A method and system for monitoring and assessing risks in urban geological survey

Abstract

The application discloses a kind of urban geology survey monitoring risk assessment method and system, it is related to urban geology survey monitoring data processing technical field, including in account mapping under access rainfall and various monitoring, survey and management data, unified field, unit, time scale and registration frequency and effective period, construct data input set;Based on data input set, window slicing and space merging are carried out, and the trigger and response basic sequence and quality label of the output object corresponding window are output;Score is calculated using the response basic sequence and quality label and forms early warning event according to threshold and continuous rule, then the event is mapped into task list and sorting scheduling, write back and archive are completed.The method described in the application completes score grading, continuous early warning event output and task dispatch write back archive closed loop under the constraint of quality label, so as to improve the traceability, stability and management landing efficiency of urban geology risk assessment.

Description

Technical Field

[0001] This invention relates to the field of urban geological survey and monitoring data processing technology, specifically to a method and system for risk assessment in urban geological survey and monitoring. Background Technology

[0002] With the continuous increase in the intensity of urbanization, the development of underground space, the dense layout of rail transit and municipal pipelines have led to urban geological risks such as ground subsidence, landslides, karst collapses, and foundation pit disturbance exhibiting characteristics of multi-source induction, spatiotemporal coupling, and chain diffusion. To support refined governance, related technologies have gradually evolved from traditional surveys and static zoning to an integrated system of investigation, monitoring, and early warning: on the one hand, GIS and spatiotemporal databases are used for object boundary solidification and results management; on the other hand, multi-source monitoring methods such as rainfall, displacement / subsidence, groundwater level / pore pressure, and InSAR achieve near real-time acquisition; at the same time, information-based methods such as data fusion, anomaly identification, threshold determination, and early warning issuance are widely used in urban safety management, providing basic support for risk identification, dynamic classification, and emergency dispatch.

[0003] Despite the established technical system, current urban geological risk assessment still faces challenges in achieving a closed-loop engineering approach for key aspects. First, multi-source data is often managed in a fragmented manner at the site, project, or departmental level, lacking a unified coding and boundary-fixing mechanism centered on the risk object. This makes it difficult to consistently categorize rainfall, deformation, groundwater, and construction monitoring data into the same governance unit, hindering consistent standards in cross-period comparisons and cross-departmental collaboration. Second, existing methods often lack adequate data quality control, failing to standardize the handling of missing data, outliers, equipment offline issues, and data expiration dates. Scoring results are easily affected by noise and data interruptions, making it difficult to establish a traceable chain of evidence and credibility constraints. Third, many assessment schemes rely heavily on single-window thresholds or experience-based judgments, lacking standardized rules for continuous triggering, maintenance, and deactivation. Warning levels are prone to frequent fluctuations, making it difficult to support a stable governance rhythm and accountability. Finally, assessment output is disconnected from task scheduling: risk levels often remain at the alarm display level, failing to form a closed loop with task templates, priority ranking, resource constraint matching, execution write-back, and archiving mechanisms. This makes it difficult to implement the assessment, handling, review, and reassessment process. The aforementioned shortcomings make it difficult for existing technologies to achieve object-oriented merging for urban governance, quality-controllable scoring, continuous rule-driven early warning eventization, and a management closed loop that connects with task lists and scheduling write-back. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by this invention are: existing urban geological risk assessment and early warning management methods have problems such as lack of unified coding and boundary solidification of risk objects, making it difficult to stably merge multi-source data by object; lack of standardized quality labeling for missing data, anomalies, offline data, and the validity period of results, resulting in insufficient credibility and untraceability of assessment results; lack of continuous triggering, maintenance, and release rules and event-based output, resulting in large fluctuations in early warning and difficulty in connecting with governance processes; and how to map assessment results into an executable task list and complete the closed loop of sorting, scheduling, writing back, and archiving.

[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a method for risk assessment in urban geological surveys and monitoring, comprising: classifying and encoding risk objects; solidifying boundaries using GIS; and simultaneously establishing a mapping ledger of attributes, exposures, and monitoring points; accessing rainfall and various monitoring, survey, and remediation data under the ledger mapping; unifying fields, units, and time scales; registering frequencies and validity periods; constructing a data input set; performing window slicing and spatial merging based on the data input set; completing deduplication verification, anomaly and missing measurement processing; and outputting the trigger and response base sequences and quality labels for the corresponding windows of the objects; calculating scores using the response base sequences and quality labels; forming early warning events according to thresholds and continuity rules; and then mapping the events into a task list and completing sorting, scheduling, write-back, and archiving.

[0007] As a preferred embodiment of the urban geological survey and monitoring risk assessment method described in this invention, the risk object division and coding includes determining the object boundary according to one of the following: administrative division unit, road mileage segmentation, rail transit section mileage segmentation, underground pipe gallery segmentation, or hazard point boundary, and solidifying the object boundary with geographic information vector elements; closed surface boundaries are used for face-shaped objects, center lines are used for linear objects and combined with start and end mileage to form segment boundaries, and point or small-scale objects are used for point coordinates and combined with influence radius to generate influence range boundaries; the object category is determined according to the object's main type and written into the enumeration field, and a unique code is generated for the object and the coding rules are solidified. The coding field includes administrative region code, object type code, spatial unit sequence number, and boundary version information, and the effective time and parent-child relationship are recorded by version appending for object splitting, merging, and boundary updates.

[0008] As a preferred embodiment of the urban geological survey and monitoring risk assessment method described in this invention, the synchronous establishment of attribute, exposure, and monitoring point mapping ledgers includes registering a set of basic attribute fields and a set of carrier exposure fields for each risk object. The basic attribute fields include geological zoning type, strata and soil type, slope grade, groundwater sensitivity, soft soil or fill thickness grade, karst development grade, historical hidden dangers and verification record number and update time. The carrier exposure fields include road grade, rail transit line or station, important pipeline category and grade, list of key units, impact range boundary and buffer parameters. A mapping table between objects and data sources is established, which includes data source type, data source identifier, spatial location, association method, association confidence marker, and a list of main monitoring points and backup monitoring points. The mapping relationship is managed synchronously with the object version.

[0009] As a preferred embodiment of the urban geological survey and monitoring risk assessment method described in this invention, the following steps are included: Accessing rainfall and various monitoring, survey, and governance data under the ledger mapping includes registering each data source and assigning a unique identifier, registering the acquisition method, sampling cycle, allowable delay, coverage area, coordinate system, data format, and validity period of the results; the accessed data includes rain gauge triggered data, displacement or settlement monitoring data, water level or pore pressure monitoring data, remote sensing deformation results, foundation pit construction monitoring data, geological survey and exploration results, and urban governance ledger data; a unified field dictionary, unit, and timestamp benchmark are used for various types of data, the original source identifier and update time are retained, and the priority and switching rules of the main data source are solidified for cases where the same object is associated with multiple similar data sources, while recording data source switching events and effective times.

[0010] As a preferred embodiment of the urban geological survey and monitoring risk assessment method described in this invention, the following steps are included: window slicing and spatial merging based on the data input set, which involves unifying the time zone and correcting the time accuracy; slicing the time series according to a preset window length; and managing the validity period of low-frequency results using time slices. Data merging is achieved by performing spatial inclusion, nearest neighbor, buffer intersection, or mileage projection matching based on object boundaries and mapping relationships. Duplicate records with the same timestamp are retained according to quality identification or last-written rules, and conflict markers are recorded. Unit verification is performed on numerical fields, and outliers are marked as outliers or steps according to verifiable statistical rules. Short-term missing measurements are interpolated according to fixed rules, and the interpolation ratio is recorded. Long-term missing measurements are only marked without completion. Simultaneously, the missing measurement rate, outlier ratio, offline percentage, delay anomalies, and measurement point switching are output and bound to the cleaning rule version.

