Land reserve project value assessment and risk control system

Abstract

The application discloses a land reserve project value evaluation and risk management and control system and relates to the technical field of land reserve management. The system comprises a multi-source data preprocessing module, a land value dynamic evaluation module, a risk transmission reasoning module and a closed-loop management and control execution module. The multi-source data preprocessing module performs standardized processing on multi-source heterogeneous business data and spatial data and outputs a standardized land feature data set. The land value dynamic evaluation module calculates the value through an adaptive feature weight model and dynamically corrects the value in combination with risk parameters. The risk transmission reasoning module completes risk transmission reasoning based on a Bayesian network, realizes two-way linkage of value and risk and iteratively optimizes model parameters based on disposal results to form a closed-loop optimization mechanism. The application can realize collaborative management and control of value and risk and improve the fine management level of land reserve projects.

Description

Technical Field

[0001] This invention relates to the field of land reserve management technology, and in particular to a land reserve project value assessment and risk management system. Background Technology

[0002] With the continuous development of the digital governance system for national land space, land reserve, as a core link in natural resource asset management and urban development, is gradually transforming its business management from the traditional manual ledger model to a digital and information-based model. Various land reserve business management systems have been widely applied to process control and data collection throughout the entire project lifecycle, playing a positive role in improving management efficiency.

[0003] In existing land reserve management systems, value assessment and risk management are typically deployed as two independent functional modules. Each is calculated independently based on its own business indicator system. The value calculation process does not incorporate risk influencing factors, and the risk assessment process does not integrate with the project's value dimension for joint analysis. This fragmented operational model fails to reflect the intrinsic relationship between risk and value, making it difficult to achieve coordinated management of land reserve project value and risk, and thus unable to provide comprehensive and accurate support for project decision-making. Summary of the Invention

[0004] The purpose of this invention is to provide a land reserve project valuation and risk management system to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a land reserve project value assessment and risk management system, comprising a multi-source data preprocessing module, a land value dynamic assessment module, a risk transmission reasoning module, and a closed-loop management execution module; The multi-source data preprocessing module is used to collect multi-source heterogeneous business data and spatial data of the target land reserve project, perform standardization processing on the multi-source heterogeneous business data and spatial data, and output a standardized land parcel feature dataset. The dynamic land value assessment module is used to receive the standardized land parcel feature dataset, calculate the initial land value assessment result through an adaptive feature weight model, receive risk correction parameters to correct the initial land value assessment result, and output the dynamic land value assessment result. The risk transmission reasoning module is equipped with a risk transmission reasoning model, which is built based on a Bayesian network. It is used to receive the standardized land parcel feature dataset and the dynamic land value assessment results, establish a directed topology based on the causal relationship between risk factors and perform forward probability propagation to calculate the comprehensive risk level of the project. At the same time, it generates the risk correction parameters and feeds them back to the dynamic land value assessment module. The closed-loop management and control execution module is used to receive the dynamic land value assessment results and the project's comprehensive risk level, generate corresponding management and control instructions, match disposal plans, and track the execution progress. After disposal, it generates model correction parameters based on the deviation between the predicted results and the actual execution results, and feeds the model correction parameters back to the dynamic land value assessment module and the risk transmission reasoning module respectively, to complete the iterative optimization of the model parameters and form a closed-loop optimization mechanism for value assessment and risk management.

[0006] Preferably, the multi-source data preprocessing module includes a data acquisition unit, a spatiotemporal reference alignment unit, and a semantic mapping unit; The data acquisition unit is used to collect multi-source heterogeneous business data and spatial data corresponding to the target land reserve project; The spatiotemporal reference alignment unit is used to unify all spatial data to the same spatial reference and elevation reference, and establish spatial topological relationships between land parcels. The semantic mapping unit is used to perform semantic standardization transformation on multi-source heterogeneous business data and generate the standardized land parcel feature dataset by combining the spatial topological relationship of the land parcels.

[0007] Preferably, the land value dynamic assessment module includes a feature weight calculation unit, a policy semantic parsing unit, and a value correction unit; The feature weight calculation unit is used to receive the standardized land parcel feature dataset, learn the contribution weight of each value feature indicator through the adaptive attention mechanism built into the adaptive feature weight model, and calculate the initial land value assessment result. The policy semantic parsing unit is used to extract the constraints and change information in the policy documents, generate value adjustment parameters and input them into the feature weight calculation unit to dynamically update the feature weights. The value correction unit is used to receive the risk correction parameters, perform risk discount correction on the initial land value assessment result, and output the dynamic land value assessment result.

