A system and method for mutual checking and verification of multi-source settlement monitoring data of a face rockfill dam during construction period

Abstract

The application discloses a kind of face panel rock-fill dam construction period multi-source settlement monitoring data mutual checking and verifying system and method, including multi-source data acquisition module, space-time reference uniform module, construction event marking module, multi-level mutual checking and verifying module, credibility evaluation module and abnormal data processing module;Multi-source data acquisition module acquires water pipe type settlement gauge, electromagnetic type settlement tube, array displacement meter, distributed optical fiber, surface deformation monitoring and construction process data;Space-time reference uniform module establishes space and time index;Construction event marking module is marked to filling, stop filling and filling, heavy rainfall, parameter change and construction disturbance;Multi-level mutual checking and verifying module carries out homologous continuity, spatial coordination, heterogenous consistency and construction response rationality checking;Credibility evaluation module determines comprehensive credibility grade;Abnormal data processing module carries out abnormal data classification processing;The application will scattered heterogeneous construction period settlement monitoring data be converted into comparable, credible, traceable data results.

Description

Technical Field

[0001] This invention belongs to the field of safety monitoring and data processing technology for water conservancy and hydropower projects, and in particular relates to a system and method for cross-verification of multi-source settlement monitoring data during the construction period of a panel rockfill dam. Background Technology

[0002] Concrete-faced rockfill dams are a type of dam with rockfill as the main load-bearing structure and concrete panels as the seepage prevention system. They are widely used in large-scale water conservancy and hydropower projects, pumped storage projects, and high dam projects under complex terrain and geological conditions. With increasing dam height and increasingly complex construction environments, settlement deformation during dam construction exhibits significant stage-based, nonlinear, time-varying, and spatial variations. Construction-period settlement not only affects the dam's filling quality and structural deformation coordination but also impacts the subsequent stress on the concrete panels, the safety of the seepage prevention system, and its long-term operational performance.

[0003] Current settlement monitoring during the construction period of rockfill dams typically employs various methods, including water-tube settlement meters, electromagnetic settlement tubes, array displacement gauges, distributed optical fibers, and surface deformation monitoring. These methods differ in their measurement principles, sampling frequencies, spatial representativeness, accuracy levels, and error characteristics. For example, water-tube settlement meters and electromagnetic settlement tubes primarily reflect settlement changes at specific measuring points or along specific lines; array displacement gauges and distributed optical fibers can reflect continuous deformation characteristics within a certain range; and surface deformation monitoring mainly reflects the external deformation response at monitoring points on the dam face or crest. Furthermore, the data from different monitoring methods are not on the same time and spatial scale, making direct comparison and cross-verification difficult.

[0004] In practical engineering, different monitoring data are often collected by different equipment, systems, or personnel, which can easily lead to problems such as inconsistent observation times, inconsistent sampling frequencies, inconsistent representation of measuring point locations, and unclear correspondences between construction stages. Meanwhile, settlement changes during the construction period of a rockfill dam are closely related to construction events such as fill elevation, fill rate, cessation of filling, refilling, changes in dam material, adjustments to compaction parameters, heavy rainfall, and local construction disturbances. If the monitoring data is not correlated with information from the construction process, judging anomalies solely based on the monitoring curves themselves can easily misinterpret the actual construction response as anomalies, and may also introduce erroneous data caused by instrument malfunctions, reading deviations, or construction disturbances into subsequent analyses.

[0005] Existing data processing methods often focus on organizing single monitoring data, removing outliers, or analyzing trends. They typically rely on human experience to determine data usability, and fail to adequately consider the cross-verification relationships between different monitoring sources, the coordination relationships between spatially adjacent monitoring points, and the response relationships between monitoring data and construction events. Such methods struggle to systematically verify multi-source settlement monitoring data during construction and fail to generate traceable and reliable evaluation results, impacting the reliability of subsequent settlement pattern identification, parameter inversion, predictive model training, and construction process control.

[0006] Therefore, it is necessary to propose a multi-source settlement monitoring data cross-verification system and method for the construction period of panel rockfill dams. This system unifies settlement monitoring data from different sources under the same dam body spatial framework, temporal framework, and construction event framework. Through multi-level cross-verification and verification of source continuity, spatial coordination, heterogeneous source consistency, and construction response rationality, the system evaluates the credibility of the monitoring data, classifies and labels the data, and improves the consistency, reliability, and traceability of settlement monitoring data during the construction period. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to provide a system and method for cross-verification of multi-source settlement monitoring data during the construction period of a rockfill dam with a concrete panel. The aim is to overcome the problems in the existing technology, such as the scattered sources of settlement monitoring data during the construction period, inconsistent time references, different spatial representativeness, insufficient correlation of construction events, reliance on human experience for anomaly identification, difficulty in quantifying data credibility, and lack of credible data foundation for subsequent settlement analysis and prediction modeling.