[0011] As a preferred embodiment of the urban geological survey and monitoring risk assessment method described in this invention, the following steps are included: calculating scores using response base sequences and quality labels, and forming early warning events according to thresholds and continuity rules. This includes solidifying trigger and response criteria based on the main object type. The trigger criterion is fixed using a cumulative or rolling cumulative rainfall index, and the response criterion is fixed using a rate of change index corresponding to the object type. A criterion reference relationship is established in the object type rule table. A threshold table is established, and the grading threshold version and effective time are solidified according to object type or geological zone. The threshold sources include code references, historical event calibration, or zone statistics. Data availability markers are introduced during the calculation process. When the window data has low reliability, it will be included in the data verification branch and a verification event record will be generated. The remaining windows will be graded according to the threshold table, and the score composition fields and evidence chain snapshots will be output.

[0012] As a preferred embodiment of the urban geological survey and monitoring risk assessment method described in this invention, the following steps are included: calculating scores using response base sequences and quality labels, forming early warning events according to thresholds and continuous rules, mapping events to task lists, and completing sorting, scheduling, writing back, and archiving. This includes maintaining the early warning status of objects and solidifying triggering, maintaining, and deactivating rule parameters; generating triggering, escalating, downgrading, deactivating, or data verification event records based on continuous window conditions; establishing a mapping table from events to task templates and instantiating tasks according to object type, early warning level, dominant type, and low-confidence markers. Task fields at least include object location index, event time, deadline, responsible unit, collaborating unit, and evidence chain summary; hierarchically sorting according to early warning level, scoring order, exposure level, deadline, and resource availability, and generating a dispatch plan under constraints of personnel, vehicles, equipment, and available time periods; and recording arrival time, verification results, image evidence index, retest value, measure type, and false alarm / missed alarm markers using structured fields for task execution writing back, and archiving them in a versioned manner with event number and rule version.

[0013] Another objective of this invention is to provide a risk assessment system for urban geological survey and monitoring, which can calculate scores using response base sequences and quality labels and generate early warning events according to thresholds and continuous rules. This solves the problems of current urban geological risk assessment and early warning management methods, which lack continuous triggering, maintenance and cancellation rules and event-based output, resulting in large fluctuations in early warnings and difficulty in connecting with governance processes.

[0014] As a preferred embodiment of the urban geological survey and monitoring risk assessment system of the present invention, it includes: an object database encoding module, a data access preprocessing module, and an assessment and early warning dispatch closed-loop module; the object database encoding module is used to classify urban geological risk objects, solidify GIS boundaries and generate unique codes, and establish attribute, exposure and measurement point mapping ledgers; the data access preprocessing module is used to map and access rainfall and monitoring, survey and management data according to the ledger, unify field units and time scales and perform window slicing, spatial merging, deduplication verification, anomaly and missing measurement processing, and output trigger and response basic sequences and quality labels; the assessment and early warning dispatch closed-loop module is used to calculate scores based on basic sequences and quality labels and form early warning events according to thresholds and continuity rules, and then map and generate task lists and complete priority sorting, resource scheduling, execution write-back and archiving.

[0015] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement a method for risk assessment of urban geological survey and monitoring.

[0016] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a method for risk assessment of urban geological survey and monitoring.

[0017] The beneficial effects of this invention are as follows: The urban geological survey and monitoring risk assessment method provided by this invention divides and encodes urban geological risk objects and solidifies their boundaries using GIS. It simultaneously establishes a mapping ledger of attributes, exposures, and monitoring points, realizing a unified governance unit with the risk object as the primary key. This unit stably merges multi-source information such as rainfall, deformation / settlement, water level / pore pressure, InSAR, foundation pit construction monitoring, and survey and governance ledgers into a single object, maintaining consistent caliber across different periods. This avoids the problem of unverifiable assessments caused by data dispersion and difficulty in alignment in existing technologies. Furthermore, by unifying data fields, units, and time scales under the ledger mapping and registering sampling frequency, allowable delays, and result validity periods, a directly calculable data input set is formed, eliminating the need for multi-source data scaling. The system addresses the issues of degree discrepancies and the misuse of outdated low-frequency results, thereby improving the interpretability and auditability of the assessment process. Window slicing and spatial merging are then used to perform deduplication verification, anomaly identification, and missing data processing. The output includes trigger / response base sequences and quality labels for the corresponding windows, which are used to synchronously bind scoring inputs with data credibility, thus reducing the misleading impact of noise, offline data, and missing data on early warnings. Finally, scores are calculated based on the response base sequences and quality labels, and early warning events are generated under threshold and continuous trigger / hold / release rules. These events are then mapped to a task list and prioritized, resource-scheduled, executed, and archived. This ensures that assessment results are directly applied to executable governance tasks and form a closed-loop evidence chain, thereby achieving an event-driven, traceable, and implementable urban geological risk management process. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 The first embodiment of the present invention provides an overall flowchart of a risk assessment method for urban geological survey and monitoring. Detailed Implementation

[0020] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0021] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for risk assessment of urban geological survey and monitoring is provided, comprising: S1: Define and code risk objects, solidify boundaries using GIS, and simultaneously establish attribute, exposure, and measurement point mapping ledgers.

[0022] Furthermore, a risk object database is established within the urban area, using manageable spatial units as boundaries. These spatial units are determined by one of the following: administrative divisions, road mileage segments, rail transit section mileage segments, underground utility tunnel segments, or boundaries of geological hazard points. Each spatial unit is registered as a risk object. The boundaries of risk objects are fixed using GIS vector data, which can be either polygon or line features: For polygonal objects such as settlement zones, karst-prone areas, soft soil areas, and fill areas, closed polygonal features are used as boundaries; for linear objects such as roads, rail transit, and utility tunnels, centerlines are used in conjunction with mileage markers or the start and end points of intervals to form segmented boundaries; for point-like or small-scale objects such as landslides, collapses, and ground subsidence hazard points, point coordinates are used in conjunction with the radius of influence or on-site verification boundaries to form polygonal features. Once the object boundary is determined, the coordinate system, data source, collection time, and version number must be recorded, and the boundary geometry fields and visualization index fields must be permanently stored in the object database.

[0023] During the risk object classification process, objects are divided into at least the following categories based on object type, and category fields are registered: ground settlement objects, landslide / slope objects, karst collapse-prone objects, soft soil / fill-affected objects, foundation pit construction-affected objects, road and underground pipeline corridor objects, and rail transit section objects. Object type fields are fixed using enumerated values. For situations where multiple types may overlap within the same spatial boundary, a main type combined with a set of subtypes is used for registration. The main type is used to determine the unique response quantity, and the subtypes are used to supplement the registration of associated risk labels. The main type determination rules are executed according to the fixed priority order, such as prioritizing foundation pit construction-affected objects over settlement objects, settlement objects over soft soil objects, and landslide objects as independent priority, etc., and these priority tables are fixed in the rule version.

[0024] To ensure consistency of objects during cross-departmental data merging and cross-period updates, for each risk object... A unique code is generated and its rules are solidified. The unique code is constructed by concatenating fixed fields, which must include at least: an administrative region code field, an object type code field, a spatial unit number field, and a boundary version field. For linear objects, a mileage start point field and a mileage end point field are added; for point objects, a point position number field or a grid index field is added. The code uses a parsable string encoding format or an integer encoding format, and specifies the range and format of field values ​​used during code generation. For example, the administrative region code uses the national / local standard administrative division code, the object type code uses a preset enumeration code table, and the spatial unit number is auto-incremented within the same administrative region and the same object type and cannot be repeated. For situations where object boundaries change, a method of keeping the code unchanged and incrementally incrementing the boundary version is used for management; for situations where objects are split or merged, a method of retaining historical codes and adding new derived codes is used for management.

[0025] Regarding the standardization of fields in the object ledger, structured fields are provided for data access, fixed definitions, and association with threshold tables, registering a set of basic attribute fields for each risk object. The basic attribute field set includes at least: object name, object type, geological zoning, strata and soil type, terrain slope grade, groundwater sensitivity grade, karst development grade, soft soil / fill thickness grade, historical hazard / risk record number, on-site verification time, and verification conclusion. The sources of these fields are limited to: geological survey results, engineering geological zoning maps, borehole and exploration reports, DEM or surveying results, historical hazard ledgers and inspection records, etc., and the source file number or system source identifier and field update timestamp are recorded in the ledger for each field. For missing fields, a missing marker and supplementary data collection task status are used for registration; null values ​​are not used as substitutes.