[0008] Preferably, the risk transmission reasoning module includes a risk factor management unit, a topology construction unit, and a probability reasoning calculation unit; The risk factor management unit is used to store the hierarchically set underlying risk factors and corresponding judgment thresholds, and to extract corresponding data from the standardized land parcel feature dataset to generate real-time risk values ​​for each underlying risk factor. The topology construction unit is used to establish a directed topology based on the causal relationship between each risk factor and to set the basic inference parameters of the risk transmission inference model. The probabilistic reasoning calculation unit is used to take the real-time risk values ​​of each underlying risk factor and the dynamic land value assessment results as input nodes, perform forward probability propagation, output the comprehensive risk level of the project, and generate the risk correction parameters.

[0009] Preferably, the closed-loop control execution module includes a graded early warning unit, a response plan matching unit, and a parameter iteration optimization unit; The graded early warning unit is used to trigger an early warning signal of the corresponding level and generate the control instructions based on the fluctuation range of the dynamic land value assessment results and the comprehensive risk level of the project. The disposal plan matching unit is used to match the disposal plan according to the risk type and warning level, push it to the terminal node with corresponding permissions, and track the execution progress. The parameter iteration optimization unit is used to receive the actual execution result after the processing is completed, compare it with the prediction result to calculate the deviation, generate the model correction parameters and feed them back to the corresponding module.

[0010] Preferably, the probabilistic reasoning calculation unit is also used to perform reverse diagnostic reasoning, calculate the risk contribution ranking of each underlying risk factor to the overall risk level of the project, locate key risk causes, and generate a risk transmission path map; The risk transmission reasoning module also includes a risk coupling degree calculation unit, which is used to quantify the coupling effect between different risk dimensions. When the coupling degree of two or more types of risks exceeds a preset threshold, the overall risk level of the project is increased.

[0011] Preferably, the closed-loop control execution module further includes a disposal cancellation verification unit, which is used to re-collect real-time data of the corresponding underlying risk factor for secondary verification after disposal is completed. If the verification passes, the cancellation of the corresponding risk status is completed; if the verification fails, the control instruction is regenerated. The parameter iterative optimization unit uses the gradient descent algorithm to generate model correction parameters, which respectively correct the feature weights of the land value dynamic assessment module and the conditional probability table of the risk transmission inference module.

[0012] Preferably, the topology construction unit determines the directed topology structure by jointly using business prior causal logic and historical risk case data; wherein the business prior causal logic defines the direct correlation between underlying risk factors, and the historical risk case data corrects the conditional probability parameters of the topology connections through a structure learning algorithm.

[0013] Preferably, the system further includes a data storage module, which stores the standardized land parcel feature dataset, the dynamic land value assessment results, the project's comprehensive risk level, underlying risk factors and corresponding judgment thresholds, the model parameters, the disposal plan and disposal execution records, and provides data read and write support for each module.

[0014] Preferably, it also includes a spatiotemporal visualization module, which uses a spatiotemporal map as a base to overlay and display land value data, risk distribution data, and business process data, and supports map-related linked queries and interactive operations.

