenarios to identify whether the actual operation will significantly amplify flood risk or weaken ecological connectivity under stress scenarios such as heavy rainfall.

[0095] By combining the results from the low-rank and sparse components, the system calculates an impact index for each actual operation and each performance indicator. This impact index is obtained by weighting the stable contribution of the low-rank component with the abnormal contribution of the sparse component according to preset weights, and distinguishes between positive improvement and negative deterioration effects based on the sign of the impact value. For actual operations with an impact exceeding a preset threshold, the system generates rural planning rectification suggestions based on the type of the actual operation, the scope of the rural entities involved, and the type of performance indicator affected. For example, when a certain type of new construction land operation consistently shows a decline in agricultural productivity in the low-rank component and a significant increase in flood risk under heavy rainfall scenarios in the sparse component, the system can propose suggestions such as reducing the expansion of this type of land use, strengthening drainage and water storage facilities, and adjusting land use layout. When a certain type of ecological restoration operation shows improved ecological connectivity in multiple scenarios, it can be marked as a positive operation to be prioritized for promotion. Through the above analysis process, this invention extracts complex information on multiple operations, indicators, and scenarios into actionable corrective measures for planning implementation management, which facilitates planning management departments to optimize rural spatial layout and construction strategies in a targeted manner during subsequent adjustments and new rounds of planning.

[0096] In practical applications, taking the eastern part of a certain county as an example, the county administers three administrative villages, namely A, B, and C, which together constitute approximately 300 hectares. 2The target rural area was divided into four rural entities, E1 to E4, by combining 0.5m resolution remote sensing imagery, 5m resolution digital elevation model, approved rural planning maps, and the latest land ownership data. The boundaries of the planned land parcels were overlaid with the boundaries of the land parcels to ensure that each rural entity has a clear land use scope and corresponding ownership.

[0097] E1 has an area of ​​approximately 100 hectares. 2 The area is planned as arable land, with 70% being high-standard farmland. The average altitude is 120m, and it is 1.2km away from the main river. The planned population is 600 people, and no new construction land is allowed.

[0098] E2 has an area of ​​approximately 80 hectares. 2 Planning 50 hm 2 For arable land, 30 hm 2 To concentrate construction land, the new village will be divided into three clusters, with a planned residential population of 1,500 and a total building area controlled at 80,000 square meters. 2 ;

[0099] E3 covers an area of ​​approximately 60 hectares. 2 40 hm 2 For the protection of forest land, 20 hm 2 It is designated for ecological use such as wetlands, with an average elevation of 115m. 40% of it is located on the floodplain, serving as an ecological corridor between the upstream and downstream areas, with a planned population of 200.

[0100] E4 has an area of ​​approximately 60 hectares. 2 Planning 30 hm 2 Land for the construction of township service centers (schools, health centers, comprehensive service stations, etc.), 30 hectares 2 The area is designated for ecological use, including parks and water bodies, and is planned to serve a population of 3,000.

[0101] The information such as land use category, functional use, population capacity, building capacity, and land use constraints is encoded into a planning state vector. Each rural entity forms a planning state vector with a dimension of approximately 20, which includes indicators such as cultivated land area, forest land area, construction land area, ecological land area, planned residential population, planned number of public service beds, whether construction is prohibited, and whether it is within the ecological red line.

[0102] Combining remote sensing interpretation results from 2020 to 2024, land ownership change records, and field surveys, an observation state vector was constructed. The 2024 observation results deviated significantly from the planning:

[0103] Approximately 20 hm in E1 2 High-standard arable land has been occupied by scattered construction, resulting in approximately 150 self-built rural houses, increasing the construction land area from 0 to 20 hectares. 2The cultivated land area has decreased to 80 hectares. 2 The residential population has increased from the planned 600 to approximately 900.

[0104] E2 was actually implemented for only about 15 hectares. 2 The new village construction land will form a 40,000 square meter area. 2 Housing, still 15 hm 2 The planned construction land will continue to be used as arable land, with the actual arable land area being 65 hectares. 2 The permanent population is approximately 1,300.