[0026] Regarding the registration of carrier and exposure information, a carrier list field and an exposure field are fixed for each risk object. The carrier list field includes at least a road grade field, a rail transit line / station field, a category and grade field for important pipelines such as gas / water supply / drainage, and a field for important buildings or important units. The exposure field includes at least an impact radius field or an impact boundary field, and a number and grade field for key units within the impact area. The source of carrier and exposure fields is limited to ledger data from housing and construction, urban management, municipal pipelines, rail companies, or emergency departments, and the data version number and update time are recorded. For determining the impact range, the object boundary is directly used as the impact range for area objects; for linear objects, the impact range boundary is generated using a centerline buffer method and the buffer distance parameter is fixed; for point objects, a circular buffer is generated using the impact radius parameter and the radius value is fixed. The buffer distance or radius value can be set with default parameters according to the object type and can be updated after on-site verification, but the parameter version and the reason for the update must be recorded.

[0027] Regarding the pre-setting of correlations between monitoring and survey data, a data source mapping table is registered for each risk object to link data sources such as rain gauges, groundwater level gauges, GNSS points, inclinometer boreholes, crack gauges, and construction monitoring points to each risk object. Establish fixed associations. The data source mapping table should include at least the following fields: data source type field, data source unique identifier field, data source coordinate field, association rule field, and association confidence flag field. Association rules can adopt one of spatial containment, nearest neighbor, buffer intersection, or mileage projection matching, and specify spatial threshold parameters used when fixing associations in the object library, such as maximum nearest neighbor distance, buffer radius, mileage projection tolerance, etc. When the same object is associated with multiple data sources of the same type, the primary test point flag field is used to determine the primary data source in the unique response quantity caliber. The primary test point flag is determined by verifiable fields such as stable online rate, historical missing rate, and installation quality records, and the primary test point selection rule version is fixed.

[0028] S2: Under the ledger mapping, access rainfall and various monitoring, survey and governance data, unify fields, units, time scales and register frequency and validity period to build a data input set.

[0029] Furthermore, urban geological survey results, online monitoring data, and urban governance ledger data are integrated in a structured manner according to data source, fields, timestamps, and object codes. This ensures that trigger quantities are calculated under the same caliber. Response volume The threshold matching relationship is used to define the access scope, field dictionary, unit specifications, time base and update cycle for each type of data source, and the data source metadata (source unit, interface method, data version number, sampling frequency, data delay and coverage) is fixed in the data source registry.

[0030] Regarding data source registration and access methods, a unique data source identifier is assigned to each data source, and a data source type enumeration field is registered. The data source types include at least: rain gauges / weather stations, groundwater level gauges / pore pressure gauges, GNSS displacement points / leveling points, inclinometers / crack gauges, InSAR settlement results, foundation pit construction monitoring systems, geological survey results documents, engineering survey reports, bearing capacity and pipeline ledgers, key unit and population distribution data, etc. Each data source identifier corresponds to a metadata record, which includes at least: data acquisition method (interface / file / database synchronization), data update time, sampling period, data format (JSON / CSV / table), coordinate system and geographic location fields, coverage period and coverage area fields, and quality fields (whether it includes a built-in quality identifier and whether it includes an equipment status field). The metadata records are then associated with a data source mapping table, enabling each risk object to... It can retrieve the set of data source identifiers associated with it through mapping relationships.

[0031] Regarding the integration of survey results data, geological survey results, engineering geological zoning maps, borehole logging, laboratory test results, geological profiles, and hazard point verification records will be integrated as static or slowly varying data. Integration will be performed for each risk object. A field table for survey results should be established using the primary key. This table should include at least the following fields: stratigraphic type code, weak interlayer marker, weak soil / fill thickness grade, karst development grade, slope grade, groundwater sensitivity grade, historical hazard density grade, survey unit, survey time, and result document number. For survey results originating from documents, in addition to structured fields, document indexes and page ranges or layer numbers should also be registered to ensure that original evidence can be located during verification. For cases where the same object has been updated multiple times, field values, version numbers, and effective dates should be appended to avoid overwriting historical field values, allowing the assessment to trace back to the previous survey version based on the window time.

[0032] Regarding the access of trigger factor data, rain gauge data is used as the default trigger source, and a unified standard is established. Specifically, the raw sequence fields of rain gauges are accessed. The specifications stipulate that the unit is mm, the timestamp is the local standard time zone, and the sampling period is in minutes or hours. Rain gauge data must include at least the station number, latitude and longitude, cumulative or incremental rainfall fields, and equipment status field. For incremental rainfall data, no cumulative calculation is performed during access; only the original incremental sequence is entered into the database by timestamp. For cumulative rainfall data, the accumulation method and zeroing rule are recorded during access to avoid aggregation errors. For cases where multiple rain gauges cover the same object, the primary rain gauge is determined based on established nearest neighbor / buffer intersection / spatial inclusion association rules, and its marker is written into the object-data source mapping table. If the primary rain gauge is offline, the mapping table automatically switches to alternative rain gauges according to priority, and the switching event and effective time are recorded to ensure... The calculations can explain the changes in its data sources.

[0033] Regarding the access to response factor data, a unique data source access list with distinct response criteria is established based on object type to facilitate calculation. Inconsistencies in indicator selection should be avoided. For ground settlement objects, leveling or GNSS settlement sequences should be prioritized, with fields including at least the measurement point number, coordinates, settlement value, sampling time, and measurement point status. For areas with InSAR results, InSAR products should be accessed in the form of time slice—settlement raster / vector point—solution version, and the image pair time and orbit information should be registered. The spatial overlay relationship between InSAR measurement points and object boundaries should be solidified into a mapping table. For landslide / slope objects, GNSS displacement, inclinometer displacement, or crack gauge readings should be accessed, with fields including at least the measurement point number, displacement, direction / component identifier, timestamp, and equipment status. For karst collapse-prone areas, groundwater level gauge or pore pressure gauge readings should be accessed, with fields including at least the water level / pore pressure value, temperature compensation identifier, timestamp, and equipment status. For foundation pit impact areas, construction monitoring system data should be accessed, with fields including at least the values ​​of monitoring items such as retaining displacement, surrounding settlement, support axial force, and pore pressure, sampling time, construction stage marker, and alarm status. The interface address and data latency limit of the data source should also be recorded. For situations where there are multiple monitoring points of the same caliber for the same object, the indicators are not synthesized at this stage. Instead, all point sequences are entered into the database, and the main monitoring points and backup monitoring points are marked in the object-data source mapping table. The selection criteria for the main monitoring points are verifiable fields such as stable online rate, missing rate statistics, and installation quality records.

[0034] Regarding the access to urban governance and exposure data, road grades, rail transit lines and stations, gas / water / drainage pipeline categories and grades, and lists of important buildings and key units are used as sources for object exposure fields, and each risk object is used as a basis for data collection. Alternatively, the boundary of the object's impact area can be used as a spatial constraint for merging. During data access, the carrier data is standardized into structured fields: carrier category code, level code, facility name, facility unique number, spatial location or network number, supervising unit, update timestamp, etc., and an association table is established between these and the object code. For key units and population distribution data, at least the data source (POI database / emergency list / population raster), statistical scope (counts / density classification within the impact area), update time, and version number are recorded during access to generate... The level allows for retrospective analysis.

[0035] To achieve windowed computation and availability coefficient statistics, a unified time base and field standard are applied to all time series data. Specifically, all data timestamps are converted to the same time zone and the original time zone field is registered, requiring timestamp accuracy to at least the minute. For fields with inconsistent units, unit unification is achieved during data access; for example, settlement / displacement is unified to mm, water level to m, and rainfall to mm. Time intervals are uniformly described by the sampling period field. For data sources lacking equipment status fields, a status missing marker field is added during access and used as a source of uncertainty in quality annotation. For multi-component displacement data, component naming rules (e.g., east-west / north-south / vertical) are defined and component fields are fixed to ensure the accuracy of calculated response quantities. It is possible to clearly identify which component to use.