[0015] Compared with the prior art, the beneficial effects of the present invention are: The multi-source data preprocessing module standardizes the heterogeneous business and spatial data of the target land reserve project, outputting a standardized land parcel feature dataset to achieve unified adaptation of heterogeneous data and provide a consistent and reliable data foundation for subsequent calculations. The dynamic land value assessment module calculates the initial land value assessment result based on an adaptive feature weight model, and outputs the dynamic assessment result after correction by combining risk correction parameters, improving the objectivity and dynamic adaptability of land value assessment. The risk transmission inference module, configured with a Bayesian network risk transmission inference model, completes risk transmission inference and comprehensive risk level calculation, and generates risk correction parameters to feed back to the value assessment module, realizing two-way linkage between value and risk. The closed-loop control execution module generates control instructions and tracks the disposal progress, and iteratively optimizes the model parameters based on the disposal results, forming a closed-loop optimization mechanism for value assessment and risk control, improving the refinement and intelligence level of land reserve project management. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of a land reserve project valuation and risk management system provided in an embodiment of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 This invention provides a land reserve project value assessment and risk management system, including a multi-source data preprocessing module, a land value dynamic assessment module, a risk transmission reasoning module, and a closed-loop management execution module; The multi-source data preprocessing module is used to collect multi-source heterogeneous business data and spatial data of the target land reserve project, perform standardization processing on the multi-source heterogeneous business data and spatial data, and output a standardized land parcel feature dataset. The dynamic land value assessment module is used to receive the standardized land parcel feature dataset, calculate the initial land value assessment result through an adaptive feature weight model, receive risk correction parameters to correct the initial land value assessment result, and output the dynamic land value assessment result. The risk transmission reasoning module is equipped with a risk transmission reasoning model, which is built based on a Bayesian network. It is used to receive the standardized land parcel feature dataset and the dynamic land value assessment results, establish a directed topology based on the causal relationship between risk factors and perform forward probability propagation to calculate the comprehensive risk level of the project. At the same time, it generates the risk correction parameters and feeds them back to the dynamic land value assessment module. The closed-loop management and control execution module is used to receive the dynamic land value assessment results and the project's comprehensive risk level, generate corresponding management and control instructions, match disposal plans, and track the execution progress. After disposal, it generates model correction parameters based on the deviation between the predicted results and the actual execution results, and feeds the model correction parameters back to the dynamic land value assessment module and the risk transmission reasoning module respectively, to complete the iterative optimization of the model parameters and form a closed-loop optimization mechanism for value assessment and risk management.

[0019] Specifically, the system executes five stages in sequence according to the data flow: data preprocessing, initial value calculation, risk transmission reasoning, dynamic value correction, and closed-loop management execution. Among them, the land value dynamic assessment module and the risk transmission reasoning module form a two-way data interaction. The land value assessment result is used as an input variable for risk reasoning in the calculation of financial risks. The risk correction parameters output by risk reasoning are input back into the value assessment module to complete the risk discount correction, realizing the linkage update of value assessment and risk management.

[0020] Furthermore, the model correction parameters output by the closed-loop management and control execution module are simultaneously applied to the land value dynamic assessment module and the risk transmission reasoning module. Based on the actual disposal results, the model parameters of the two modules are synchronously iterated and optimized, so that the accuracy of value assessment and risk reasoning are improved simultaneously, forming a self-optimizing complete closed loop.

[0021] In one optional embodiment, the multi-source data preprocessing module includes a data acquisition unit, a spatiotemporal reference alignment unit, and a semantic mapping unit; The data acquisition unit is used to collect multi-source heterogeneous business data and spatial data corresponding to the target land reserve project; The spatiotemporal reference alignment unit is used to unify all spatial data to the same spatial reference and elevation reference, and establish spatial topological relationships between land parcels. The semantic mapping unit is used to perform semantic standardization transformation on multi-source heterogeneous business data and generate the standardized land parcel feature dataset by combining the spatial topological relationship of the land parcels.

[0022] Specifically, the three units execute processing sequentially according to the data flow order, realizing the transformation of raw heterogeneous data into a standardized dataset, and providing a unified format of data input for subsequent value assessment and risk reasoning.

[0023] In this embodiment, the spatiotemporal benchmark alignment uses a coordinate transformation algorithm to uniformly map spatial vector data from different sources and coordinate systems to the same spatial and elevation benchmarks. Then, a spatial overlay analysis algorithm is used to identify the topological relationships between plots, such as adjacency, inclusion, and overlap, and to establish spatial topological associations between plots. The semantic mapping unit uses a pre-built standardized semantic dictionary to normalize and map data fields with different names and meanings in different business systems, eliminating differences in data fields between departments. Finally, the standardized attribute data and spatial data are bound based on the unique identifier of the plot to generate a complete standardized plot feature dataset.

[0024] Optionally, the data acquisition unit supports interfaces with various official data sources such as government data sharing and exchange platforms, land and space basic information platforms, and real estate registration systems. It also supports two data acquisition modes: offline batch import and real-time interface calls, adapting to the government data opening conditions in different regions.