[0105] Approximately 10 hm in E3 2 Protective forest land was converted into arable land, reducing the forest area to 30 hectares. 2 10 hectares of new arable land 2 , of which 6\hm 2 Located within the floodplain area;

[0106] E4 has basically completed the construction of its service center according to plan, with an actual construction land area of ​​approximately 32 hectares. 2 Public service facilities have a building area of ​​50,000 square meters. 2 It can serve a population of approximately 3,500, with ecological land reduced slightly to 28 hectares. 2 .

[0107] Subtracting the planned state vector from the observed state vector component by component yields the difference vector for each rural entity: E1 represents "farmland reduced by 20 hm²". 2 Construction land increased by 20 hectares 2 "Population increased by 300 people, violation of the prohibition on construction"; E2 is "less than 15 hectares of construction land implemented". 2 15 hectares of arable land should be retained. 2 "Building capacity reduced by 40,000 m²"; E3 is "Forest land reduced by 10 hm²". 2 10 hm² of arable land 2 "The ecological space of the river floodplain has been compressed"; E4 is "the service capacity slightly exceeds the plan, and the ecological land has been reduced by 2 hm²". 2 "Based on the ownership change time from 2018 to 2024, the E1 construction spread mainly occurred from 2021 to 2023, the E3 forest land reclamation was concentrated from 2020 to 2022, the E2 new village construction progressed slowly, and the E4 service center was completed in 2022. Based on this, corresponding planning operations and actual operations were generated for each type of difference, forming an operation sequence ordered by year."

[0108] During the performance evaluation phase, two external scenarios are introduced: Scenario S1 represents the 30-year average meteorological conditions with an annual precipitation of approximately 900 mm; Scenario S2 represents a year of heavy rainfall with an annual precipitation of approximately 1300 mm and an increase of 50% in the number of extreme rainfall days. Under each scenario, four types of performance indicators are calculated based on the planning state and the observation state: agricultural production capacity, flood risk, ecological connectivity, and public service accessibility.

[0109] In terms of agricultural production capacity, it is assumed that the yield of high-standard arable land under S1 is approximately 7 t / hm. 2 The yield of ordinary cultivated land is about 6 tons per hectare. 2 The amount of water used in S2 was reduced to 6 t / hm. 2 and 5t / hm 2 Under the planned conditions, the combined arable land area of ​​E1 and E2 is 150 hectares. 2 Of which, 90 hectares are high-standard arable land. 2 The total output of region S1 is approximately 90×7+60×6=630+360=990t.

[0110] Plans E3 and E4 both have no arable land. Under S2, the calculated yield is approximately 840 tons. Under the observed conditions, E1 has 80 hectares of arable land. 2 (The proportion of high-standard arable land has been reduced to 60%), E2 arable land 65 hm² 2 E3 arable land 10 hm 2 Total 155 hm 2 Of which, approximately 83 hectares are high-standard arable land. 2 The total output under S1 is approximately 83×7+72×6=581+432=1013t.

[0111] Under S2 conditions, the output is approximately 83×6 + 72×5 = 498 + 360 = 858t. It can be seen that the short-term output under the observed conditions is about 20t higher than the planned output.

[0112] The flood risk index is represented by a value between 0 and 1. A standardized index of the probability of flooding and potential loss is calculated for each entity, taking into account elevation, slope, distance from the river channel, and land cover. Under S1, the individual risk indices in the planning state are approximately E1: 0.20, E2: 0.25, E3: 0.15, and E4: 0.28. Under the observation state, the risk index of E1 rises to 0.45 due to the hardening of low-lying farmland into construction land, and the risk index of E3 rises from 0.15 to 0.30 due to the reclamation of floodplain forest land. E2 and E4 show relatively small changes, approximately 0.26 and 0.30 respectively. Under S2, the overall risk increases. The risk indices of the four entities under the planning state are approximately 0.35, 0.40, 0.30, and 0.42, and under the observation state, they are approximately 0.60, 0.45, 0.50, and 0.48 respectively. Using area as the weight for regional weighting, the regional risk index under planning status S2 is approximately (100×0.35+80×0.40+60×0.30+60×0.42) / 300≈0.37.

[0113] Under the observed conditions, the value is approximately (100×0.60+80×0.45+60×0.50+60×0.48) / 300≈0.50. The overall regional flood risk increases by approximately 0.13.