[0036] Regarding update frequency and data delay registration, the update cycle and allowable delay threshold are fixed for each data source. Rain gauges are typically updated every 5–60 minutes, groundwater level gauges and pore pressure gauges are typically updated hourly or daily, GNSS and crack gauges can be updated from minutes to hours, leveling and InSAR are updated weekly / monthly / quarterly, and foundation pit construction monitoring is typically updated hourly or daily. For low-frequency data sources (such as InSAR and leveling), the validity period of their time slices must be registered, i.e., the start and end date range of the usable data within the window; for high-frequency data sources, the upper limit of data delay must be registered, and if the upper limit is exceeded, it will be recorded as a delay anomaly in the quality labeling. Both update frequency and delay threshold are written into the data source metadata table and managed with version numbers to avoid changes in quality evaluation criteria due to updates from different systems.

[0037] Regarding the verification and solidification of object-data source binding, an association verification is performed on each data source record according to the defined spatial association rules, generating each risk object. The binding relationship with the data source is defined, and fields such as association method, distance / intersection area, mileage projection error, and confidence flag are written to the binding relationship. For binding relationships with confidence flags below a preset threshold (e.g., nearest neighbor distance exceeds the upper limit, mileage projection error is too large, or buffer intersection area is too small), they are marked as bindings to be verified and will not be used as primary test points. The calculation process generates supplementary verification records for manual confirmation. After the binding relationship is confirmed, the list of primary and backup test points needs to be solidified. In the event of data source changes, test point migrations, or test point obsolescence, the effective time and reason for the change should be recorded by adding a new binding version to ensure that the evaluation records can be traced back to the data source set used at that time.

[0038] S3: Based on the data input set, perform window slicing and spatial merging to complete deduplication verification, anomaly and missing test handling, and output the trigger and response basic sequence and quality label of the corresponding window of the object.

[0039] Furthermore, regarding time base unification and window slicing, firstly, a unified time zone and time precision correction is performed on all time series data, converting timestamps to a unified local standard time zone while retaining the original time zone and original timestamp fields for audit backtracking. Subsequently, a regularized sampling index is generated for each type of data source: an equally spaced time axis is generated for minute / hourly data, and a time-slice index is generated for daily / weekly / monthly low-frequency data. The evaluation window length is then fixed at the object level. (Default is 1 day, 1 hour if necessary). During window slicing, the original sequence is aggregated into a sample set within the window according to the window boundaries, retaining the window start and end times and sample count fields. For trigger inputs with rolling cumulative calibers (e.g., 24-hour cumulative rainfall), an additional rolling window index is generated within the 1-hour window, and the rolling length parameter (e.g., 24 hours) and sliding step size (e.g., 1 hour) are recorded to ensure... The calculation caliber is decoupled from the window setting and can be recalculated.

[0040] In terms of spatial merging and measurement point mapping, based on the fixed object boundaries and the generated object-data source binding relationship, each measurement point data is merged to its respective risk object. In the data domain, for point measurement points (rain gauges, GNSS, crack gauges, water level gauges, etc.), they are directly merged according to their binding relationships; for linear or area results (InSAR graticules, settlement contour lines, geological zoning layers, etc.), spatial overlay is performed on the result layers according to the object boundaries to form statistical or representative values ​​of the object and time slices, and the overlay method (area-weighted average, maximum value, center point value, etc.) and overlay parameters are recorded. When multiple measurement points of the same caliber exist for the same object, they are not arbitrarily weighted and synthesized at this stage. Instead, an object-level master sequence is generated based on the rule of priority for primary measurement points and traceability of backup measurement points: the primary measurement point is determined by the primary measurement point marker; when the primary measurement point is missing or offline, it is switched according to the priority of backup measurement points, and the switching event, switching reason, and effective time are recorded to ensure response volume. The source of its value is clear.

[0041] Regarding duplicate record handling and unit consistency verification, duplicate record detection is performed on each data source sequence by timestamp. If multiple records exist at the same timestamp, the records to be retained are determined based on the quality identifier, equipment status, or last update time field provided by the data source. If there is no quality identifier, the record with the most recently written timestamp is retained according to fixed rules, and that timestamp is marked as a duplicate conflict. Simultaneously, unit consistency verification is performed on all numerical fields, checking whether the units for rainfall, displacement / settlement, water level, and axial force are mm, etc. Any records with inconsistent or missing units are marked as unit anomalies and included in the quality label to avoid calculation errors. , Unexplained scale errors sometimes occur.

[0042] In terms of outlier identification and denoising, verifiable statistical rules are used for outlier detection in both trigger and response sequences, without employing uninterpretable adaptive thresholds. For non-negative sequences such as rainfall, records with negative values ​​or exceeding physical upper limits are detected and marked. For response sequences such as displacement / settlement / water level, outlier detection is performed on samples within the window using the 3σ rule or IQR rule. Specifically, the mean and standard deviation, or the quartiles and interquartile ranges, are calculated on the rolling baseline segment of the same object and data source, and samples exceeding the threshold are marked as outliers. For sequences with significant step changes (e.g., sudden jumps caused by equipment resets), the difference between adjacent samples exceeding a preset upper limit is detected and marked as a step anomaly. The rule type, window length, and threshold parameters used in the above anomaly detection are all recorded using the rule version number. The marked samples retain their original values ​​and are marked with anomaly tags, without directly overwriting the original records. For scenarios requiring the generation of cleaned sequences, outliers in the cleaned sequences are handled as missing data and enter the missing data completion process to ensure that the index calculation is not contaminated by significant error values.

[0043] Regarding missing data identification and completion processing, the missing data status is statistically analyzed for each object-data source sequence within each evaluation window. Miss data identification includes: the difference between the number of samples and the actual number of samples on the timeline, the duration of consecutive missing data, and the proportion of offline status. For short-term missing data where the duration of consecutive missing data does not exceed a preset upper limit (e.g., not exceeding 20% ​​of the window length or not exceeding N consecutive sampling points), linear interpolation is used for completion, and the start and end points of the interpolation, the length of the interpolation segment, and the interpolation ratio are recorded. For long-term missing data or when the proportion of offline status exceeds the upper limit, no interpolation completion is performed; only the missing data marker is retained, and the window is marked as a low-confidence window. For low-frequency results (such as InSAR monthly / quarterly, leveling weekly / monthly), a time-slice validity period is used: within the validity period, the result is assigned a usable value and its corresponding time-slice number is recorded; after the validity period, it is marked as missing data, avoiding the incorrect use of expired results for high-frequency window evaluation.

[0044] For object-level window sequence generation, two core input sequences are generated according to object type: a trigger sequence and a response sequence. The trigger sequence generates the window cumulative rainfall by summing the incremental rainfall from the rain gauge within the window, or generates the 24-hour rolling cumulative rainfall by using a 1-hour window. During generation, the rain gauge used, the cumulative caliber (window cumulative / rolling cumulative), window boundaries, and number of samples are recorded. The response sequence selects the main measuring point sequence based on a unique caliber for each object type, extracts the response values ​​at the start and end times within the window, and obtains the window change rate by dividing the difference by the window duration. During generation, the main measuring point, response caliber type (settlement rate / displacement rate / water level change rate / containment displacement rate), start and end values, difference value, and corresponding timestamp are recorded. If a main measuring point is missing within the window but a backup measuring point is available, the system switches to the backup measuring point and records the measuring point switching marker in that window to ensure... The source is transparent.

[0045] Regarding the output and solidification of quality label sets, for each risk object In each window Output a quality tag set for reliability correction and order dispatch processing. The quality tag set must include at least: the missing test rate. (Percentage of missing samples within the window), percentage of outliers (Percentage of samples marked as outliers / step anomalies / unit anomalies within the window), percentage of offline time (Percentage of devices in offline status within the window), data latency (percentage of data exceeding the latency threshold), measurement point switching marker (whether the main measurement point has been switched to the backup measurement point), low-frequency results validity period marker (whether the InSAR / leveling results are within the validity period), unit anomaly marker, etc.