[0025] In an optional embodiment, the land value dynamic assessment module includes a feature weight calculation unit, a policy semantic parsing unit, and a value correction unit; The feature weight calculation unit is used to receive the standardized land parcel feature dataset, learn the contribution weight of each value feature indicator through the adaptive attention mechanism built into the adaptive feature weight model, and calculate the initial land value assessment result. The policy semantic parsing unit is used to extract the constraints and change information in the policy documents, generate value adjustment parameters and input them into the feature weight calculation unit to dynamically update the feature weights. The value correction unit is used to receive the risk correction parameters, perform risk discount correction on the initial land value assessment result, and output the dynamic land value assessment result.

[0026] Specifically, the three units sequentially complete the basic value calculation, policy dynamic correction, and risk discount correction, forming a three-layer value calculation link to realize the dynamic updating of land value as the market, policies, and risks change.

[0027] Furthermore, the adaptive attention mechanism is a feature weight allocation technique in the field of deep learning. In this embodiment, it is embedded in the adaptive feature weight model. The input is multi-dimensional value features such as location, planning indicators, supporting resources, and market transactions in the standardized land feature dataset. The mechanism dynamically assigns weight coefficients to each dimension of features. The weight coefficients are positively correlated with the degree of influence of the features on land value. Finally, the initial land value assessment result is obtained by multi-feature weighted summation calculation. The policy semantic parsing unit uses a fine-tuned natural language processing pre-trained model to perform entity extraction and semantic matching on official documents such as land planning, industrial policies, and regulatory requirements. It identifies key information that has a positive or negative impact on land value and converts it into quantitative value adjustment parameters.

[0028] Optionally, the risk discount coefficient of the value correction unit can be configured differently according to the project business stage. A higher discount coefficient is configured when the risk uncertainty is high in the early stage of land acquisition and storage, and a lower discount coefficient is configured when the risk gradually becomes clear in the early stage of supply, so as to adapt to the value assessment needs of different stages of the entire land reserve cycle.

[0029] In an optional embodiment, the risk transmission reasoning module includes a risk factor management unit, a topology construction unit, and a probability reasoning calculation unit; The risk factor management unit is used to store the hierarchically set underlying risk factors and corresponding judgment thresholds, and to extract corresponding data from the standardized land parcel feature dataset to generate real-time risk values ​​for each underlying risk factor. The topology construction unit is used to establish a directed topology based on the causal relationship between each risk factor and to set the basic inference parameters of the risk transmission inference model. The probabilistic reasoning calculation unit is used to take the real-time risk values ​​of each underlying risk factor and the dynamic land value assessment results as input nodes, perform forward probability propagation, output the comprehensive risk level of the project, and generate the risk correction parameters.

[0030] Specifically, the three units sequentially complete the quantification of risk factors, the establishment of a transmission network, and the comprehensive risk simulation, realizing the quantitative calculation from scattered risk factors to the comprehensive risk level of the project.

[0031] In this embodiment, the Bayesian network is a graphical model based on probabilistic reasoning. It represents the causal dependencies between variables in the form of a directed acyclic graph. Each node in the network corresponds to a risk factor, and the directed edges between nodes correspond to the causal transmission path of the risk. Each node is configured with a corresponding conditional probability table. The real-time risk values ​​of each bottom-level risk factor are normalized and converted into the probability of occurrence of risk in the range of 0-1, which is used as the input node values ​​of the Bayesian network. Forward probability propagation is the core reasoning method of the Bayesian network. That is, starting from the real-time probability values ​​of the bottom-level risk factor nodes, the probability of occurrence of the upper-level nodes is calculated layer by layer along the edge direction of the directed topology, and finally the probability of occurrence of the comprehensive project risk is derived. It is then mapped to the corresponding comprehensive project risk level through a preset probability grading threshold.

[0032] It should be noted that the underlying risk factors stored in the risk factor management unit are classified and categorized according to business stage and risk category, forming a hierarchical factor system. Each factor is configured with a clear numerical classification judgment threshold. The unit automatically extracts the values ​​of the corresponding fields from the standardized land feature dataset, matches them with the judgment threshold, and converts them into standardized real-time risk values, completing the mapping from raw business data to risk quantification values, and providing a unified format input for subsequent probabilistic inference.