[0114] The ecological connectivity index combines factors such as ecological land area, number of patches, and ecological path length on the entity map. Under the planning state, E3 and E4 are connected by approximately 50 hectares of ecological pathways. 2 The continuous ecological land forms a main corridor with a connectivity index of 0.80. Under the observed conditions, the ecological land decreased to approximately 45 hectares after reclamation at E3. 2 The main corridor was cut into two sections, reducing the connectivity index to 0.60. Regarding public service accessibility, using a 30-minute travel time as a threshold, under the planned conditions, 85% of the population could reach the E4 service center within 30 minutes, resulting in an accessibility index of approximately 0.85. Under the observed conditions, due to some new construction concentrated in areas close to E4, the accessibility index increased to approximately 0.90.

[0115] At the operational level, four types of actual operations were selected: O1 represents the "new 20\hm" that occurred in E1 between 2021 and 2023. 2 Construction land (occupying high-standard cultivated land), O2 is E2 "only 15 hm" in 2020–2024 2 The "inadequate implementation of planning" in the "new village construction" project; O3 refers to E3's "reclamation of 10 hectares" from 2020 to 2022. 2 "Protective forest land is arable land", and O4 is the service center built by E4 in 2022 according to the plan. For each operation, an observation state of "cancel the operation" is constructed under each scenario, the regional performance indicators are recalculated, and compared with the original observation state to obtain the change value.

[0116] For example, under S2, cancel O1, and set E1 to 20\hm 2 Construction land was restored to arable land, maintaining the same population and building capacity. Assessing flood risk changes solely from a land use perspective, the regional flood risk index decreased from 0.50 to approximately 0.44, a change of approximately -0.06; the agricultural output index increased slightly by about 5 tons, a change recorded as +5 tons. O3 was revoked, and 10 hm² of E3 was allocated to... 2 Farmland was restored to protective forest land. The flood risk index in the S2 area decreased from 0.50 to approximately 0.46, a change of approximately -0.04, while the ecological connectivity index increased from 0.60 to approximately 0.72, a change of approximately +0.12. O2 was withdrawn, assuming 30 hectares were completed as planned. 2 With the construction of new villages and an equivalent reduction in E2 arable land, the agricultural output index under S1 will decrease from 1013t to approximately 990t, a change of approximately -23t, but the public service accessibility index will increase from 0.90 to approximately 0.93. For O4, assuming no service center is built, the public service accessibility index will decrease from 0.90 to approximately 0.65, a change of approximately -0.25, while the agricultural and flood indicators will change less.

[0117] The aforementioned changes are written into a third-order tensor in the order of "actual operation - performance indicators - external scenarios". For example, O1 has an element of -0.06 in "flood risk - S2" and +5 in "agricultural production capacity - S2"; O3 has an element of +0.12 in "ecological connectivity - S2" and -0.04 in "flood risk - S2"; O4 has an element of +0.20 in "public service accessibility - S1" and +0.25 in "public service accessibility - S2". After performing low-rank and sparse decomposition on this third-order tensor, it can be seen that "occupying low-lying farmland for construction" and "reclaiming floodplain forest land" like O1 and O3 generally show an increase in flood risk and a decrease in ecological connectivity under multiple scenarios, while "concentrated public service construction" like O4 generally shows a significant improvement in public service accessibility. Based on the magnitude and distribution of the changes, O1 and O3 can be marked as high-risk operations requiring priority rectification, while O4 can be marked as a positive operation type that can be promoted. Furthermore, specific adjustment suggestions are given, such as land return and ecological restoration for E1 and E3, and accelerating the implementation of new village construction for E2. With such data support, the method of this invention demonstrates a complete chain within a specific area, from identifying deviations between planning and reality, to analyzing the impact at the operational level, and finally to generating spatial rectification suggestions.