[0046] It should be noted that, in terms of feature system organization, features are divided into four categories and stored in separate tables: trigger-type features. Response-type features Survey and vulnerability characteristics With exposure-type features Among them, trigger-type and response-type features change with the window and are fast variables; investigation and vulnerability-type features, and exposure-type features are object-level slow variables, which only change when the ledger is updated or the investigation results are updated.

[0047] In response feature extraction, a unique response criterion is fixed according to object type, and this criterion is written into the object type-feature mapping table at the feature level to ensure computation. Multiple caliber selections are not always present. Specifically, for ground settlement objects, the settlement rate is extracted using a fixed response-type feature. and its window difference component; for landslide / slope objects, displacement rate is extracted by fixing the response class features. Or the displacement rate of the specified directional component; for karst collapse-prone areas, the groundwater level change rate is extracted using fixed response features. For the impact zone of the foundation pit, the displacement rate of the retaining wall is extracted using fixed response features. When multiple measurement points exist for the same object, response features are generated only for the main measurement point sequence, and the main measurement point identifier and measurement point switching marker are written into the feature record. When a backup measurement point switch occurs, the main measurement point identifier after the switch and the switch event number are written into the feature record simultaneously, enabling window calculation to trace changes in the source of measurement points. The units and reasonable ranges of the response features are also considered.

[0048] In terms of investigation and vulnerability feature extraction, object-level slow variable fields are read from the object ledger and investigation result field table, and converted into structured features that can be used for grading. These fields include at least: geological zoning type code, stratigraphic type code, soft soil / fill thickness grade, slope grade, karst development grade, groundwater sensitivity grade, historical hazard density grade, and historical hazard record count. For investigation results sourced from text or maps, semantic inference is not performed; only structured or directly coded fields are used, and the source file number and field update timestamp are retained in the feature record. For missing fields, the missing field marking method is used instead of default values, and the missing field marking is written into the feature record so that grading rules can identify situations where evidence is insufficient.

[0049] For exposure-related feature extraction, a list of carriers within the object's influence range is read from the carrier-exposure association table and converted into structured fields required for exposure level generation. These fields include at least: road grade code, presence markers for rail transit lines / stations, grade codes for important pipelines such as gas / water supply / drainage, presence markers for important buildings or key units, segmented values ​​for the number of key units, version number of the influence range boundary, and buffer parameters. For the influence range of linear objects, fixed buffer parameters are used to generate buffer boundaries and the generation parameter version is recorded; for the influence range of point objects, influence radius parameters are used to generate influence area boundaries and the radius value and version number are recorded. Exposure-related features do not generate a final level; only necessary fields for rule-based classification are retained to avoid introducing difficult-to-trace comprehensive indicators at the feature layer.

[0050] Regarding the standardization of feature calculation parameters and value ranges, a feature dictionary table is established. For each feature, the field name, data type, unit, source table, calculation method, window method (window cumulative / rolling cumulative / start and end difference), reasonable value range, out-of-range handling flag, missing value handling flag, and version number are clearly defined. For trigger features requiring rolling cumulative values, the feature dictionary fixes the rolling length parameter (e.g., 24h, 3d) and step size parameter. For response features, the feature dictionary fixes the component rules used (e.g., vertical component for settlement objects, horizontal composite quantity or specified direction component for slope objects), and binds these rules to the object type association table to avoid errors caused by differences in data source components. Calculate the change in aperture.

[0051] Regarding the pre-setting of object types, characteristic calibers, and threshold reference relationships, a relationship table is established to bind object types with the required trigger calibers, response calibers, and threshold table reference rules. Specifically, this includes: trigger caliber identifier, response caliber identifier (settlement rate / displacement rate / water level change rate / containment displacement rate), trigger threshold table ID, response threshold table ID, and their applicable partition fields.

[0052] Regarding feature record generation and version management, one or more feature records are generated for each object per window. These records include: object code, window number, window start and end times, trigger feature value, response feature value, corresponding data source, cleansing rule version number, measurement point switching flag, low-frequency result validity period flag, missing flag, and out-of-range flag. For object-level slow variable features, a new slow variable version is generated and its effective time is recorded when the object is updated. This version is then linked to the currently valid slow variable version in the window feature record via a foreign key, ensuring that the feature set of any window can reconstruct its original survey and exposure conditions.

[0053] S4: Calculate scores using response base sequences and quality labels, generate early warning events based on thresholds and continuity rules, map events to task lists, and complete sorting, scheduling, write-back, and archiving.

[0054] Furthermore, to avoid the risk that trigger indicators are difficult to obtain or verify, the trigger amount will be... The data is fixed as aggregated results from rain gauge stations that can be obtained stably over a long period. Specifically, it incorporates minute-level or hourly rainfall sequences from meteorological stations or regional rain gauge stations. (Unit: mm), and the trigger value is obtained by summing the values ​​within the window range. ;when Timing, This refers to the cumulative rainfall for that day; when When the hour is used and rolling accumulation is employed, This represents the cumulative rainfall in the 24 hours preceding the end of that hour.

[0055] For ground settlement objects, the settlement rate is fixed, and the monitoring quantity used is the settlement value. Divide by the difference between the start and end values ​​of the window. get The displacement rate is fixed for the landslide / slope object, and the monitoring quantity used is displacement. Obtained in the same way In karst collapse-prone areas, the rate of change in groundwater level is fixed, and the monitoring quantity used is water level. ,get The displacement rate of the retaining wall is fixed in the affected area of ​​the foundation pit, and the monitoring quantity used is the retaining wall displacement. ,get To facilitate project implementation, the common usable ranges for response quantities are given: settlement / displacement rates typically fall within [-50, 50] mm / day, and water level change rates typically fall within [-2, 2] m / day.

[0056] in, Indicates the start time of the window; Indicates the end time of the window; and These represent the settlement readings at the start and end of the window, respectively; , Indicates the displacement readings at the start and end times of the window; , Indicates the starting and ending water level readings of the window; , This indicates the starting and ending displacement readings of the enclosure.

[0057] For trigger threshold For example, rainfall thresholds in local regulations or geological disaster early warning rules should be used first; if there are no unified regulations, historical calibration methods should be used: collect records of hazard escalation or dangerous situations in the city over the past 3-5 years, and extract the corresponding window before the event occurred. The sample set was used, and the 0.90 quantile was taken as... When there are significant geological zoning differences within a city, threshold tables can be created for soft soil areas, karst areas, and mountainous areas. (Regarding response thresholds...) Generally speaking, alarm thresholds from design documents or construction monitoring procedures should be used first (especially applicable to foundation pits, retaining structures, and settlement-related objects); if calibration is required, the corresponding window should be extracted from the confirmed escalation events. The sample was then used to form the 0.90 quantile. To facilitate initial deployment, the initial engineering interval is allowed as threshold version 0 in the absence of historical samples: the settlement / displacement rate threshold is generally set at [2,10] mm / day, the water level change rate threshold is generally set at [0.05,0.30] m / day, and the foundation pit retaining displacement rate threshold is generally set at [1,5] mm / day.

[0058] In terms of data credibility, a data availability coefficient is introduced. As a credibility correction item, the It is calculated directly from the quality label of the output, without subjective weighting. The quality label is fixed as the missing rate. Outlier ratio Offline time percentage All of them are statistically significant proportions between 0 and 1; Fixed deduction rules are adopted: If the calculation result is less than 0, it is set to 0; if it is greater than 1, it is set to 1, thus ensuring... And it is completely recalculated.

[0059] Converting triggering and response into over-limit ratios, the hazard level is expressed as:

[0060] in, It indicates a dangerous situation.

[0061] Introducing vulnerability levels With exposure level Both are generated according to rules from verifiable fields in the object ledger, rather than being subjectively assigned. The vulnerability level uses at least the following available fields: soft soil / fill thickness level (from borehole / geological survey report), slope level (from DEM / surveying), karst development level (from exploration / general survey), and historical hazard density level (from hazard ledger), and is determined as level 1–5 based on the number of entities meeting high-risk conditions; the exposure level uses at least the following available fields: road level, rail transit line / station, important pipeline levels such as gas / water supply, and number of key units (hospitals / schools / refugees), and is determined as level 1–5 based on the existence and quantity of critical facilities. To facilitate integration with the hazard level at the same scale, a fixed mapping is used to linearly normalize the levels to [0,1]. Its expression is:

[0062] in, This represents the vulnerability score, ranging from [0,1]. This represents the exposure score, ranging from [0,1].