[0033] In one optional embodiment, the closed-loop control execution module includes a graded early warning unit, a response plan matching unit, and a parameter iterative optimization unit; The graded early warning unit is used to trigger an early warning signal of the corresponding level and generate the control instructions based on the fluctuation range of the dynamic land value assessment results and the comprehensive risk level of the project. The disposal plan matching unit is used to match the disposal plan according to the risk type and warning level, push it to the terminal node with corresponding permissions, and track the execution progress. The parameter iteration optimization unit is used to receive the actual execution result after the processing is completed, compare it with the prediction result to calculate the deviation, generate the model correction parameters and feed them back to the corresponding module.

[0034] Specifically, the three units are responsible for early warning judgment, process execution, and feedback optimization, respectively, forming the execution layer of the control closed loop and opening up the entire process link from risk identification to model optimization.

[0035] Furthermore, the tiered early warning unit adopts a dual-dimensional judgment rule that combines risk level with value fluctuation. Based on the overall risk level of the project and the magnitude of land value fluctuation, it triggers different levels of early warning signals and generates corresponding control instructions with different control intensities. The disposal plan matching unit has a built-in workflow engine that matches corresponding disposal plans from the contingency plan library according to the risk type and early warning level, automatically generates disposal work orders, and dispatches them to the system terminal nodes with corresponding permissions. It tracks the processing nodes, processing time, and processing status of the work orders throughout the entire process, achieving full-link traceability of the disposal process.

[0036] Optionally, the parameter iteration optimization unit supports two operating modes: full iteration and incremental iteration. Full iteration retrains the model parameters based on all historical disposal cases accumulated by the system, while incremental iteration updates the corresponding parameters only based on newly added single disposal results. This can be flexibly selected according to the amount of data and computing power, taking into account both model accuracy and computational efficiency.

[0037] In an optional embodiment, the probabilistic reasoning calculation unit is further configured to perform reverse diagnostic reasoning, calculate the risk contribution ranking of each underlying risk factor to the overall risk level of the project, locate key risk triggers, and generate a risk transmission path map; The risk transmission reasoning module also includes a risk coupling degree calculation unit, which is used to quantify the coupling effect between different risk dimensions. When the coupling degree of two or more types of risks exceeds a preset threshold, the overall risk level of the project is increased.

[0038] Specifically, reverse diagnostic reasoning is used to locate the root causes of risks, and risk coupling degree calculation is used to quantify the superimposed amplification effect of multiple types of risks. Together, they improve the comprehensiveness and accuracy of risk assessment.

[0039] In this embodiment, the reverse diagnostic reasoning is based on the posterior probability calculation capability of Bayesian networks. Given the overall risk level of the project, the posterior probability of each underlying risk factor is derived in reverse. This is used to calculate the contribution ratio of each factor to the overall risk. The key risk causes are obtained by sorting them from high to low contribution. A risk transmission path map with directional transmission arrows is automatically drawn, and the contribution value of each node is marked to help managers quickly locate the source of risk.

[0040] It should be noted that the risk coupling degree calculation adopts the coupling coordination degree model. The comprehensive risk index of four types of risks—compliance risk, natural risk, financial risk, and market risk—is used as the input variable. The coupling degree calculation formula is used to calculate the coupling degree value between two or more types of risks. The value ranges from 0 to 1. The closer the value is to 1, the stronger the superposition and amplification effect between risks. The preset threshold is usually set to 0.7. If the value exceeds the threshold, it is determined that there is a significant coupling effect. At this time, the system automatically increases the comprehensive risk level of the project to ensure the rigor of risk assessment.

[0041] In an optional embodiment, the closed-loop control execution module further includes a disposal cancellation verification unit. The disposal cancellation verification unit is used to re-collect real-time data of the corresponding underlying risk factor for secondary verification after disposal is completed. If the verification passes, the cancellation of the corresponding risk status is completed. If the verification fails, the control instruction is regenerated. The parameter iterative optimization unit uses the gradient descent algorithm to generate model correction parameters, which respectively correct the feature weights of the land value dynamic assessment module and the conditional probability table of the risk transmission inference module.

[0042] Specifically, the handling and verification unit is used to ensure the substantive effectiveness of risk handling, while the parameter iteration and optimization unit is used to realize the automatic correction of model parameters. Together, they improve the final stage of the control loop.

