A land space planning management system and method based on big data

Abstract

The application discloses a kind of land space planning management system and method based on big data, it is related to spatial planning management technical field, in the flood season waterlogging and project approval collaborative scene, using risk and utility double threshold value automatic parameter estimation desensitization, sensitive data can be shared and available are realized, through event time anchoring synchronization, time uncertainty and version pedigree, make cross-source data alignment traceable, can be calculated, control line and index threshold are constructed as constraint atlas, conflict identification and repair candidate are automatically output, and optimal planning scheme is selected with non-dominated solution set, generates approval and land adjustment signal issue, improves decision interpretability and landing efficiency.

Description

Technical Field

[0001] This invention relates to the field of spatial planning and management technology, and in particular to a land spatial planning management system and method based on big data. Background Technology

[0002] Coastal prefecture-level cities often face the simultaneous increase in urban flooding risk and expedited approval demands for projects such as drainage pumping stations and sponge city facilities during the flood season. Existing land spatial planning management often relies on fragmented systems across multiple departments. Inconsistencies in spatial coordinates and event timelines in remote sensing data on land cover, hydrology and meteorology, surface deformation, and approval / land use change data lead to a lack of unified basis for risk identification and approval adjustments. Furthermore, cross-departmental data sharing involves sensitive locations and project identifiers, often resorting to coarse-grained anonymization or manual screening, impacting usability and making it difficult to quantify privacy risks. In addition, control lines and indicator threshold constraints are typically based on static layers or manual review, making it difficult to provide recalculated evidence chains and executable remedial solutions for conflicting plans. This results in a disconnect between prediction, decision-making, and execution, slow iteration, and difficulty in tracing responsibility.

[0003] Currently, Chinese patent application number CN202410636571.8 discloses a land spatial planning management system and method based on big data. The content includes: collecting raw data; performing preliminary formatting and preprocessing on the raw data; integrating the preprocessed dataset; synchronizing the integrated dataset to obtain a synchronized dataset; updating historical land spatial planning data in real time based on the synchronized dataset; and performing pattern recognition and trend prediction based on the real-time updated data to obtain recognition and prediction results; providing planning and suggestions based on the recognition and prediction results to further support policy and resource allocation decisions. This solves the technical problems of existing technologies using large datasets in land spatial planning, which struggle to maintain high quality, consistency, and traceability throughout the processing, while also exhibiting poor data confidentiality and failing to provide accurate data analysis and prediction support.

[0004] The aforementioned technologies are insufficient to achieve traceable and aligned event times, computable and diagnosable planning constraints, and output executable and recalculated optimal planning adjustment schemes to support collaborative management of flood risk and project approval during the flood season, provided that sensitive data can be shared across departments. Summary of the Invention

[0005] The technical problem solved by this invention is that existing technologies are difficult to achieve traceable and aligned event times, calculable diagnosis of planning constraints, and output of executable and recalculated optimal planning adjustment schemes to support collaborative management of flood risk and project approval during the flood season, under the premise that sensitive data can be shared across departments.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: A big data-based land spatial planning management system includes a basic model building module, a risk constraint desensitization module, a spatiotemporal anchoring synchronization module, a planning constraint map module, a scheme generation and decision-making module, and a result output and execution module. The basic model building module is used to build a basic risk model based on historical planning data and historical monitoring data; The risk-constrained desensitization module is used to perform risk assessment on real-time monitoring data and real-time business data, and to calculate desensitization parameters under the constraints of risk threshold and utility error threshold, thereby generating a desensitized shared dataset. The spatiotemporal anchoring synchronization module is used to perform event time alignment on the desensitized shared dataset and generate time uncertainty and version spectrum, and output the spatiotemporal aligned dataset. The planning constraint graph module is used to construct a constraint graph based on planning control data and perform conflict diagnosis on the spatiotemporal alignment dataset, outputting conflict triggering identifiers and repair candidate data; The scheme generation and decision-making module is used to generate a set of candidate planning schemes based on the basic risk model, constraint map and repair candidate data and select the optimal planning scheme. The result output execution module is used to output adjustment signals based on the optimal planning scheme and send them to the notification terminal.

[0007] Preferably, the basic model building module includes a historical data acquisition unit and a model training unit; The historical data acquisition unit is used to collect historical planning data and historical monitoring data. The historical planning data includes historical planning zoning data and historical construction indicator data. The historical monitoring data includes historical remote sensing land cover data, historical hydrological and meteorological data, and historical surface deformation data. The model training unit is used to train the basic risk model with historical planning data and historical monitoring data as input data, output risk baseline data, and write the risk baseline data into the basic risk model.