[0063] It should be noted that, upon obtaining information about the danger, as well as Then, the three factors were fused using unweighted geometric mean, and then... After making credibility adjustments, the overall risk score is obtained. The geometric mean setting ensures that risk, vulnerability, and exposure are all indispensable, avoiding the inflating of overall risk by a single indicator. It also better aligns with the comprehensive assessment of consequences and conditions in urban governance, and is expressed as:

[0064] in, Represents the risk score of the object in the window. After completing the risk scoring and quality label output, the system uses a tiered threshold table, a continuous rule table, a warning status table, and a warning event table as the core data structures to classify and process the scoring sequence of each risk object through a state machine, generating warning event records that can be directly invoked for task generation. The tiered thresholds are fixed by managing them separately according to object type and geological zone. Any threshold adjustments are recorded as version updates, including the effective time and source basis, ensuring that historical assessment results are traceable and recalculated.

[0065] In constructing the tiered threshold system, a set of tiered thresholds is established for each type of risk object, including at least the thresholds for low-risk and medium-risk boundaries, the thresholds for medium-risk and high-risk boundaries, and the high-risk trigger threshold. The threshold table should simultaneously record the object type, geological zone, threshold version, effective date, threshold source category, and source description. The threshold source must have traceable evidence, allowing for reference to local regulations or industry standard clauses, historical hazard and hidden danger escalation sample calibration, or determination based on statistical quantiles of geological zones. When using historical calibration, the sample time range, sample size, extraction calibrator, and calibration method parameters must be recorded. When using statistical quantiles, the start and end dates of the statistical period, zone version, and quantile value must be recorded. The threshold table, trigger threshold, and response threshold are managed separately to avoid misuse of thresholds for different purposes, which could lead to caliber drift.

[0066] Regarding the classification criteria, risk score is used as the sole input for classification, while data credibility constraints are introduced. For data windows marked as low credibility, the system does not directly trigger a higher-level judgment, but instead enters the data verification branch and generates a verification event. For credible windows, the applicable version is read according to the threshold table, and an initial classification of the window is performed. The initial classification result is only used as input to the state machine and is not directly used as the final warning level output to avoid frequent jumps caused by single-window fluctuations.

[0067] Regarding the solidification of continuous triggering and deactivation rules, continuous window parameters are solidified for each object type. These parameters include at least the number of windows that must be continuously satisfied to trigger, the hold conditions after triggering, and the number of windows that must be continuously satisfied to deactivate. This rule is then written into the object type rule table in a versioned manner. A reference relationship is established between continuous rules and threshold versions to ensure that after threshold upgrades or rule adjustments, the old and new rules can correctly switch according to their effective time, and historical windows can still reproduce the changes in warning status according to the rules at that time. For low-confidence windows, it is stipulated that they are not included in the continuous trigger count, but are included in the data gap count. When the data gap count exceeds the threshold, a data interruption event is generated, and the duration of the interruption and the affected data source are recorded.

[0068] In terms of state machine processing, a set of early warning states is maintained for each risk object, including at least five categories: normal, watchful, medium risk, high risk, and data verification. Entry conditions, maintenance conditions, and exit conditions are fixed for each state. Entry conditions are determined by the initial window classification and continuous counting. For example, high risk requires a certain number of consecutive windows reaching the high risk threshold, and the windows must be reliable; the same applies to medium risk. The data verification state is triggered by the continuous appearance of low-reliability windows or data interruption. Maintenance conditions are used to control the handling of short-term declines, avoiding frequent switching of early warning states due to noise. Exit conditions are used for the determination of cancellation or downgrading. For example, cancellation of high risk requires a certain number of consecutive windows below the medium risk threshold, and the windows must be reliable. Each state transition must record the transition time, a summary of the triggering reason, the threshold version used, and the continuous rule version, and simultaneously record a summary of key evidence for that window (e.g., trigger-dominant or response-dominant markers, key raw quantity snapshots, and data reliability snapshots) to form a complete chain of evidence.

[0069] Regarding the generation of early warning event records, when a critical transition occurs such as entering, escalating, downgrading, canceling, or undergoing data verification, an early warning event record is generated and written to the early warning event table. The event record must include at least: object code, event time, event type, event level, trigger condition summary (continuous count and threshold comparison results), score and hazard snapshot, trigger dominant type marker, key raw quantity summary, data credibility and low credibility marker, threshold version number, and rule version number. For segments that continuously maintain the same early warning status, a "continuous record" can be added. The continuous record must include at least the number of continuous windows, the highest score during the period, the lowest credibility during the period, the number of measurement point switches, and the number of data interruptions, facilitating task priority sorting and review arrangements.

[0070] After the early warning event record is generated, the early warning event is transformed into an executable task list, with the task template library, event-to-template mapping table, task instance table, resource ledger table, dispatch plan table, execution write-back table, and closed-loop archive table as the core. Under the resource constraints of personnel, vehicles, equipment and collaborating units, the early warning event is sorted and scheduled. At the same time, a structured write-back and archive mechanism for the execution results is established to form a closed-loop record of the entire chain from assessment to disposal.

[0071] Regarding the establishment of the task template library, a set of task templates covering urban geological risk management is pre-built, and each template is standardized with its applicable object type, applicable warning level, applicable dominant type, required fields, default time limit, and required resource type. Task templates should at least include routine inspections, post-rain inspections, encrypted monitoring, emergency inspections, temporary control suggestions, collaborative disposal suggestions (categories such as drainage and pressure reduction, backfilling and reinforcement, and construction control), data verification, and equipment maintenance. Each template must describe the execution requirements using structured fields, without relying on free text, and template iterations should be managed through version appending to ensure that historical tasks can be traced back to their creation criteria.

[0072] Regarding event-to-template mapping, a mapping rule table is established. Based on the warning event level, event type, object type, trigger dominance type, and low-confidence marker, a list of task templates to be generated is output. The mapping rules use enumerated conditions for fixed implementation, avoiding runtime ad-hoc judgments that could lead to unverifiable results. For example, when high risk and response dominance occur, emergency verification and temporary control tasks are generated, and collaborative handling tasks can be overlaid; when medium risk and trigger dominance occur, post-rain verification and encrypted monitoring tasks are generated; when entering data verification status, only data verification and equipment maintenance tasks are generated, and high-risk handling tasks are suppressed. The mapping rule table needs to record the rule number, conditional expression, output template list, version number, and effective time.

[0073] Regarding task instantiation and field population, each alert event is instantiated into a task list according to mapping rules. The primary key of a task instance consists of the object code, event number, template number, and generation time, and the task status field (pending assignment, assigned, executing, completed, closed) is fixed. A task instance must include at least: object location index (boundary or mileage range), alert level and event time, score and hazard snapshot, trigger dominant type marker, key raw quantity summary, data credibility marker, threshold and rule version number, responsible unit, collaborating unit, default deadline, and verification direction fields. For tasks with missing required fields, a missing marker and supplementary information requirement fields must be written to avoid generating empty tasks that cannot be executed.

[0074] Regarding prioritization, to avoid subjective sorting, a hierarchical sorting strategy is adopted and the sorting strategy version is fixed. Sorting is performed at least in the following order: first, sort by warning level in descending order; within the same level, sort by score in descending order; then sort by exposure level in descending order; then sort by task deadline in ascending order; low-confidence related tasks are placed in a separate queue and sorted in descending order by quality labels such as missing test rate and offline percentage, so as to prioritize data verification.