[0043] Furthermore, once the disposal work order is marked as completed, the disposal cancellation verification unit automatically triggers the data re-collection process for the corresponding risk factor, and performs a second verification on the updated data according to the original risk judgment threshold. If the verification result shows that the risk has dropped below the safety threshold, the cancellation process for the corresponding risk status is officially completed. If the verification fails, it is determined that the disposal has not achieved the expected effect, and the control instructions are automatically regenerated and a new round of disposal process is started to ensure that the risk is substantially resolved.

[0044] In this embodiment, the gradient descent algorithm is a classic parameter optimization algorithm in the field of machine learning. In this embodiment, for the land value dynamic assessment module, a loss function is constructed based on the deviation between the value prediction value and the actual transaction value, and the weight coefficients of each value feature are updated iteratively through gradient descent. For the risk transmission inference model, a loss function is constructed based on the deviation between the risk prediction result and the actual risk occurrence, and the conditional probability table parameters corresponding to each node in the Bayesian network are updated iteratively, so that the model output gradually approaches the true result, continuously improving the prediction accuracy and adaptability of the model.

[0045] In an optional embodiment, the topology construction unit determines the directed topology by jointly using business prior causal logic and historical risk case data; wherein the business prior causal logic defines the direct correlation between underlying risk factors, and the historical risk case data corrects the conditional probability parameters of the topology connections through a structure learning algorithm.

[0046] Specifically, the topology structure adopts a joint construction approach that combines expert prior knowledge with data-driven methods, taking into account both the rationality of business logic and the objectivity of actual data, while supporting model cold start and continuous iterative optimization.

[0047] In this embodiment, the prior causal logic of the business is defined by senior business experts in the land reserve field based on actual business rules. It follows the hierarchical transmission rules of underlying factors, intermediate risk events, and comprehensive risks, clarifying the direct causal relationship and transmission logic between each underlying risk factor. For example, the ownership clarity factor points to the risk of expropriation progress, the risk of expropriation progress points to the risk of cost overrun, and the risk of cost overrun ultimately points to the comprehensive risk of the project. This forms the initial skeleton of the directed topology structure, ensuring that the network structure conforms to the real business logic. Historical risk case data is used to optimize and correct the initial topology through structural learning algorithms. Specifically, classic Bayesian network structure learning algorithms such as the K2 algorithm and hill climbing algorithm can be used. Using the sample data of historical risk cases as input, the connection relationship and conditional probability parameters of the nodes in the initial topology are iteratively optimized, so that the fit between the topology structure and the actual risk occurrence pattern is gradually improved.

[0048] It should be noted that this joint construction method enables the cold start operation of the model. In the early stages of project launch, when there is a lack of historical risk cases, basic risk reasoning can be completed solely based on the prior causal logic of the business, ensuring the rapid launch and operation of the system. As historical cases are continuously accumulated during the project operation, the system can automatically iterate and optimize the topology and probability parameters, gradually improving the inference accuracy of the model.

[0049] In an optional embodiment, a data storage module is further included. The data storage module is used to store the standardized land parcel feature dataset, the dynamic land value assessment results, the project comprehensive risk level, the underlying risk factors and corresponding judgment thresholds, the model parameters, the disposal plan and disposal execution records, and to provide data read and write support for each module.

[0050] Specifically, the data storage module provides underlying data support for the entire system, uniformly storing various types of business data, model parameters, and process records, and serves as the carrier for data exchange between modules.

[0051] In this embodiment, the data storage module adopts a hybrid storage architecture of relational database and spatial database. Structured business data, model parameters, and disposal process records are stored in the relational database, while land parcel spatial vector data and spatial base map data are stored in the spatial database. The module provides a unified data access interface to the outside world, providing standardized data read and write services to the upper-level business modules, and realizing centralized management and efficient access to data.

[0052] Optionally, the data storage module is configured with a timed data backup and data desensitization mechanism, which regularly performs off-site backup of core business data and desensitizes government data involving sensitive information to ensure the security and compliance of system data.

[0053] In an optional embodiment, it also includes a spatiotemporal visualization module, which uses a spatiotemporal map as a base to overlay and display land value data, risk distribution data, and business process data, and supports map-related linked queries and interactive operations.

[0054] Specifically, the spatiotemporal visualization module serves as the human-computer interaction carrier of the system, spatializing and visualizing abstract value data, risk data, and business process data to enhance the intuitiveness of the data.