[0008] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A multi-source settlement monitoring data verification system for a rockfill dam construction period includes a multi-source data acquisition module, a spatiotemporal benchmark unification module, a construction event annotation module, a multi-level verification module, a credibility evaluation module, an anomaly data processing module, and a credible settlement dataset output module. The multi-source data acquisition module collects data from water-tube settlement meters, electromagnetic settlement tubes, array displacement meters, distributed fiber optic monitoring, surface deformation monitoring, and construction process data, forming a multi-source raw dataset. The spatiotemporal benchmark unification module performs spatial indexing, temporal indexing, and sampling frequency matching on the multi-source raw dataset based on dam cross-section, station number, elevation, dam material zoning, fill layer, relationship between adjacent measuring points, sampling frequency, and construction period. The system is designed to match the construction period; the construction event labeling module is used to label settlement monitoring data in multi-source raw datasets with construction events based on construction process data, forming a spatiotemporally aligned and event-labeled dataset; the multi-level mutual verification module is used to perform source continuity verification, spatial coordination verification, heterogeneous source consistency verification, and construction response rationality verification on the spatiotemporally aligned and event-labeled datasets, and generate mutual verification results; the credibility evaluation module is used to determine the comprehensive credibility level based on the mutual verification results; the anomaly data processing module is used to classify and process the monitoring data according to the comprehensive credibility level, forming a settlement monitoring dataset with credibility labels; and the credible settlement dataset output module is used to output the settlement monitoring dataset with credibility labels to the engineering application module.

[0009] As a preferred embodiment, the unified spatiotemporal benchmark module includes a spatial index construction unit, a dam cross-section mapping unit, a station-to-elevation mapping unit, a dam material zoning mapping unit, a filling layer mapping unit, an adjacent measuring point relationship construction unit, a time index construction unit, a sampling frequency matching unit, and a construction period matching unit. The spatial index construction unit is used to establish the correspondence between monitoring points and the spatial location of the dam body; the dam cross-section mapping unit is used to establish the correspondence between monitoring points and dam cross-sections; the station-to-elevation mapping unit is used to establish the correspondence between monitoring points and station numbers and elevations; the dam material zoning mapping unit is used to establish the correspondence between monitoring points and dam material zoning zones; the filling layer mapping unit is used to establish the correspondence between monitoring points and construction filling layers; the adjacent measuring point relationship construction unit is used to establish the spatial adjacency relationship between monitoring points on the same cross-section, at the same elevation, in the same dam material zoning zone, or in adjacent filling layers; the time index construction unit is used to establish a unified time index for multi-source monitoring data; the sampling frequency matching unit is used to perform time scale matching for data with different sampling frequencies; and the construction period matching unit is used to match monitoring data with corresponding construction days, shifts, or stages.

[0010] As a preferred embodiment, the construction event labeling module includes a construction event tag set, filling event tags, cessation and refilling event tags, heavy rainfall event tags, dam material or compaction parameter change event tags, and abnormal construction disturbance event tags. The construction event labeling module is used to identify the type, occurrence time, affected area, and duration of construction events based on construction process data, and to associate the corresponding construction event tags with settlement monitoring data within the corresponding time and spatial range.

[0011] As a preferred option, the multi-level mutual verification module includes a same-source continuity verification unit, a spatial coordination verification unit, a heterogeneous source consistency verification unit, and a construction response rationality verification unit. The same-source continuity verification unit is used to verify the continuity of time series within the same monitoring source. The spatial coordination verification unit is used to verify the spatial coordination between adjacent measuring points, measuring points on the same cross section, measuring points at the same elevation, or measuring points in the same dam material zone. The heterogeneous source consistency verification unit is used to verify the consistency of the changing trends of different monitoring sources in the same or similar spatial locations and at the same or similar construction stages. The construction response rationality verification unit is used to determine whether the response relationship between changes in monitoring data and construction events conforms to the settlement evolution law during the construction period.

[0012] As a preferred embodiment, the co-source continuity verification unit includes a cumulative settlement continuity judgment item, a stage settlement change judgment item, and a settlement rate mutation judgment item. The cumulative settlement continuity judgment item is used to determine whether there is a sudden jump, rebound, or drift in the cumulative settlement curve of the same measuring point without construction support. The stage settlement change judgment item is used to determine whether the settlement increment between adjacent construction periods exceeds the preset range. The settlement rate mutation judgment item is used to determine whether the settlement rate undergoes an abnormal mutation in a short period of time.

[0013] As a preferred option, the spatial coordination verification unit includes an adjacent measuring point coordination judgment item, which is used to determine whether the settlement change direction, settlement magnitude and settlement rate between monitoring points of the same dam cross section, the same elevation, the same dam material zone or adjacent filling layer have spatial coordination.

[0014] As a preferred embodiment, the heterogeneous consistency verification unit includes a heterogeneous trend consistency judgment item, which is used to determine whether the direction, magnitude and rate of settlement change are consistent among the data of water pipe setter, electromagnetic setter, array displacement meter, distributed optical fiber monitoring and surface deformation monitoring.

[0015] As a preferred option, the construction response rationality verification unit includes a construction event response judgment item, which is used to determine whether the response relationship between settlement change and filling event label, cessation and refilling event label, heavy rainfall event label, dam material or compaction parameter change event label and abnormal construction disturbance event label is reasonable.

[0016] As a preferred option, the credibility evaluation module includes data integrity evaluation items, temporal continuity evaluation items, spatial coordination evaluation items, heterogeneous consistency evaluation items, construction response rationality evaluation items, and comprehensive credibility level. The credibility evaluation module comprehensively determines the comprehensive credibility level of the monitoring data based on the data integrity evaluation items, temporal continuity evaluation items, spatial coordination evaluation items, heterogeneous consistency evaluation items, and construction response rationality evaluation items.

[0017] As a preferred approach, the abnormal data processing module categorizes monitored data into normal data, suspicious data, and abnormal or missing data based on the overall credibility level. Normal data is directly retained and used for subsequent analysis and modeling. Suspicious data is retained with reduced weight, manually reviewed, or undergoes secondary verification. Abnormal data is corrected or removed. Missing data is marked as missing and is appropriately supplemented when necessary.