[0075] Regarding resource constraints and scheduling output, a resource ledger is integrated and a resource availability table is established. The resource ledger includes at least the following: patrol personnel, vehicles, equipment, team capabilities, available time periods and maintenance status, as well as the response time and contact information of collaborating units. During scheduling, resources are matched according to their required type and availability to generate a dispatch plan. The dispatch plan includes at least the task number, dispatch recipient, planned start and finish times, work area or mileage range, required equipment list, and rendezvous point information. For multiple tasks that are spatially adjacent or belong to the same responsible unit, tasks can be merged based on the intersection of object boundaries or distance thresholds, generating task packages and recording merging rule parameters, including the task list and merged version, to reduce duplicate attendance.

[0076] Regarding write-back and evidence collection, a unified list of write-back fields is standardized for different task types. The executing end is required to write back using structured fields, not free text. Write-back fields must include at least arrival time, departure time, patrol route or point list, image evidence index, remeasurement readings, hazard discovery markers and hazard levels, type of measures taken, whether an escalation warning is recommended, and false alarm / missed alarm markers. For encrypted monitoring tasks, additional write-back information is required, including the new monitoring point number, coordinates, sampling period, online time, and data source registration information. For equipment maintenance tasks, the cause of the fault, repair time, replacement part number, and maintenance record number must be written back. All write-backs must include a write-back timestamp and write-back account information, and be associated with the task number and event number.

[0077] Regarding closed-loop archiving and sample accumulation, task instances, dispatch plans, execution write-backs, and attachment indexes are uniformly written into the closed-loop archiving table, and foreign key relationships are established through object codes and event numbers. Archiving adopts an append-only versioning approach, and task status changes are saved as event records, prohibiting the overwriting of historical records. Data that can be used for rule updates is structured and accumulated into monitoring samples and handling effect records. For example, tags that are confirmed as real hazards or false alarms after being triggered are bound to the score at that time, key raw quantities, object types, and partition information. Handling measures are bound to the retest improvement quantity, schedule, and cost and written into the effect table data.

[0078] Example 2, an embodiment of the present invention, provides a method for risk assessment of urban geological survey and monitoring. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiment.

[0079] First, the central urban area of ​​a certain city and its surrounding transition zone were selected as the implementation area, covering typical urban geological scenarios such as soft soil subsidence areas, hazard points on the piedmont slopes, karst-prone areas in the old city, the impact area of ​​rail station foundation pits, and the underground utility tunnel section of the ring line and the existing rail section. Based on existing engineering geological zoning results, hazard survey results, and the distribution of urban infrastructure, the treatment unit was divided into six risk objects. The boundaries of these objects were then drawn and solidified using a geographic information system: isometric objects used closed polygon boundaries, linear objects used centerlines and start / end mileage to form segmental boundaries, and point hazards used location and influence radius to form the influence range boundary. Simultaneously, a unique code was generated for each object, and an object ledger was established. The ledger recorded the object type, stratum and soft soil classification fields, exposed load-bearing structures fields (road grade, rail transit and important pipelines, distribution of key units), and the mapping relationship of measuring points (main measuring point, backup measuring point, associated distance or mileage projection error) and version change records. Subsequently, data was integrated under the constraints of the ledger mapping: two rain gauge stations were used as primary and backup stations, and the sampling period and allowable delay were recorded; settlement objects were integrated with GNSS vertical settlement sequences and phased leveling results; slope objects were integrated with GNSS displacement and crack gauge readings; karst objects were integrated with groundwater level gauges; foundation pit objects were integrated with the retaining displacement of the construction monitoring system; pipe gallery sections were integrated with surface settlement points and inspection records along the route; and track sections were integrated with section settlement monitoring points. The survey results and treatment ledgers (pipeline grade, key unit list, road grade) were structured and stored in the database, with unified field dictionaries, units, and timestamp benchmarks, and the validity period of low-frequency results was registered. The experiment was conducted over a continuous 30-day period, using daily windows for window slicing. High-frequency data was aggregated by window, while low-frequency results were assigned values ​​based on their validity period. Spatial merging was performed based on the mapping relationship between object boundaries and measurement points. Duplicate records were deduplicated and checked. Unit anomalies were marked, and outliers and step anomalies were identified according to verifiable statistical rules. Short-term missing measurements were filled by interpolation, and the interpolation ratio was recorded. Long-term missing measurements were only marked without filling. Simultaneously, the trigger base sequence and response base sequence for each object in each window were output, and quality labels such as missing measurement rate, anomaly point ratio, offline percentage, and measurement point switching were generated. Then, response criteria (settlement rate, displacement rate, water level change rate, retaining wall displacement rate, etc.) and trigger criteria (mainly based on 24-hour cumulative rainfall) were solidified according to object type. The threshold table was initially formed using the method of "procedure alarm value + quantile calibration of the city's hazard sample in the past two years" and the effective time was recorded. When calculating the score, the quality labels were introduced into the credibility constraint, and low-credibility windows entered the data verification branch.Then, based on the hierarchical threshold table and the rules for continuous triggering, maintaining, and deactivation, early warning status and early warning event records are generated. Early warning events are instantiated into task lists according to event-template mapping, and are sorted hierarchically according to early warning level, score, exposure level, deadline, and resource availability to generate a dispatch plan. The execution end writes back structured fields such as arrival time, retest readings, image evidence index, hazard discovery, and measure type according to the template. Finally, the tasks, write-back, and evidence chain are versioned and archived to form a complete closed-loop data chain that can be reviewed.

[0080] Table 1 Experimental Data

[0081] The table lists six object categories covering the most common scenarios in urban geological risk assessment with significant differences in governance chains: soft soil settlement and track sections represent continuous deformation risks; slope hazards represent localized, sudden risks; karst-prone areas represent hydrological and geological coupled risks; and the foundation pit impact zone and utility tunnel corridor exhibit significant engineering disturbance and equipment operation dependence characteristics. Comparative data shows that the advantage of this method is not simply a matter of higher or lower scores, but rather manifests in verifiable differences across four levels: object-based merging, quality constraints, event-based early warning, and task closure. Firstly, the online rates of the main monitoring points in the foundation pit impact zone and utility tunnel section are over 70% and slightly over 70%, respectively, while the missing monitoring rate and the proportion of outliers are significantly higher than other objects, representing a typical situation of unstable data, noise, and offline data coexisting. Existing single-window threshold methods are easily affected by short-term spikes and drifts in such scenarios, resulting in high risk and approximately seven alarm jumps per week, as reflected in the table. Furthermore, the dispatch hit rate is only around 20% to 10%, indicating that alarms are primarily driven by data quality issues and fail to pinpoint actual hazards. In contrast, this method incorporates quality labels into credibility constraints and sets up a data verification branch. Pit and pipe gallery objects are identified as data verification events, significantly reducing the number of alarm jumps and increasing the dispatch hit rate to over 60%. Moreover, the on-site verification conclusions correspond to offline maintenance, sensor replacement, and supplementary data collection, showing a shift in task objectives from blind handling to data link repair. This approach is more aligned with engineering practice from a governance perspective and explains why the increased hit rate stems from the correct matching of task types rather than simply suppressing alarms. Secondly, for soft soil settlement areas and slope hazard points, this method maintained a high score and triggered high-risk events with a low number of jumps (once a week), consistent with the high online rate and low missing data. Simultaneously, the hit rate reached approximately 80%, consistent with the results of on-site verification showing uneven settlement and crack expansion. This indicates that for objects with reliable data and clear response indicators, this method did not suppress risk by introducing quality constraints, but rather outputted actionable events more stably. Thirdly, in karst-prone areas, the peak rainfall during the test period was not low, but the response and water level changes fluctuated with a high number of missing data. Existing methods directly gave high risk with frequent jumps, and the verification showed no signs of collapse, which is a typical false alarm. This method, however, gave a sustained medium risk and shifted the focus of remediation to water level verification. This reflects that the object-oriented response caliber and continuous rules make the early warning more consistent with the continuity of the hydrological process, avoiding equating a single rainfall trigger with a collapse risk. Finally, based on two quantifiable indicators, the number of warning jumps and the order hit rate, this method generally reduces jumps and improves the hit rate across all six objects, with the most significant improvement on objects with unstable data. This demonstrates the advantages of the link from the data input set to the quality label and then to the event and task mapping: warnings no longer remain at the display layer, but close the evidence chain through task templates and write-back archiving, thereby providing a traceable data foundation for subsequent threshold and rule updates.In summary, the tabular data supports the development of a more stable, interpretable, executable, and traceable risk management process in urban scenarios with complex multi-source data of varying quality and limited governance resources. Furthermore, it demonstrates verifiable improvements over the weaknesses of existing technologies in areas such as false alarm suppression, task hits, and governance closure.