[0118] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for comparing multi-source data of rural planning results, characterized in that, Includes the following steps: Acquire multi-source spatial data and attribute data of the target rural area, divide the target rural area into rural entities based on the multi-source spatial data and construct an entity map, and generate planning state vector and observation state vector for each rural entity according to the attribute data; The attribute data includes environmental time series data and socio-economic time series data. The external scenario is generated based on the environmental time series data and socio-economic time series data through scenario segmentation and scenario extraction. The external scenario includes at least an external scenario representing normal meteorological conditions and an external scenario representing heavy rainfall conditions. Based on the entity graph and the planned state vector and observed state vector, the difference between the planned state vector and observed state vector is calculated to determine the planned operation sequence and the actual operation sequence, establish the correspondence between the planned operation sequence and the actual operation sequence, and obtain the operation difference. When calculating the difference between the planned state vector and the observed state vector, a difference vector is constructed for each rural entity. The difference vector includes differences in land use category, functional use, capacity index, and connectivity. Based on the difference vector, the operations in the planned operation sequence and the actual operation sequence are divided into land use conversion operation, function adjustment operation, capacity adjustment operation, and connectivity adjustment operation. Based on entity graphs, planning state vectors, observation state vectors, operational differences, and external scenarios, planning performance indicators and observation performance indicators are calculated under each external scenario. Third-order tensors corresponding to actual operations, performance indicators, and external scenarios are constructed. The third-order tensors are analyzed to determine the degree of influence of actual operations on performance indicators under different external scenarios, and the planning implementation deviation comparison results are output. When calculating planning performance indicators and observation performance indicators under each external scenario, the performance indicator values ​​of each rural entity are calculated based on the planning state vector, observation state vector and parameters of the external scenario. The performance indicator values ​​of rural entities include agricultural production capacity indicators, flood risk indicators, ecological connectivity indicators and public service accessibility indicators. The performance indicator values ​​of rural entities are weighted according to the area weight or population weight of the rural entities to obtain the regional performance indicator values. When analyzing the third-order tensor, low-rank decomposition and sparse decomposition are performed on the third-order tensor. The third-order tensor is decomposed into a low-rank part representing the general influence pattern and a sparse part representing the local abnormal influence. Based on the change values ​​corresponding to each actual operation in the low-rank part and sparse part, the influence degree of each actual operation on each performance indicator under different external scenarios is calculated. Based on the actual operations whose influence degree exceeds the preset threshold, suggestions for rural planning rectification are generated.

2. The method according to claim 1, characterized in that, Multi-source spatial data includes remote sensing image data, digital elevation data, and rural planning map data. Attribute data includes land ownership data and rural planning text data. Based on remote sensing image data and digital elevation data, spatial units are divided into target rural areas. Based on rural planning map data and land ownership data, spatial units are matched with land boundaries to generate rural entities. The spatial adjacency relationships between rural entities and the connectivity relationships between rural entities and roads and waterways are recorded in the entity map.

3. The method according to claim 1, characterized in that, When generating planning state vectors and observation state vectors for each rural entity, the planning state vectors and observation state vectors include land use category information, functional use information, capacity index information, and constraint information. Land use category information is used to represent one of cultivated land, forest land, water area land, and construction land. Functional use information is used to represent one or more of residential land, industrial land, public service land, and ecological land.

4. The method according to claim 1, characterized in that, Multi-source spatial and attribute data include remote sensing image time series data, land ownership change records, and construction approval records. When determining the planning operation sequence and the actual operation sequence, time sequence identifiers are assigned to the planning operations and the actual operations based on the remote sensing image time series data, land ownership change records, and construction approval records. Operations belonging to the same rural entity or spatially adjacent rural entities are sorted according to the time sequence identifiers and the spatial adjacency relationship between rural entities to form the planning operation sequence and the actual operation sequence.

5. The method according to claim 1, characterized in that, Operational differences include the classification results of whether each planned operation in the planned operation sequence has a corresponding actual operation. The classification results include operations executed according to the plan, operations not executed according to the plan, actual operations added beyond the plan, and actual operations opposite to the planned objective.

6. The method according to claim 1, characterized in that, When constructing the third-order tensor corresponding to actual operations, performance indicators, and external scenarios, for each actual operation, an observation state without the corresponding actual operation is constructed under each performance indicator and each external scenario. The change value is obtained by comparing the regional performance indicator value under the observation state without the corresponding actual operation with the regional performance indicator value under the observation state with the corresponding actual operation, and the change value is written into the corresponding element in the third-order tensor.

Images