[0018] Preferably, the settlement monitoring dataset with credibility labels includes data source labels, construction event labels, credibility level labels, and processing status labels; the data source label is used to record the source of the monitoring data, the construction event label is used to record the corresponding construction event type, the credibility level label is used to record the overall credibility level, and the processing status label is used to record the status of retention, downgrading, correction, removal, missing or completion.

[0019] A method for cross-verification of multi-source settlement monitoring data during the construction period of a panel rockfill dam includes the following steps: S1 collects data from water pipe type settlement meter, electromagnetic settlement pipe, array type displacement meter, distributed optical fiber monitoring, surface deformation monitoring and construction process data through the multi-source data acquisition module to form a multi-source raw dataset; S2 constructs a spatial index for the multi-source original dataset through the spatial index construction unit, dam body cross-section mapping unit, station number and elevation mapping unit, dam material zoning mapping unit, filling layer mapping unit and adjacent measuring point relationship construction unit in the unified spatiotemporal reference module. S3, through the time index construction unit, sampling frequency matching unit and construction period matching unit in the spatiotemporal benchmark unification module, performs time index construction, sampling frequency matching and construction period matching on multi-source raw datasets; S4, through the construction event labeling module, based on the filling event labels, cessation and refilling event labels, heavy rainfall event labels, dam material or compaction parameter change event labels and abnormal construction disturbance event labels in the construction event label set, the multi-source original dataset is labeled with construction events to form a spatiotemporally aligned and event-labeled dataset; S5 performs multi-level mutual verification on the spatiotemporal alignment and event labeling dataset through the same-source continuity verification unit, spatial coordination verification unit, heterogeneous consistency verification unit and construction response rationality verification unit in the multi-level mutual verification verification module, and forms mutual verification verification results; S6. The overall credibility level is determined by the data integrity evaluation item, time continuity evaluation item, spatial coordination evaluation item, heterogeneous consistency evaluation item, and construction response rationality evaluation item in the credibility evaluation module. S7, through the abnormal data processing module, divides the monitoring data into normal data, suspicious data and abnormal or missing data according to the comprehensive credibility level, and performs retention, downweighting, review, correction or removal, missing marking or reasonable completion processing respectively to form a settlement monitoring dataset with credibility labels. S8 outputs a settlement monitoring dataset with credibility labels to the engineering application module through the credible settlement dataset output module. This dataset is used for settlement pattern identification, parameter inversion, prediction model training, differential settlement risk identification, and construction process control.

[0020] The beneficial effects of this invention are as follows: 1. This invention integrates data from water-tube settling meters, electromagnetic settling tubes, array displacement gauges, distributed optical fiber monitoring, surface deformation monitoring, and construction process data into a unified multi-source raw dataset through a multi-source data acquisition module; and establishes the correspondence between monitoring points and dam cross-sections, station numbers, elevations, dam material zones, filling layers, adjacent measuring points, and construction periods through a unified spatiotemporal reference module.

[0021] 2. This invention uses a construction event annotation module to associate construction events such as filling, cessation of filling and refilling, heavy rainfall, changes in dam material or compaction parameters, and abnormal construction disturbances with settlement monitoring data, forming a comparable and interpretable spatiotemporal aligned and event-annotated dataset. A multi-level mutual verification module is used to jointly verify the monitoring data from four aspects: continuity of common source, spatial coordination, consistency of different sources, and rationality of construction response.

[0022] 3. This invention uses a credibility evaluation module and an anomaly data processing module to perform credibility grading and classification of data, ultimately forming a settlement monitoring dataset with credibility labels. This invention transforms multi-source settlement monitoring data during the construction period from a scattered, heterogeneous, and difficult-to-comparate original state into comparable, reliable, and traceable data results. This helps reduce the uncertainty caused by misjudgment of single monitoring data and human experience judgment, improves the quality of settlement monitoring data during the construction period, and provides a reliable data foundation for settlement pattern identification, parameter inversion, prediction model training, differential settlement risk identification, and construction process control. Attached Figure Description