[0055] Furthermore, the map-attribute linkage query specifically implements a two-way mutual query function: when a user selects a target plot on the spatiotemporal map, the system simultaneously displays the full attribute information of the plot, such as the value assessment result, risk level, and business process progress; when a user searches and filters plots by attribute conditions, the system automatically locates and highlights the spatial location of the corresponding plot on the spatiotemporal map, realizing two-way linkage between querying attributes by map and locating maps by attributes.

[0056] Optionally, the spatiotemporal visualization module supports various thematic map display modes such as risk heat maps, value distribution maps, and acquisition progress maps. It can switch the corresponding visualization form according to different business scenarios to meet diverse display needs such as daily monitoring, risk investigation, and decision reporting.

[0057] In this embodiment, a multi-source data preprocessing module performs standardized processing on the multi-source heterogeneous business data and spatial data of the target land reserve project, outputting a standardized land parcel feature dataset to achieve unified adaptation of heterogeneous data and provide a consistent and reliable data foundation for subsequent calculations. A land value dynamic assessment module calculates the initial land value assessment result based on an adaptive feature weight model, and after correction by combining risk correction parameters, outputs the dynamic assessment result, improving the objectivity and dynamic adaptability of land value assessment. A risk transmission inference module configured with a Bayesian network risk transmission inference model completes risk transmission inference and comprehensive risk level calculation, while generating risk correction parameters and feeding them back to the value assessment module to achieve two-way linkage between value and risk. A closed-loop control execution module generates control instructions and tracks the disposal progress, and iteratively optimizes model parameters based on the disposal results, forming a closed-loop optimization mechanism for value assessment and risk control, improving the refinement and intelligence level of land reserve project management.

[0058] Furthermore, it should be noted that the combination of the various technical features in this case is not limited to the combination methods described in the claims of this case or the combination methods described in the specific embodiments. All technical features described in this case can be freely combined or combined in any way, unless they contradict each other.

[0059] It should be noted that the above examples are merely specific embodiments of the present invention, and the present invention is obviously not limited to the above embodiments, with many similar variations. All modifications that can be directly derived or conceived by those skilled in the art from the content disclosed in this invention should fall within the protection scope of this invention.

[0060] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A land reserve project valuation and risk management system, characterized in that, It includes a multi-source data preprocessing module, a land value dynamic assessment module, a risk transmission reasoning module, and a closed-loop management and control execution module; The multi-source data preprocessing module is used to collect multi-source heterogeneous business data and spatial data of the target land reserve project, perform standardization processing on the multi-source heterogeneous business data and spatial data, and output a standardized land parcel feature dataset. The dynamic land value assessment module is used to receive the standardized land parcel feature dataset, calculate the initial land value assessment result through an adaptive feature weight model, receive risk correction parameters to correct the initial land value assessment result, and output the dynamic land value assessment result. The risk transmission reasoning module is equipped with a risk transmission reasoning model, which is built based on a Bayesian network. It is used to receive the standardized land parcel feature dataset and the dynamic land value assessment results, establish a directed topology based on the causal relationship between risk factors and perform forward probability propagation to calculate the comprehensive risk level of the project. At the same time, it generates the risk correction parameters and feeds them back to the dynamic land value assessment module. The closed-loop management and control execution module is used to receive the dynamic land value assessment results and the project's comprehensive risk level, generate corresponding management and control instructions, match disposal plans, and track the execution progress. After disposal, it generates model correction parameters based on the deviation between the predicted results and the actual execution results, and feeds the model correction parameters back to the dynamic land value assessment module and the risk transmission reasoning module respectively, to complete the iterative optimization of the model parameters and form a closed-loop optimization mechanism for value assessment and risk management.

2. The land reserve project valuation and risk management system according to claim 1, characterized in that, The multi-source data preprocessing module includes a data acquisition unit, a spatiotemporal reference alignment unit, and a semantic mapping unit; The data acquisition unit is used to collect multi-source heterogeneous business data and spatial data corresponding to the target land reserve project; The spatiotemporal reference alignment unit is used to unify all spatial data to the same spatial reference and elevation reference, and establish spatial topological relationships between land parcels. The semantic mapping unit is used to perform semantic standardization transformation on multi-source heterogeneous business data and generate the standardized land parcel feature dataset by combining the spatial topological relationship of the land parcels.