[0008] Preferably, the logic of the model training unit is as follows: Historical planning data and historical monitoring data are divided into training set, validation set and test set according to a preset division ratio; Use the training set to adjust the parameters of the spatiotemporal regression model, and use the validation set to select parameter combinations. The basic risk model is trained using historical planning zoning data, historical construction indicator data, historical remote sensing land cover data, historical hydrological and meteorological data, and historical land surface deformation data as input data, and risk baseline data as output data.

[0009] Preferably, the risk constraint desensitization module includes a field classification unit, a risk assessment unit, a parameter solving unit, and a desensitization generation unit; The field classification unit is used to generate field classification identifiers for real-time monitoring data and real-time business data. The risk assessment unit is used to calculate re-identification risk indicators based on field hierarchical identifiers and output risk assessment data; The parameter solving unit is used to calculate the desensitization parameters based on risk assessment data and utility error threshold. The de-identification generation unit is used to generate a de-identified shared dataset based on the de-identification parameters for real-time monitoring data and real-time business data, and write the de-identification parameter identifier.

[0010] Preferably, the logic of the parameter solving unit is as follows: The desensitization parameters are iteratively updated under the constraints that the re-identification risk indicators do not exceed the risk threshold and the planning statistical error of the desensitized shared dataset relative to the real-time monitoring data and the real-time business data does not exceed the utility error threshold. If the risk indicators for re-identification exceed the risk threshold, the spatial generalization granularity or time window granularity corresponding to the field classification identifier will be increased. If the planning statistical error exceeds the utility error threshold, then reduce the spatial generalization granularity or the time window granularity. Until both the risk threshold and the utility error threshold are met simultaneously, the desensitization parameters are output and solidified as desensitization parameter identifiers.

[0011] Preferably, the spatiotemporal anchoring synchronization module includes an event time extraction unit, a window alignment unit, an uncertainty generation unit, and a version genealogy unit; The event time extraction unit is used to extract event time data and collection time data from the de-identified shared dataset; The window alignment unit is used to generate alignment time data based on a preset time window; The uncertainty generation unit is used to generate time uncertainty based on the time deviation parameter between the event time data and the acquisition time data; The version genealogy unit is used to generate version identifiers for alignment time data and record the version genealogy, outputting a spatiotemporal alignment dataset.

[0012] Preferably, the collaborative logic between the window alignment unit and the version lineage unit is as follows: The de-identified shared dataset is aggregated into windows based on event time data, and the time deviation parameter between the collection time data and the event time data within the window is calculated. The event time data is timestamped based on the time deviation parameter to obtain aligned time data; When the event time data of late data falls into the output window, the version identifier is updated and the difference data is written into the version genealogy.

[0013] Preferably, the planning constraint graph module includes a constraint parsing unit, a conflict diagnosis unit, and a repair candidate generation unit; The constraint parsing unit is used to collect planning control data and parse it into constraint nodes and constraint relationships to construct a constraint graph. The planning control data includes planning control line data and planning index threshold data. The conflict diagnosis unit is used to perform spatial overlay detection and index threshold comparison on the spatiotemporal alignment dataset based on the constraint graph, and outputs a conflict triggering identifier. The repair candidate generation unit is used to generate repair candidate data based on the conflict triggering identifier. The repair candidate data includes candidate land use adjustment data and candidate project layout data, and the repair candidate data is transmitted to the scheme generation decision module.

[0014] Preferably, the scheme generation decision module includes a scheme evaluation unit and a non-dominated selection unit, and the result output execution module includes a signal generation unit and a signal transmission unit; The scheme evaluation unit is used to input the repair candidate data into the basic risk model to obtain the risk prediction results, and combine the planning indicator threshold data to generate scheme evaluation data. The non-dominated selection unit is used to generate a set of non-dominated solutions based on the solution evaluation data, under the premise of satisfying the constraint map constraints, and to select the optimal planning solution from the set of non-dominated solutions. The signal generation unit is used to generate adjustment signals based on the optimal planning scheme, and the adjustment signals include approval adjustment signals and land use adjustment signals. The signal sending unit is used to send the adjustment signal, along with the version identifier, the desensitization parameter identifier, and the conflict trigger identifier, to the notification end.

[0015] A big data-based land spatial planning management method includes the following steps: Step S1: Establish a basic risk model based on historical planning data and historical monitoring data; Step S2: Conduct a risk assessment on real-time monitoring data and real-time business data, and determine the desensitization parameters under the constraints of risk threshold and utility error threshold to generate a desensitized shared dataset; Step S3: Perform event time alignment on the desensitized shared dataset and generate time uncertainty and version genealogy, outputting the spatiotemporal aligned dataset; Step S4: Construct a constraint map based on the planning control data and perform conflict diagnosis on the spatiotemporal alignment dataset, outputting conflict triggering identifiers and repair candidate data; Step S5: Generate a set of candidate planning schemes based on the basic risk model, constraint map, and remediation candidate data, and select the optimal planning scheme; Step S6: Output an adjustment signal based on the optimal planning scheme and send it to the notification terminal.