[0023] Figure 1This is a schematic diagram of the overall process of the present invention; Figure 2 This is a schematic diagram illustrating the spatiotemporal unification of multi-source data and the annotation of construction events in this invention; Figure 3 This is a schematic diagram illustrating the multi-level mutual verification, credibility evaluation, and anomaly handling of the present invention; Figure 4 This is a schematic diagram comparing multi-source raw settlement monitoring data in an embodiment of the present invention; Figure 5 This is a schematic diagram of the automatic identification results of abnormal points in an embodiment of the present invention; Figure 6 This is a schematic diagram illustrating the dynamic changes of the heterogeneous trend consistency index in an embodiment of the present invention; Figure 7 This is a schematic diagram comparing the correction effects of the method of the present invention and the traditional method in an embodiment of the present invention. The diagram shows: 1. Multi-source data acquisition module; 2. Multi-source raw dataset; 3. Spatiotemporal benchmark unification module; 4. Construction event annotation module; 5. Spatiotemporal alignment and event annotation dataset; 6. Multi-level mutual verification module; 7. Mutual verification results; 8. Credibility evaluation module; 9. Anomaly data processing module; 10. Settlement monitoring dataset with credibility labels; 11. Credible settlement dataset output module; 12. Engineering application module; 13. Water pipe settlement meter data; 14. Electromagnetic settlement pipe data; 15. Array displacement meter data; 16. Distributed fiber optic monitoring data; 17. Surface deformation monitoring data; 18. Construction process data; 19. Spatial index construction unit; 20. Dam cross-section mapping unit; 21. Station number and elevation mapping unit; 22. Dam material zoning mapping unit; 23. Fill layer mapping unit; 24. Adjacent measuring point relationship construction unit; 25. Time index construction unit; 26. Sampling frequency matching unit; 27. 28. Construction Period Matching Unit; 29. ​​Construction Event Tag Set; 30. Filling Event Tag; 31. Stopped Filling and Refilling Event Tag; 32. Heavy Rainfall Event Tag; 33. Dam Material or Compaction Parameter Change Event Tag; 34. Abnormal Construction Disturbance Event Tag; 35. Same Source Continuity Verification Unit; 36. Spatial Coordination Verification Unit; 37. Different Source Consistency Verification Unit; 38. Construction Response Rationality Verification Unit; 39. Cumulative Settlement Continuity Judgment Item; 40. Stage Settlement Change Judgment Item; 41. Settlement Rate Abrupt Change Judgment Item; 42. Adjacent Measuring Point Coordination Judgment Item; 43. Different Source Trend Consistency Judgment Item; 44. Construction Event Response Judgment Item; 45. Data Integrity Evaluation Item; 46. Temporal Continuity Evaluation Item; 47. Spatial Coordination Evaluation Item; 48. Different Source Consistency Evaluation Item; 49. Construction Response Rationality Evaluation Item; 50. Comprehensive Reliability Level; 51. Normal Data; 52. Suspicious Data; 53. Abnormal or Missing Data. Detailed Implementation

[0024] Example 1: like Figure 1 As shown, a method for cross-verification of multi-source settlement monitoring data during the construction period of a rockfill dam includes a multi-source data acquisition module 1, a multi-source raw dataset 2, a spatiotemporal benchmark unification module 3, a construction event labeling module 4, a spatiotemporal alignment and event labeling dataset 5, a multi-level cross-verification module 6, cross-verification results 7, a credibility evaluation module 8, an anomaly data processing module 9, a settlement monitoring dataset with credibility labels 10, a credibility settlement dataset output module 11, and an engineering application module 12. The multi-source data acquisition module 1 is used to collect data from water-tube settlement gauges 13, electromagnetic settlement tubes 14, array displacement gauges 15, distributed fiber optic monitoring data 16, surface deformation monitoring data 17, and construction process data 18, and aggregates these data to form the multi-source raw dataset 2. The spatiotemporal benchmark unification module 3 is used to perform spatial and temporal indexing processing on the multi-source raw dataset 2, unifying settlement monitoring data from different sources, with different sampling frequencies, and different spatial representations under the same dam body spatial framework and construction time framework. The construction event annotation module 4 is used to annotate settlement monitoring data with construction events based on construction process data 18, forming a spatiotemporal aligned and event-annotated dataset 5. The multi-level mutual verification module 6 is used to verify the spatiotemporal aligned and event-annotated dataset 5 for source continuity, spatial coordination, heterogeneous consistency, and construction response rationality, forming a mutual verification result 7. The credibility evaluation module 8 is used to determine the comprehensive credibility level 49 based on the mutual verification result 7. The anomaly data processing module 9 is used to classify and process the monitoring data according to the comprehensive credibility level 49, forming a settlement monitoring dataset 10 with credibility labels. The credible settlement dataset output module 11 is used to output the credible settlement monitoring dataset 10 to the engineering application module 12 for use in settlement pattern identification, parameter inversion, prediction model training, differential settlement risk identification, and construction process control.

[0025] like Figure 2As shown, the spatiotemporal reference unification module 3 includes a spatial index construction unit 19, a dam cross-section mapping unit 20, a station and elevation mapping unit 21, a dam material zoning mapping unit 22, a filling layer mapping unit 23, an adjacent measuring point relationship construction unit 24, a time index construction unit 25, a sampling frequency matching unit 26, and a construction period matching unit 27. The spatial index construction unit 19 is used to establish a spatial index table for monitoring points, associating each monitoring point with the dam cross-section, station, elevation, dam material zoning, filling layer, and the relationship with adjacent measuring points. The dam cross-section mapping unit 20 is used to determine the dam cross-section to which each monitoring point belongs. The station and elevation mapping unit 21 is used to determine the station and elevation corresponding to each monitoring point. The dam material zoning mapping unit 22 is used to determine the dam material zoning where the monitoring point is located. The filling layer mapping unit 23 is used to match the monitoring point with the corresponding construction filling layer. The adjacent monitoring point relationship construction unit 24 is used to establish the adjacency relationship between monitoring points at the same cross-section, elevation, dam material zone, or adjacent filling layers. The time index construction unit 25 is used to unify the time index of data from different sources. The sampling frequency matching unit 26 is used to match the sampling frequencies of automatically acquired data, manually observed data, and construction process data. The construction period matching unit 27 is used to correlate monitoring data with construction days, shifts, or stages.

[0026] Preferably, the spatial index table of monitoring points generated by the spatial index construction unit 19 includes at least the monitoring point number, monitoring source type, dam cross-section, station number, elevation, dam material zoning, filling layer, upstream and downstream location, and adjacent monitoring point number. The time index generated by the time index construction unit 25 includes at least the observation time, construction date, construction shift, construction stage, and effective observation period. Through the combined effect of the spatial and time indices, the data from the water-tube settling meter 13, electromagnetic settling pipe 14, array displacement meter 15, distributed fiber optic monitoring 16, and surface deformation monitoring 17 are unified under the same analysis unit, providing a basis for subsequent cross-verification and verification.