3. The land reserve project valuation and risk management system according to claim 1, characterized in that, The dynamic land value assessment module includes a feature weight calculation unit, a policy semantic analysis unit, and a value correction unit. The feature weight calculation unit is used to receive the standardized land parcel feature dataset, learn the contribution weight of each value feature indicator through the adaptive attention mechanism built into the adaptive feature weight model, and calculate the initial land value assessment result. The policy semantic parsing unit is used to extract the constraints and change information in the policy documents, generate value adjustment parameters and input them into the feature weight calculation unit to dynamically update the feature weights. The value correction unit is used to receive the risk correction parameters, perform risk discount correction on the initial land value assessment result, and output the dynamic land value assessment result.

4. The land reserve project valuation and risk management system according to claim 1, characterized in that, The risk transmission reasoning module includes a risk factor management unit, a topology construction unit, and a probability reasoning calculation unit. The risk factor management unit is used to store the hierarchically set underlying risk factors and corresponding judgment thresholds, and to extract corresponding data from the standardized land parcel feature dataset to generate real-time risk values ​​for each underlying risk factor. The topology construction unit is used to establish a directed topology based on the causal relationship between each risk factor and to set the basic inference parameters of the risk transmission inference model. The probabilistic reasoning calculation unit is used to take the real-time risk values ​​of each underlying risk factor and the dynamic land value assessment results as input nodes, perform forward probability propagation, output the comprehensive risk level of the project, and generate the risk correction parameters.

5. The land reserve project valuation and risk management system according to claim 1, characterized in that, The closed-loop control execution module includes a hierarchical early warning unit, a response plan matching unit, and a parameter iteration optimization unit; The graded early warning unit is used to trigger an early warning signal of the corresponding level and generate the control instructions based on the fluctuation range of the dynamic land value assessment results and the comprehensive risk level of the project. The disposal plan matching unit is used to match the disposal plan according to the risk type and warning level, push it to the terminal node with corresponding permissions, and track the execution progress. The parameter iteration optimization unit is used to receive the actual execution result after the processing is completed, compare it with the prediction result to calculate the deviation, generate the model correction parameters and feed them back to the corresponding module.

6. The land reserve project valuation and risk management system according to claim 4, characterized in that, The probabilistic reasoning calculation unit is also used to perform reverse diagnostic reasoning, calculate the risk contribution ranking of each underlying risk factor to the comprehensive risk level of the project, locate key risk causes and generate a risk transmission path map. The risk transmission reasoning module also includes a risk coupling degree calculation unit, which is used to quantify the coupling effect between different risk dimensions. When the coupling degree of two or more types of risks exceeds a preset threshold, the overall risk level of the project is increased.

7. The land reserve project valuation and risk management system according to claim 5, characterized in that, The closed-loop control execution module also includes a disposal cancellation verification unit. The disposal cancellation verification unit is used to re-collect real-time data of the corresponding underlying risk factor after disposal is completed for secondary verification. If the verification passes, the cancellation of the corresponding risk status is completed. If the verification fails, the control instruction is regenerated. The parameter iterative optimization unit uses the gradient descent algorithm to generate model correction parameters, which respectively correct the feature weights of the land value dynamic assessment module and the conditional probability table of the risk transmission inference module.

8. The land reserve project valuation and risk management system according to claim 4, characterized in that, The topology construction unit determines the directed topology by jointly using business prior causal logic and historical risk case data; wherein the business prior causal logic defines the direct correlation between underlying risk factors, and the historical risk case data corrects the conditional probability parameters of the topology connections through a structure learning algorithm.

9. The land reserve project valuation and risk management system according to claim 1, characterized in that, It also includes a data storage module, which is used to store the standardized land parcel feature dataset, the dynamic land value assessment results, the project comprehensive risk level, the underlying risk factors and corresponding judgment thresholds, the model parameters, the disposal plan and disposal execution records, and provides data read and write support for each module.

10. The land reserve project valuation and risk management system according to claim 1, characterized in that, It also includes a spatiotemporal visualization module, which uses a spatiotemporal map as its base and overlays and displays land value data, risk distribution data, and business process data, supporting map-related query and interactive operation.

Images