[0016] The beneficial effects of this invention are as follows: In the scenario of flood season waterlogging and project approval collaboration, this invention adopts automatic parameter desensitization using dual thresholds of risk and utility to achieve the sharing and usability of sensitive data. Through event time anchoring synchronization, time uncertainty and version spectrum, cross-source data alignment is made traceable and recalculated. Control lines and indicator thresholds are constructed into a constraint graph, which automatically outputs conflict identifiers and repair candidates. The optimal planning scheme is selected from the non-dominated scheme set, and approval and land use adjustment signals are generated and issued, improving the interpretability of decisions and the efficiency of implementation. Attached Figure Description

[0017] Figure 1 A basic flowchart of a land spatial planning management system based on big data is provided as an embodiment of the present invention; Figure 2 The present invention provides a flowchart of a land spatial planning and management method based on big data, which is an embodiment of the present invention. Detailed Implementation

[0018] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] Example 1, refer to Figure 1 This paper presents a land spatial planning management system based on big data, including a basic model building module, a risk constraint desensitization module, a spatiotemporal anchoring synchronization module, a planning constraint map module, a scheme generation and decision-making module, and a result output and execution module. The basic model building module is used to build a basic risk model based on historical planning data and historical monitoring data.

[0020] The risk-constrained desensitization module is used to perform risk assessment on real-time monitoring data and real-time business data, and to determine desensitization parameters under the constraints of risk threshold and utility error threshold, thereby generating a desensitized shared dataset.

[0021] The spatiotemporal anchoring synchronization module is used to align the events in the desensitized shared dataset and generate time uncertainty and version spectrum, outputting a spatiotemporally aligned dataset.

[0022] The planning constraint graph module is used to construct constraint graphs based on planning control data and perform conflict diagnosis on spatiotemporally aligned datasets, outputting conflict triggering identifiers and repair candidate data.

[0023] The scheme generation and decision-making module is used to generate a set of candidate planning schemes based on the basic risk model, constraint map and repair candidate data, and select the optimal planning scheme.

[0024] The result output execution module is used to output adjustment signals based on the optimal planning scheme and send them to the notification end.

[0025] In the context of flood season waterlogging and project approval collaboration, this invention employs automatic parameter desensitization using dual thresholds of risk and utility to ensure that sensitive data is shareable and usable. Through event time anchoring synchronization, time uncertainty, and version spectrum, cross-source data alignment is made traceable and recalculated. Control lines and indicator thresholds are constructed into a constraint graph, automatically outputting conflict identifiers and repair candidates. The optimal planning scheme is selected from a set of non-dominated schemes, generating approval and land use adjustment signals for issuance, thereby improving the interpretability of decisions and the efficiency of implementation.

[0026] The basic model building module includes a historical data acquisition unit and a model training unit.

[0027] The historical data acquisition unit is used to collect historical planning data and historical monitoring data. Historical planning data includes historical planning zoning data (planning zoning identifiers, zoning boundary data, zoning area data, zoning type data, zoning control intensity data, adjacent zoning set data, zoning center point coordinate data, and zoning bounding box data) and historical construction indicator data (construction land area data, resident population size data, road density data, green space ratio summary data, data source department, and update time). Historical monitoring data includes historical remote sensing coverage data (historical time markers, planning zoning identifiers, multi-band reflectance data, vegetation index data, built-up area index data, water...). The data includes: body index data, zonal statistical summary data, zonal missing measurement ratio data and adjacent time change data, historical hydrological and meteorological data (daily rainfall data, cumulative rainfall data, peak rainfall data, average rainfall data, number of consecutive rainfall days, river water level data, duration of water level exceeding warning level data, wind speed data, wind direction data, air pressure data, temperature data, typhoon impact period data and rainstorm level indicators), and historical surface deformation data (cumulative sedimentation data, subsidence rate data, uplift data, deformation variance data, deformation standard deviation data, abrupt change period data, observation quality indicators, mean deformation within a zonal area, maximum deformation within a zonal area, and deformation quantile within a zonal area).

[0028] The model training unit is used to train the basic risk model with historical planning data and historical monitoring data as input data, output risk baseline data, and write the risk baseline data into the basic risk model.

[0029] The logic of the model training unit is as follows: Historical planning data and historical monitoring data are divided into training set, validation set and test set according to preset division ratios.

[0030] The parameters of the spatiotemporal regression model are adjusted using the training set, and the parameter combinations are selected using the validation set.