[0027] like Figure 2As shown, the construction event labeling module 4 includes a construction event label set 28, a filling event label 29, a stop-fill and refill event label 30, a heavy rainfall event label 31, a dam material or compaction parameter change event label 32, and an abnormal construction disturbance event label 33. The construction event label set 28 is used to uniformly manage different types of construction event labels. The filling event label 29 is used to identify the filling operations that occurred within the corresponding time period. The stop-fill and refill event label 30 is used to identify the interruption of construction and the refilling process. The heavy rainfall event label 31 is used to identify rainfall events that may affect the dam's settlement response. The dam material or compaction parameter change event label 32 is used to identify changes in dam material type, dam material zoning, number of compaction passes, compaction thickness, or compaction parameters. The abnormal construction disturbance event label 33 is used to identify construction disturbances, damaged instrument protection, partial rework, or other abnormal construction situations. After being processed by the construction event annotation module 4, the multi-source original dataset 2 is converted into a spatiotemporally aligned and event-annotated dataset 5, so that each settlement monitoring data has spatial location, unified time, data source and construction event information.

[0028] Preferably, the construction event labeling module 4 determines the event type, occurrence time, duration, and affected area of ​​a construction event based on the filling records, work stoppage records, work resumption records, rainfall records, dam material supply records, compaction construction records, construction logs, and abnormal situation records in the construction process data 18, and associates the corresponding tags with the monitoring data corresponding to that time and area. When the same monitoring data corresponds to multiple construction events, the construction event labeling module 4 can set multiple event tags for that monitoring data and record the time range and impact of each event tag.

[0029] like Figure 3 As shown, the multi-level mutual verification module 6 includes a same-source continuity verification unit 34, a spatial coordination verification unit 35, a heterogeneous source consistency verification unit 36, and a construction response rationality verification unit 37. The same-source continuity verification unit 34 is used to verify the continuity of data within the same monitoring source. It includes a cumulative settlement continuity judgment item 38, a stage settlement change judgment item 39, and a settlement rate mutation judgment item 40. The cumulative settlement continuity judgment item 38 is used to determine whether there are sudden jumps, rebounds, or drifts in the cumulative settlement curve of the same measuring point without construction reasons. The stage settlement change judgment item 39 is used to determine whether the settlement increment between adjacent construction periods exceeds a preset range. The settlement rate mutation judgment item 40 is used to determine whether the settlement rate undergoes an abnormal mutation in a short period of time. When the settlement data of a certain measuring point within the same monitoring source shows a sudden jump, obvious rebound, or continuous drift, and no corresponding changes are observed in adjacent time periods or adjacent same-source measuring points, the same-source continuity verification unit 34 marks the data as suspicious same-source data.

[0030] like Figure 3As shown, the spatial coordination verification unit 35 includes an adjacent measuring point coordination judgment item 41. The adjacent measuring point coordination judgment item 41 is used to determine whether the settlement change direction, settlement magnitude, and settlement rate between measuring points at the same cross-section, same elevation, same dam material zone, or adjacent filling layer have spatial coordination based on the adjacent relationship established by the adjacent measuring point relationship construction unit 24. When a measuring point experiences a significant abrupt change, but its adjacent measuring points, measuring points at the same cross-section, or measuring points at the same elevation do not show corresponding changes, and the construction event label cannot explain the abrupt change, the spatial coordination verification unit 35 marks this data as data with insufficient spatial coordination.

[0031] like Figure 3 As shown, the heterogeneous source consistency verification unit 36 ​​includes a heterogeneous source trend consistency judgment item 42. This item 42 compares the settlement change trends between different monitoring sources, including trend comparisons between water-tube settling meter data 13 and electromagnetic settling tube data 14, trend comparisons between array displacement meter data 15 and surface deformation monitoring data 17, and trend comparisons between distributed fiber optic monitoring data 16 and point monitoring data. The heterogeneous source consistency verification does not require the absolute values ​​of different monitoring sources to be completely consistent; rather, it judges whether the direction, stage, rate of change, and construction event response are mutually supportive. When multiple monitoring sources show a consistent change trend at the same cross-section, similar elevation, or the same construction stage, the heterogeneous source consistency verification unit 36 ​​increases the reliability of the corresponding data; when only a single monitoring source shows an abnormal change, while other monitoring sources do not show corresponding changes, the heterogeneous source consistency verification unit 36 ​​decreases the reliability of that data.

[0032] like Figure 3 As shown, the construction response rationality verification unit 37 includes a construction event response judgment item 43. The construction event response judgment item 43 is used to determine whether the correspondence between settlement changes and construction event labels is reasonable. For example, within the time period corresponding to the filling event label 29, an increase in settlement increment or settlement rate in the relevant area can be determined as having a reasonable construction response; within the time period corresponding to the stop-fill and refill event label 30, a gradual decrease in settlement rate during the stop-fill stage and a renewed increase in settlement rate during the refill stage can be determined as having a reasonable construction response; within the time periods corresponding to the heavy rainfall event label 31, the dam material or compaction parameter change event label 32, and the abnormal construction disturbance event label 33, the relevant settlement changes need to be further judged in conjunction with spatial coordination and heterogeneous source consistency. When there is no reasonable correspondence between settlement changes and construction events, or when a sudden settlement change cannot be explained by the construction event, the construction response rationality verification unit 37 marks the data as unreasonable construction response data.