[0082] Example 3, an embodiment of the present invention, provides a risk assessment system for urban geological survey and monitoring, including an object database encoding module, a data access preprocessing module, and an assessment, early warning, and dispatch closed-loop module.

[0083] The object database encoding module is used to classify urban geological risk objects, solidify GIS boundaries, and generate unique codes, while establishing attribute, exposure, and measurement point mapping ledgers. The data access preprocessing module is used to map and access rainfall, monitoring, survey, and governance data according to the ledgers, unify field units and time scales, and perform window slicing, spatial merging, deduplication verification, anomaly and missing measurement processing, and output trigger and response basic sequences and quality labels. The assessment and early warning dispatch closed-loop module is used to calculate scores based on basic sequences and quality labels, and form early warning events according to thresholds and continuity rules, then map and generate task lists and complete priority sorting, resource scheduling, execution write-back, and archiving.

[0084] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0085] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0086] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0087] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc. 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 scope of the claims of the present invention.

[0088] 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 scope of the claims of the present invention.

Claims

1. A method for risk assessment of urban geological survey and monitoring, characterized in that, include: Risk objects are identified and coded, their boundaries are solidified using GIS, and attribute, exposure, and measurement point mapping ledgers are established simultaneously. Under the ledger mapping, rainfall and various monitoring, survey and governance data are integrated, fields, units, and time scales are unified, and frequency and validity period are recorded to construct a data input set; Based on the data input set, window slicing and spatial merging are performed to complete deduplication verification, anomaly and missing test handling, and output the trigger and response basic sequence and quality label of the corresponding window of the object. The system calculates scores using response base sequences and quality labels, generates early warning events based on thresholds and continuity rules, maps these events to a task list, and completes sorting, scheduling, write-back, and archiving. The process of calculating scores using response base sequences and quality labels and forming early warning events according to thresholds and continuous rules includes fixing trigger criteria and response criteria according to the main object type. The trigger criteria are fixed to use the cumulative rainfall index of window or rolling accumulation, and the response criteria are fixed to use the rate of change index corresponding to the object type. A criterion reference relationship is established in the object type rule table. Establish a threshold table and solidify the grading threshold versions and effective times according to object type or geological zone. Threshold sources include code references, historical event markings, or zone statistics. Data availability markers are introduced during the calculation process. When the window data is of low confidence, it will be included in the data verification branch and a verification event record will be generated. The remaining windows will be judged according to the threshold table and the score composition fields and evidence chain snapshot will be output. The process involves calculating scores using response base sequences and quality labels, forming early warning events based on thresholds and continuous rules, mapping the events to a task list, and completing sorting, scheduling, writing back, and archiving. This includes maintaining the early warning status of objects and solidifying trigger, maintain, and cancel rule parameters. Based on the continuous window meeting the conditions, trigger, upgrade, downgrade, cancel, or data verification event records are generated. Establish a mapping table from events to task templates and instantiate tasks according to object type, warning level, dominant type and low confidence mark. The task fields should at least include object location index, event time, deadline, responsible unit, collaborating unit and evidence chain summary. The distribution plan is generated by stratifying and sorting the data according to warning level, scoring order, exposure level, deadline, and resource availability, and under the constraints of personnel, vehicles, equipment, and available time periods. The task execution write-back uses structured fields to record the arrival time, verification results, image evidence index, retest value, measure type and false alarm / missed alarm flags, and archives them in a versioned manner with event number and rule version.

2. The urban geological survey and monitoring risk assessment method as described in claim 1, characterized in that: The classification and coding of risk objects includes determining the object boundary according to one of the following: administrative division unit, road mileage segment, rail transit section mileage segment, underground pipe gallery segment, or hazard point boundary, and solidifying the object boundary with geographic information vector elements; For face-shaped objects, closed surface boundaries are used; for linear objects, centerlines are used and start and end mileage are combined to form segmented boundaries; for point-shaped or small-area objects, point coordinates are used and influence radii are combined to generate influence range boundaries. The object category is determined by the object's main type and written into the enumeration field. At the same time, a unique code is generated for the object and the coding rules are fixed. The coding field includes the administrative region code, object type code, spatial unit number and boundary version information. The effective time and parent-child relationship are recorded by version appending for object splitting, merging and boundary updates.

3. The urban geological survey and monitoring risk assessment method as described in claim 2, characterized in that: The synchronous establishment of attribute, exposure and measurement point mapping ledgers includes registering a set of basic attribute fields and a set of carrier exposure fields for each risk object. The basic attribute fields include geological zoning type, strata and soil type, slope grade, groundwater sensitivity, soft soil or fill thickness grade, karst development grade, historical hidden dangers and verification record number and update time. The exposed fields for the carrier include road grade, rail transit line or station, important pipeline category and grade, list of key units, impact range boundary and buffer parameters; Establish a mapping table between objects and data sources. The mapping table includes data source type, data source identifier, spatial location, association method, association confidence flag, and a list of primary and backup test points. The mapping relationship is managed synchronously with the object version.

4. The urban geological survey and monitoring risk assessment method as described in claim 3, characterized in that: The process of accessing rainfall and various monitoring, survey, and governance data under the ledger mapping includes registering each data source and assigning a unique identifier, as well as registering the acquisition method, sampling cycle, allowed delay, coverage area, coordinate system, data format, and validity period of the results. The data accessed includes rain gauge trigger data, displacement or settlement monitoring data, water level or pore pressure monitoring data, remote sensing deformation results, foundation pit construction monitoring data, geological survey and exploration results, and urban governance ledger data. A unified field dictionary, unit, and timestamp baseline are used for all types of data. The original source identifier and update time are retained. For cases where the same object is associated with multiple data sources of the same type, the priority and switching rules of the main data source are fixed, and the data source switching event and effective time are recorded.

5. The urban geological survey and monitoring risk assessment method as described in claim 4, characterized in that: The window slicing and spatial merging based on the data input set includes unifying the time zone and correcting the time accuracy, slicing the time series according to the preset window length, and using time slice validity management for low-frequency results; Data merging is achieved by performing spatial inclusion, nearest neighbor, buffer intersection, or mileage projection matching based on object boundaries and mapping relationships. Duplicate records with the same timestamp are retained according to quality identifiers or last-written rules, and conflict markers are recorded. Perform unit validation on numerical fields and mark outliers or steps according to verifiable statistical rules; For short-term missing tests, interpolation is performed according to fixed rules and the interpolation ratio is recorded. For long-term missing tests, only the missing test rate is marked and not filled. At the same time, the missing test rate, the proportion of abnormal points, the offline proportion, the delay anomaly and the test point switching are output and bound to the cleaning rule version.

6. A system employing the urban geological survey and monitoring risk assessment method as described in any one of claims 1 to 5, characterized in that: It includes an object database building and encoding module, a data access and preprocessing module, and an evaluation, early warning, and order dispatch closed-loop module; The object database coding module is used to classify urban geological risk objects, solidify GIS boundaries and generate unique codes, and at the same time establish an attribute, exposure and measurement point mapping ledger; The data access preprocessing module is used to map rainfall and monitoring, survey and governance data according to the ledger, unify the field unit time scale and perform window slicing, spatial merging, deduplication and verification, anomaly and missing data processing, and output trigger and response basic sequences and quality labels. The assessment and early warning dispatch closed-loop module is used to calculate scores based on basic sequences and quality labels, and form early warning events according to thresholds and continuous rules. Then, it maps and generates a task list and completes priority sorting, resource scheduling, execution write-back and archiving.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the urban geological survey and monitoring risk assessment method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the urban geological survey and monitoring risk assessment method according to any one of claims 1 to 5.

Images