[0031] The basic risk model is trained using historical planning zoning data, historical construction indicator data, historical remote sensing land cover data, historical hydrological and meteorological data, and historical land surface deformation data as input data, and risk baseline data as output data.

[0032] This embodiment uses each planning zone in the historical planning zoning data as the calculation object to establish a basic risk model and output risk baseline data. Historical construction indicator data, historical remote sensing land cover data, historical hydrological and meteorological data, and historical surface deformation data are aligned by zone and time to form training samples. The input vector for each sample is denoted as... , among which, Plan partition index, For historical time indexing, The zoning attribute codes of historical planning zoning data, along with historical construction indicator data, historical remote sensing land cover data, historical hydrological and meteorological data, and historical surface deformation data, are used in... and The numerical values ​​are obtained by splicing together the data at the specified points. Historical observations that can be mapped to risk intensity from historical monitoring data (in this embodiment, historical waterlogging records are mapped to intensity values ​​after zoning) are used as monitoring labels. The basic risk model uses a spatiotemporal regression form with spatial adjacency effects: ; in, For normalization function, To and Adjacent partition sets, Adjacent weights and satisfying , To and The partition IDs of adjacent partitions, For adjacent partitions in the previous time Historical risk observations of the step for and The length of the shared boundary.

[0033] ; in, Historical hydrological and meteorological data, This is historical surface deformation data. For historical remote sensing coverage data, To normalize the values ​​to 0~1, , and These are the weighting coefficients.

[0034] Train parameters by minimizing the loss function. : ; in, These are used as sample weights to enhance the impact of samples during periods of high disaster incidence. After training, each partition is weighted at the latest historical moment. Calculate and record as risk baseline data By incorporating risk baseline data into the basic risk model as a benchmark for subsequent predictions, the subsequent risk prediction results are updated based on the risk baseline data and real-time disturbances rather than being re-estimated.

[0035] The risk constraint desensitization module includes a field classification unit, a risk assessment unit, a parameter solving unit, and a desensitization generation unit.

[0036] The field classification unit is used to generate field classification identifiers for real-time monitoring data and real-time business data.

[0037] The risk assessment unit is used to calculate re-identification risk indicators based on field hierarchical identifiers and output risk assessment data.

[0038] The parameter solving unit is used to calculate desensitization parameters based on risk assessment data and utility error threshold.

[0039] The de-identification generation unit is used to generate a de-identified shared dataset based on the de-identification parameters for real-time monitoring data and real-time business data, and write the de-identification parameter identifier.

[0040] The logic of the parameter solving unit is as follows: The desensitization parameters are iteratively updated under the constraints that the re-identification risk index does not exceed the risk threshold and the planning statistical error of the desensitized shared dataset relative to the real-time monitoring data and real-time business data does not exceed the utility error threshold.

[0041] If the risk indicators for re-identification exceed the risk threshold, the spatial generalization granularity or time window granularity corresponding to the field classification identifier will be increased.

[0042] If the planning statistical error exceeds the utility error threshold, then reduce the spatial generalization granularity or the time window granularity.

[0043] Until both the risk threshold and the utility error threshold are met simultaneously, the desensitization parameters are output and solidified as desensitization parameter identifiers.

[0044] The risk constraint anonymization module generates field hierarchical identifiers for real-time monitoring data and real-time business data. These identifiers must include at least a location field, a time field, and an identifier field. Quasi-identifier combinations are then extracted based on these field hierarchical identifiers. ,in It consists of spatial units represented by the location field, time buckets represented by the time field, and item types in real-time business data. All records are then processed by... Grouping to obtain equivalence classes Its scale is Define the unique proportion: ; Define the minimum equivalence class size At the same time, to characterize spatiotemporal sparsity, each record... Count the neighborhood within the same spatial unit and adjacent time buckets. And define sparsity: ; The re-identification risk indicator is ultimately defined as: ; in, , and These are the weighting coefficients. Output risk assessment data, which includes at least the following: , , , And the field classification version on which these statistics are based.