[0033] like Figure 3As shown, the multi-level mutual verification module 6 generates mutual verification result 7 based on the verification results of the same-source continuity verification unit 34, spatial coordination verification unit 35, heterogeneous consistency verification unit 36, and construction response rationality verification unit 37. Mutual verification result 7 includes the score of each verification unit, consistency results, and anomaly indication information, used for subsequent credibility evaluation. Preferably, the score of each verification unit can be determined based on a preset threshold, historical statistical range, range of differences between adjacent measuring points, degree of consistency of heterogeneous trends, and degree of matching of construction event responses.

[0034] like Figure 3 As shown, the credibility evaluation module 8 includes data integrity evaluation item 44, temporal continuity evaluation item 45, spatial coordination evaluation item 46, heterogeneous source consistency evaluation item 47, construction response rationality evaluation item 48, and comprehensive credibility level 49. Data integrity evaluation item 44 is used to evaluate whether there are missing measurements, breakpoints, or invalid values ​​in the monitoring data. Temporal continuity evaluation item 45 is used to evaluate the continuity of the monitoring data over time. Spatial coordination evaluation item 46 is used to evaluate the degree of coordination between the monitoring data and data from adjacent measuring points or adjacent areas. Heterogeneous source consistency evaluation item 47 is used to evaluate the degree of trend consistency between different monitoring sources. Construction response rationality evaluation item 48 is used to evaluate the response relationship between changes in monitoring data and construction events. The credibility evaluation module 8 comprehensively determines the comprehensive credibility level 49 based on the above evaluation items.

[0035] As a preferred approach, the overall credibility level 49 can be divided into high credibility, medium credibility, low credibility, and very low credibility. High credibility data can be directly used as normal data 50 for subsequent analysis and modeling. Medium credibility data can be retained as suspicious data 51 with reduced weight, manually reviewed, or subject to secondary verification. Low credibility and very low credibility data can be classified as abnormal or missing data 52 for correction, removal, missing data marking, or reasonable completion. Specific credibility thresholds can be set according to the engineering monitoring accuracy, historical data stability, and construction stage characteristics. The credibility evaluation module 8 can also set different weights for each evaluation item according to the actual project, so that the rationality of construction response, spatial coordination, or heterogeneous consistency have different evaluation weights at different construction stages.

[0036] like Figure 3As shown, the abnormal data processing module 9 classifies the data according to the comprehensive credibility level 49. Normal data 50 represents data with good integrity, good temporal continuity, high spatial coordination, good consistency across different sources, and reasonable construction response. Normal data 50 is directly retained and used for settlement pattern identification, parameter inversion, and prediction model training. Suspicious data 51 represents data where some verification items are inconsistent but cannot yet be determined to be erroneous. Suspicious data 51 retains its original value, while its weight is reduced, and manual review or secondary verification is performed as needed. Abnormal or missing data 52 represents data with obvious anomalies, missing measurements, or data that cannot be explained by construction events. For abnormal data, corrections can be made based on adjacent measuring points, data from different sources, or historical sequences; data that cannot be corrected is removed. For missing data, a missing marker is applied, and reasonable completion is performed when necessary.

[0037] like Figure 3 As shown, the abnormal data processing module 9 processes the data to form a settlement monitoring dataset 10 with credibility labels. The settlement monitoring dataset 10 with credibility labels includes data source labels, construction event labels, credibility level labels, and processing status labels. Data source labels record that the data comes from water-tube settlement meters, electromagnetic settlement tubes, array displacement gauges, distributed optical fibers, surface deformation monitoring, or construction process data; construction event labels record the type and stage of the corresponding construction event; credibility level labels record the overall credibility level; and processing status labels record the status of retention, downgrading, correction, removal, missing, or completion.

[0038] The credible settlement dataset output module 11 outputs a settlement monitoring dataset 10 with credibility labels. The engineering application module 12 is used to perform settlement pattern identification, parameter inversion, prediction model training, differential settlement risk identification, and construction process control based on the credible settlement monitoring dataset 10. Through these methods, the present invention can transform multi-source, heterogeneous, frequency-inconsistent, and spatially representative construction-period settlement monitoring data into a unified, comparable, credible, and traceable dataset, providing reliable support for the analysis and prediction evaluation of settlement mechanisms during the construction period of panel rockfill dams.

[0039] A method for cross-verification of multi-source settlement monitoring data during the construction period of a panel rockfill dam includes the following steps: S1, collects data from water pipe type settlement meter 13, electromagnetic settlement pipe 14, array type displacement meter 15, distributed optical fiber monitoring 16, surface deformation monitoring 17 and construction process data 18 through multi-source data acquisition module 1 to form multi-source raw dataset 2. S2, through the spatial index construction unit 19, dam cross section mapping unit 20, station number and elevation mapping unit 21, dam material zoning mapping unit 22, filling layer mapping unit 23 and adjacent measuring point relationship construction unit 24 in the spatiotemporal reference unification module 3, spatial index is constructed on the multi-source original dataset 2; S3, through the time index construction unit 25, sampling frequency matching unit 26 and construction period matching unit 27 in the spatiotemporal reference unification module 3, the time index is constructed, sampling frequency is matched and construction period is matched for the multi-source original dataset 2; S4, through the construction event annotation module 4, based on the filling event label 29, the stop-filling and refilling event label 30, the heavy rainfall event label 31, the dam material or compaction parameter change event label 32, and the abnormal construction disturbance event label 33 in the construction event label set 28, the multi-source original dataset 2 is annotated with construction events to form a spatiotemporal aligned and event annotation dataset 5. S5, through the same source continuity verification unit 34, spatial coordination verification unit 35, heterogeneous source consistency verification unit 36 ​​and construction response rationality verification unit 37 in the multi-level mutual verification module 6, multi-level mutual verification is performed on the spatiotemporal alignment and event labeling dataset 5 to form mutual verification result 7. S6. The overall credibility level 49 is determined by the data integrity evaluation item 44, time continuity evaluation item 45, spatial coordination evaluation item 46, heterogeneous consistency evaluation item 47, and construction response rationality evaluation item 48 in the credibility evaluation module 8. S7, the abnormal data processing module 9 divides the monitoring data into normal data 50, suspicious data 51 and abnormal or missing data 52 according to the comprehensive credibility level 49, and performs retention, downweighting, verification, correction or removal, missing marking or reasonable completion processing respectively to form a settlement monitoring dataset 10 with credibility label. S8 outputs the settlement monitoring dataset 10 with credibility labels to the engineering application module 12 through the credible settlement dataset output module 11, which is used for settlement pattern identification, parameter inversion, prediction model training, differential settlement risk identification and construction process control.

[0040] Example 2: This embodiment provides a specific computational example of the algorithm in this method.

[0041] Furthermore, the sampling frequency matching unit is used to perform time scale matching on data with different sampling frequencies, specifically employing a weighted average algorithm based on an adaptive time window. First, a reference time series is determined, using the time series with the highest sampling frequency in the multi-source original dataset as the reference series. The sampling interval of this reference series is denoted as... ,in This represents the baseline sampling interval. For any low-frequency monitoring data sequence to be matched, its original set of sampling time points is obtained, denoted as . ,in Indicates the first Each original sampling time, This refers to the sampling point number. For each target time point on the baseline sequence... ,in Indicates the first A base time point, As the time point number, Construct an adaptive time window centered on , The range is ,in This is the window width adjustment factor. The value range is from 1.0 to 3.0. The average sampling interval of the low-frequency monitoring data sequence. It is obtained by calculating the average of the differences between adjacent original sampling time points. (Within the window) Select all original sampling points And a weighted average of the monitoring values ​​of these sampling points is obtained. Matching value at time Its calculation expression is: ; in, Indicates the time of the original sampling. Settlement monitoring data values, These are the weighting coefficients. Determined by the time distance, its calculation formula is: ; in, The Gaussian kernel width parameter, The value is , Indicates the original sampling time.

[0042] Furthermore, the cumulative settlement continuity judgment item is used to determine whether there are sudden jumps, rebounds, or drifts in the cumulative settlement curve of the same measuring point without construction support. First, a sudden jump detection threshold is set. and rebound detection threshold , and All are determined based on the statistical characteristics of the historical settlement data of this measuring point, and the calculation method is as follows: ; ; in, This represents the standard deviation of the settlement increment sequence at adjacent observation times for this measuring point. This is the jump coefficient. The rebound coefficient is... and The values ​​range from $2.5$ to $4.0$. Calculate the current observation time. Compared to the previous observation time Settlement increment ,in and They represent and The cumulative settlement value at any given time. If If so, it is determined that there is a sudden jump anomaly at that point; if If the point is found to have a rebound anomaly, then the anomaly is determined. For the determination of drift anomalies, a sliding linear fitting method is used, with a length of... Linear fitting was performed on the cumulative settlement curve within the time window. Given a preset number of consecutive observation periods, the fitted slope is obtained. ,like The absolute value is consistently greater than the drift threshold. If no relevant construction event tags are detected, it is determined that the data has drifted. This is the allowable drift rate value set according to the instrument's accuracy.

[0043] Furthermore, the heterogeneous trend consistency judgment item is used to determine whether the direction and magnitude of settlement changes are consistent among different monitoring sources, employing a multi-source data trend consistency index. Quantification was then performed. First, two different monitoring sources were selected for comparison. and Data sequences matched at equal time intervals within the same spatial location and the same time period are denoted as follows: and ,in , To match the number of data points, Indicates the source of monitoring In the The cumulative settlement value at each time point, Indicates the source of monitoring In the The cumulative settlement values ​​at each time point. Calculate the first-order difference sequences for each of the two sequences, denoted as follows: and ,in Define the directional consistency index. The calculation formula is ; in, For a sign function, when the independent variable is greater than Time to take ,equal Time to take less than Time to take Calculate the magnitude correlation index. Pearson correlation coefficient was used. ; in, and Sequences and The mean, and Sequences and The standard deviation of the heterogeneous trend consistency index is calculated. , The range of values ​​is arrive , The closer The more consistent the changing trends of heterogeneous data, the better. Below the preset threshold When the heterogeneous consistency requirement is not met, the result is determined that the heterogeneous consistency requirement is not met. It is a constant between 0.6 and 0.8.