[0045] The utility error threshold is not determined manually, but automatically by the principle of non-reversal of planning decisions. A vector of planning statistics directly related to constraint determination is selected from historical construction indicator data and real-time business data. In this embodiment The data is calculated solely from existing data, specifically including: the incremental zoning construction indicators derived from real-time land use change data (change identifier, planning zoning identifier, change scope data, change time data, change type data, change area data, pre-change indicator value data, post-change indicator value data, project association identifier data, policy basis identifier data, execution status data, and implementation time window data) and real-time approved project data (planning zoning identifier, project location data, project scale data, project type data, approval node identifier, approval node time data, estimated node completion time data, expedited identifier data, return and correction identifier data, construction period water demand data, and construction period land and road occupation demand data), and the zoning benchmark indicators derived from historical construction indicator data. This is calculated from the original real-time data. Calculations were made on the desensitized data. And define the planning statistical error: ; in, It is a very small positive number. On the other hand, to ensure that the threshold constraint does not flip, for each partition... The threshold constraint uses planning index threshold data to give the threshold. And calculate the constraint margin based on the original data. ,in, These are the partition indicator values ​​under the influence of raw real-time data. This embodiment limits the allowable error to [value missing]. ,in A preset safety factor is used. The final utility error threshold is set as follows: ; in, This is the upper limit parameter for statistical error set by the management side. This is how it is obtained. It forms a closed loop with planning indicator threshold data, historical construction indicator data, and real-time business data, and can ensure that the closer the partition is to the threshold, the stricter the requirements.

[0046] The desensitization parameters of the desensitization generation unit are denoted as: ,in, For spatial generalization granularity, For time window granularity, Given the spatial perturbation scale, the desensitization parameters are obtained iteratively by satisfying a double threshold constraint: ; ; in, To preset risk thresholds, spatial unitization is applied to location fields during the anonymization process: coordinates in real-time approved project data and real-time land use change data are processed using spatial unitization. Calculate the spatial cell index: ; If disturbance is required, generate it first. ,calculate After boundary pruning, the spatial unit index is calculated, and time bucketing is applied to the time field: for event time data... Calculate the time bucket: ; Irreversible mapping is used to generate identifier tokens from the identifier field. ,in, The key is recorded along with the de-identified parameter identifier. For real-time monitoring data, the real-time remote sensing cover data, real-time hydrological and meteorological data, and real-time surface deformation data are aggregated at the zoning and time bucket dimensions using the zoning key in the historical planning zoning data, maintaining consistency with the subsequent basic risk model input. The final de-identified shared dataset consists of real-time monitoring data aggregated by zoning, real-time operational data after time bucketing, time buckets, and de-identified parameter identifiers, ensuring that it can be used for constraint map extrapolation and risk prediction while reducing the risk of re-identification.

[0047] The spatiotemporal anchoring synchronization module includes an event time extraction unit, a window alignment unit, an uncertainty generation unit, and a version genealogy unit.

[0048] The event time extraction unit is used to extract event time data and collection time data from the de-identified shared dataset.

[0049] The window alignment unit is used to generate aligned time data based on a preset time window.

[0050] The uncertainty generation unit is used to generate time uncertainty based on the time deviation parameter between the event time data and the acquisition time data.

[0051] The version genealogy unit is used to generate version identifiers for aligned time data and record the version genealogy, outputting a spatiotemporal aligned dataset.

[0052] The coordination logic between window alignment units and version lineage units is as follows: The de-identified shared dataset is aggregated into windows based on event time data, and the time deviation parameter between the collection time data and the event time data within the window is calculated.

[0053] The event time data is corrected by timestamp based on the time deviation parameter to obtain aligned time data.

[0054] When the event time data of late data falls into the output window, the version identifier is updated and the difference data is written into the version genealogy.

[0055] For each record in the de-identified shared dataset, retain event time data. and collection time data Calculation delay Within each time window, the statistical delay distribution is calculated, and the time uncertainty is determined by removing the quantiles. ; The time uncertainty width is defined as Window alignment employs robust center deviation correction: make Alignment time data is defined as and put This represents the temporal uncertainty of the corresponding record. The version hierarchy is based on window output granularity: each output of the spatiotemporally aligned dataset generates a version identifier. If the event time data of late data falls into the already output window, a difference data packet is generated and the version identifier is updated. The difference data packet contains at least the key of the affected record, the corrected alignment time data, and the time uncertainty, thereby ensuring that the subsequent risk prediction results and the optimal planning scheme can be recalculated on the specified version identifier.

[0056] The planning constraint map module includes constraint resolution unit, conflict diagnosis unit, and repair candidate generation unit.

[0057] The constraint parsing unit is used to collect planning control data and parse it into constraint nodes and constraint relationships to construct a constraint map. The planning control data includes planning control line data and planning indicator threshold data.

[0058] The conflict diagnosis unit is used to perform spatial overlay detection and index threshold comparison on the spatiotemporal alignment dataset based on the constraint graph, and outputs a conflict triggering identifier.

[0059] The repair candidate generation unit is used to generate repair candidate data based on the conflict triggering identifier. The repair candidate data includes candidate land use adjustment data and candidate project layout data, and the repair candidate data is transmitted to the scheme generation decision module.

[0060] The constraint parsing unit constructs a constraint graph using planning control line data and planning indicator threshold data. The nodes of the constraint graph include partition nodes, control line nodes, and business nodes mapped from real-time approved project data and real-time land use change data.

[0061] The edges of the constraint graph include spatial overlay relationship edges and indicator influence relationship edges. The spatial overlay relationship is calculated from the geometric relationship between spatial units and zoning data and planning control line data, while the indicator influence relationship is calculated from the incremental influence of real-time business data on historical construction indicator data.

[0062] The conflict diagnosis unit performs two types of judgments for each business node: Spatial Overlap Determination: Calculate the intersection of the spatial unit corresponding to the business node with the planning control line data. If the intersection is greater than zero and the control line is a prohibited constraint, output a spatial conflict trigger flag and record the intersection as evidence. Indicator threshold determination: First, calculate the updated values ​​of the zoning indicators based on real-time approved project data and real-time land use change data. Then compare with the planning indicator threshold data In comparison, if Greater than If the output indicator conflict triggers, it will record the excess quantity.

[0063] The repair candidate generation unit generates repair candidate data with the conflict triggering identifier as input: For business nodes with a true spatial conflict triggering identifier, candidate project layout data is generated, and candidate units that satisfy the condition that the spatial conflict triggering identifier is false and the indicator threshold is not exceeded are searched in the set of adjacent spatial units. Several candidates are selected with the weighted sum of migration distance and indicator cost as the target.

[0064] For partition nodes where the indicator conflict trigger flag is true, candidate land use adjustment data is generated, the amount of indicators to be reduced is calculated, and feasible reduction combinations are formed within the partition, so that the adjusted indicators return to within the threshold and do not trigger the spatial conflict trigger flag.

[0065] The solution generation and decision-making module includes a solution evaluation unit and a non-dominated selection unit, while the result output and execution module includes a signal generation unit and a signal distribution unit.

[0066] The scheme evaluation unit is used to input the remediation candidate data into the basic risk model to obtain risk prediction results, and combine the planning indicator threshold data to generate scheme evaluation data.

[0067] The non-dominated selection unit is used to generate a set of non-dominated solutions based on the solution evaluation data, under the premise of satisfying the constraint map constraints, and to select the optimal planning solution from the set of non-dominated solutions.

[0068] The signal generation unit is used to generate adjustment signals based on the optimal planning scheme. The adjustment signals include approval adjustment signals and land use adjustment signals.

[0069] The signal sending unit is used to send the adjustment signal, along with the version identifier, the desensitization parameter identifier, and the conflict trigger identifier, to the notification end.

[0070] The solution generation and decision-making module will encode the repaired candidate data into a set of candidate planning solutions. Each candidate planning scheme consists of two parts: candidate project layout data and candidate land use adjustment data. The conflict triggering indicators are verified on the constraint map and are considered as false if they are all found to be feasible.

[0071] For each feasible solution, real-time perturbation inputs under the corresponding version identifier are extracted based on the spatiotemporally aligned dataset. These perturbation inputs are then overlaid with historical construction indicator data to show the changes in indicators caused by the solution, and input into the basic risk model to obtain the risk prediction results. ; in, It is obtained by aggregating real-time monitoring data in the spatiotemporal aligned dataset along the partition dimension. Data from historical construction indicators The changes in zoning indicators by candidate planning schemes are calculated to generate scheme evaluation data, which includes at least the risk target value. and cost target value ,in, It is obtained by weighted summation of migration distance from candidate project layout data and adjustment amount from candidate land use adjustment data. This refers to the partition weight, which is a preset management preference.

[0072] Perform non-dominated ranking on the candidate programs: like exist and All of the above are not inferior and at least one is superior. ,but Dominate All undominated solutions constitute the non-dominated solution set. Finally, the optimal planning solution is selected from the non-dominated solution set based on the ideal point distance, and an adjustment signal is generated accordingly. The approval adjustment signal is used to adjust the order of approval nodes corresponding to real-time approval project data or to supplement verification. The land use adjustment signal is used to adjust the scope and type of change corresponding to real-time land use change data. At the same time, the version identifier, desensitization parameter identifier and conflict trigger identifier are sent to the notification terminal along with the adjustment signal to meet the management requirements of traceability, recalculation and interpretability.

[0073] Example 2, refer to Figure 2 This paper provides a big data-based method for land spatial planning management, which includes the following steps: Step S1: Establish a basic risk model based on historical planning data and historical monitoring data.

[0074] Step S2 involves conducting a risk assessment on real-time monitoring data and real-time business data, and determining the desensitization parameters under the constraints of risk threshold and utility error threshold to generate a desensitized shared dataset.

[0075] Step S3: Perform event time alignment on the desensitized shared dataset and generate time uncertainty and version spectrum, outputting the spatiotemporal aligned dataset.

[0076] Step S4: Construct a constraint map based on the planning control data and perform conflict diagnosis on the spatiotemporal alignment dataset, outputting conflict triggering identifiers and repair candidate data.

[0077] Step S5: Generate a set of candidate planning schemes based on the basic risk model, constraint map, and repair candidate data, and select the optimal planning scheme.

[0078] Step S6: Output an adjustment signal based on the optimal planning scheme and send it to the notification terminal.

[0079] This invention uses re-identification risk indicators and utility error thresholds as constraints to automatically iterate and calculate desensitization parameters and generate a desensitized shared dataset. This allows the shared data to be used for spatial overlay, indicator calculation and risk prediction, as well as to quantify and control re-identification risk, avoiding the irreproducibility, over-desensitization and under-desensitization caused by empirical parameter selection.

[0080] The system aligns real-time monitoring data and real-time business data with windows based on event time and acquisition time, outputs aligned time data and generates time uncertainty, and records version lineage and difference packages. It supports recalculation and replay after late data is recovered, solving the problems of risk judgment bias and untraceable responsibility caused by high-frequency data disorder and delay during the flood season.

[0081] The planning control line data and planning indicator threshold data are parsed into a constraint map. Spatial overlay detection and threshold comparison are performed on the spatiotemporal aligned dataset. Spatial conflict triggering identifiers and indicator conflict triggering identifiers are output, and conflict evidence is retained, so as to realize automated and interpretable review of project site selection and land use changes.

[0082] Based on conflict triggering indicators, candidate project layout data and candidate land use adjustment data are automatically generated. Combined with the basic risk model, risk prediction results are obtained and scheme evaluation data is formed. The optimal planning scheme is output by screening the non-dominated scheme set and selecting preferences, thus avoiding the problem of providing suggestions that cannot be implemented.

[0083] The optimal planning scheme is transformed into approval adjustment signals and land use adjustment signals, and sent to the notification terminal along with version identifiers, desensitized parameter identifiers, and conflict trigger identifiers. This enables the management of the scheme, execution, and review, and improves the timeliness, compliance, and auditability of flood season emergency project approval and risk control.

[0084] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0085] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the protection scope of the present invention.

Claims

1. A land spatial planning management system based on big data, characterized in that, It includes a basic model building module, a risk constraint desensitization module, a spatiotemporal anchoring and synchronization module, a planning constraint graph module, a scheme generation and decision-making module, and a result output and execution module. The basic model building module is used to build a basic risk model based on historical planning data and historical monitoring data; The risk-constrained desensitization module is used to perform risk assessment on real-time monitoring data and real-time business data, and to calculate desensitization parameters under the constraints of risk threshold and utility error threshold, thereby generating a desensitized shared dataset. The spatiotemporal anchoring synchronization module is used to perform event time alignment on the desensitized shared dataset and generate time uncertainty and version spectrum, and output the spatiotemporal aligned dataset. The planning constraint graph module is used to construct a constraint graph based on planning control data and perform conflict diagnosis on the spatiotemporal alignment dataset, outputting conflict triggering identifiers and repair candidate data; The scheme generation and decision-making module is used to generate a set of candidate planning schemes based on the basic risk model, constraint map and repair candidate data and select the optimal planning scheme. The result output execution module is used to output adjustment signals based on the optimal planning scheme and send them to the notification terminal.

2. The land spatial planning management system based on big data as described in claim 1, characterized in that, The basic model building module includes a historical data acquisition unit and a model training unit; The historical data acquisition unit is used to collect historical planning data and historical monitoring data. The historical planning data includes historical planning zoning data and historical construction indicator data. The historical monitoring data includes historical remote sensing land cover data, historical hydrological and meteorological data, and historical surface deformation data. The model training unit is used to train the basic risk model with historical planning data and historical monitoring data as input data, output risk baseline data, and write the risk baseline data into the basic risk model.

3. A land spatial planning management system based on big data as described in claim 2, characterized in that, The logic of the model training unit is as follows: Historical planning data and historical monitoring data are divided into training set, validation set and test set according to a preset division ratio; Use the training set to adjust the parameters of the spatiotemporal regression model, and use the validation set to select parameter combinations. The basic risk model is trained using historical planning zoning data, historical construction indicator data, historical remote sensing land cover data, historical hydrological and meteorological data, and historical land surface deformation data as input data, and risk baseline data as output data.

4. A land spatial planning management system based on big data as described in claim 3, characterized in that, The risk constraint desensitization module includes a field classification unit, a risk assessment unit, a parameter solving unit, and a desensitization generation unit. The field classification unit is used to generate field classification identifiers for real-time monitoring data and real-time business data. The risk assessment unit is used to calculate re-identification risk indicators based on field hierarchical identifiers and output risk assessment data; The parameter solving unit is used to calculate the desensitization parameters based on risk assessment data and utility error threshold. The de-identification generation unit is used to generate a de-identified shared dataset based on the de-identification parameters for real-time monitoring data and real-time business data, and write the de-identification parameter identifier.

5. A land spatial planning management system based on big data as described in claim 4, characterized in that, The logic of the parameter solving unit is as follows: The desensitization parameters are iteratively updated under the constraints that the re-identification risk indicators do not exceed the risk threshold and the planning statistical error of the desensitized shared dataset relative to the real-time monitoring data and the real-time business data does not exceed the utility error threshold. If the risk indicators for re-identification exceed the risk threshold, the spatial generalization granularity or time window granularity corresponding to the field classification identifier will be increased. If the planning statistical error exceeds the utility error threshold, then reduce the spatial generalization granularity or the time window granularity. Until both the risk threshold and the utility error threshold are met simultaneously, the desensitization parameters are output and solidified as desensitization parameter identifiers.

6. A land spatial planning management system based on big data as described in claim 5, characterized in that, The spatiotemporal anchoring synchronization module includes an event time extraction unit, a window alignment unit, an uncertainty generation unit, and a version genealogy unit; The event time extraction unit is used to extract event time data and collection time data from the de-identified shared dataset; The window alignment unit is used to generate alignment time data based on a preset time window; The uncertainty generation unit is used to generate time uncertainty based on the time deviation parameter between the event time data and the acquisition time data; The version genealogy unit is used to generate version identifiers for alignment time data and record the version genealogy, outputting a spatiotemporal alignment dataset.

7. A land spatial planning management system based on big data as described in claim 6, characterized in that, The collaborative logic between the window alignment unit and the version lineage unit is as follows: The de-identified shared dataset is aggregated into windows based on event time data, and the time deviation parameter between the collection time data and the event time data within the window is calculated. The event time data is timestamped based on the time deviation parameter to obtain aligned time data; When the event time data of late data falls into the output window, the version identifier is updated and the difference data is written into the version genealogy.

8. A land spatial planning management system based on big data as described in claim 7, characterized in that, The planning constraint graph module includes a constraint parsing unit, a conflict diagnosis unit, and a repair candidate generation unit. The constraint parsing unit is used to collect planning control data and parse it into constraint nodes and constraint relationships to construct a constraint graph. The planning control data includes planning control line data and planning index threshold data. The conflict diagnosis unit is used to perform spatial overlay detection and index threshold comparison on the spatiotemporal alignment dataset based on the constraint graph, and outputs a conflict triggering identifier. The repair candidate generation unit is used to generate repair candidate data based on the conflict triggering identifier. The repair candidate data includes candidate land use adjustment data and candidate project layout data, and the repair candidate data is transmitted to the scheme generation decision module.

9. A land spatial planning management system based on big data as described in claim 8, characterized in that, The scheme generation and decision-making module includes a scheme evaluation unit and a non-dominated selection unit, and the result output and execution module includes a signal generation unit and a signal distribution unit. The scheme evaluation unit is used to input the repair candidate data into the basic risk model to obtain the risk prediction results, and combine the planning indicator threshold data to generate scheme evaluation data. The non-dominated selection unit is used to generate a set of non-dominated solutions based on the solution evaluation data, under the premise of satisfying the constraint map constraints, and to select the optimal planning solution from the set of non-dominated solutions. The signal generation unit is used to generate adjustment signals based on the optimal planning scheme, and the adjustment signals include approval adjustment signals and land use adjustment signals. The signal sending unit is used to send the adjustment signal, along with the version identifier, the desensitization parameter identifier, and the conflict trigger identifier, to the notification end.

10. A big data-based land spatial planning management method, applied in a big data-based land spatial planning management system as described in any one of claims 1-9, characterized in that, Includes the following steps: Step S1: Establish a basic risk model based on historical planning data and historical monitoring data; Step S2: Conduct a risk assessment on real-time monitoring data and real-time business data, and determine the desensitization parameters under the constraints of risk threshold and utility error threshold to generate a desensitized shared dataset; Step S3: Perform event time alignment on the desensitized shared dataset and generate time uncertainty and version genealogy, outputting the spatiotemporal aligned dataset; Step S4: Construct a constraint map based on the planning control data and perform conflict diagnosis on the spatiotemporal alignment dataset, outputting conflict triggering identifiers and repair candidate data; Step S5: Generate a set of candidate planning schemes based on the basic risk model, constraint map, and remediation candidate data, and select the optimal planning scheme; Step S6: Output an adjustment signal based on the optimal planning scheme and send it to the notification terminal.

Images