[0044] Furthermore, the construction event response judgment item in the construction response rationality verification unit quantifies the response relationship by establishing a construction response model. Firstly, for the filling loading process corresponding to the filling event label, a correlation model between settlement increment and filling parameters is established, using the response ratio index. Make a judgment. The expression is ; in, This represents the actual observed settlement increment during the period affected by the filling event. This indicates the predicted settlement increment obtained based on the historical settlement statistics of similar dam materials in the region under similar filling conditions. The calculation expression is: ; in, This represents the historical average compression rate within the dam material zone. The thickness of this fill layer, This is the depth correction factor. Depending on the burial depth of the measuring point As it increases, it decreases. , This is the empirical decay factor; if The calculation results are in If the settlement response falls within the specified range, then the response to the corresponding filling event is deemed reasonable. For tolerance coefficient, Take a value between $0.3$ and $0.5$. Next, for the tags indicating the cessation and refilling events, determine whether the settlement rate during the cessation phase decreases over time, and calculate the time decay coefficient after cessation. , ,in The settlement rate at the initial moment of cessation of filling. The residual settlement rate is the rate at which the filling continues until the stabilization point. Greater than the preset attenuation threshold If so, the response is reasonable. Again, regarding the heavy rainfall event label, it is determined whether there is a temporary increase in the settlement rate shortly after rainfall and whether it returns to normal levels within a reasonable time. The reasonableness judgment is based on the post-rainfall increment ratio. ,in To set the settlement increment within a specified period after the rain, This represents the normal settlement increment over the same period of time under conditions of no rainfall. Exceeding If the response falls outside the acceptable range, it is deemed unreasonable. The rainfall response threshold is set based on engineering experience.

[0045] Furthermore, the credibility evaluation module comprehensively determines the overall credibility level of the monitoring data based on data integrity evaluation items, temporal continuity evaluation items, spatial coordination evaluation items, heterogeneous consistency evaluation items, and construction response rationality evaluation items. The overall credibility level is calculated using a weighted scoring method to determine the credibility score. Its expression is ; in, Scoring for data integrity For scoring of time continuity, Scoring for spatial harmony, Scoring for heterogeneous consistency For the construction response rationality score, all scores are normalized to... arrive interval; , , , , For the corresponding weight coefficients, satisfying The weighting coefficients are dynamically adjusted according to the construction stage, as follows: During the peak filling period, the weight of the rationality of the construction response is increased. And spatial coordination weight During the intermittent filling period, increase the weight of time continuity. and heterogeneous consistency weight Increase the heterogeneous consistency weight after heavy rainfall or abnormal disturbances. And spatial coordination weight The overall credibility level is based on the score. Delineation: When Above the high confidence threshold The time is set to high reliability, when Within the medium confidence interval When the time is determined to be reliable, In the low confidence interval When the time is set to low confidence, Below The time setting is considered to be of extremely low confidence, among which , , This is the preset grading threshold.

[0046] Example 3: This embodiment uses a rockfill dam with a concrete face on the upper reservoir of a pumped storage power station as the application object. The dam is 120 meters high and 450 meters long at the top, with a dam construction period of 180 days. During the construction period, four types of settlement monitoring instruments were deployed: three water-tube settlement meters, numbered P1 to P3, buried at elevations of EL.80m, EL.100m, and EL.120m respectively; two electromagnetic settlement tubes, numbered E1 to E2, with five measuring rings per hole and a ring spacing of 2 meters; one distributed optical fiber, laid along the 0+200 section of the dam body from the dam foundation to the dam top, equivalent to 20 sensing sections; and eight surface deformation monitoring prisms, numbered S1 to S8, deployed at different elevations on the slope behind the dam.

[0047] The experimental platform, built based on the method of this invention, includes a multi-source data acquisition module, a spatiotemporal benchmark unification module, a construction event annotation module, a multi-level mutual verification module, a credibility evaluation module, an anomaly data processing module, and a credible settlement dataset output module. Data processing is run in the MATLAB R2022a environment, with a computer configuration of an Intel Core i7-13700H processor and 32GB of memory. The multi-source raw data is uniformly stored in an SQLite database, the spatial index uses a three-dimensional mapping table of station number-elevation-dam material partitioning, and the time index is uniformly set to a daily scale.

[0048] Construction process data is sourced from the construction log system, recording filling days, cessation of filling days, days of heavy rainfall, and days of dam material changes. The construction event tag set includes four categories: filling event tags, cessation and refilling event tags, heavy rainfall event tags, and dam material or compaction parameter change event tags. In the spatiotemporal benchmark unification module, the adaptive time window adjustment coefficient of the sampling frequency matching unit... Set to 2.0, Gaussian kernel width parameter according to Adaptive calculation; abrupt jump coefficient in the cumulative settlement continuity judgment term. Set to 3.0, rebound coefficient Set to 3.0; Heterogeneous Trend Consistency Index threshold Set to 0.6; Allowable deviation coefficient in the judgment of construction response rationality Set to 0.4, rainfall response threshold The overall credibility evaluation weights are set to 0.3. During the peak filling period, the weights are set to w1=0.10, w2=0.10, w3=0.25, w4=0.15, and w5=0.40. During the intermittent filling period, the weights are adjusted to w1=0.15, w2=0.30, w3=0.15, w4=0.25, and w5=0.15. After heavy rainfall or abnormal disturbances, the weights are adjusted to w1=0.10, w2=0.10, w3=0.30, w4=0.35, and w5=0.15. The overall credibility level classification thresholds are QH=0.80, QM=0.55, and QL=0.30.

[0049] To verify the technical effectiveness of the method of this invention, four preset anomalies were artificially injected in the experiment: a 15.0 mm sudden jump was injected into the P2 measuring point of the water-tube sedimentation meter on day 55; a -12.5 mm rebound was injected into the E1-3 measuring ring of the electromagnetic sedimentation tube on day 80; a 0.4 mm daily linear drift was superimposed on the F8 segment of the distributed optical fiber from day 95 to day 115; and the S5 measuring point of the surface deformation monitoring was missing for 6 consecutive days from day 130 to day 135. All raw data were superimposed with Gaussian noise corresponding to the instrument type: the standard deviation of noise for the water-tube sedimentation meter was 1.0 mm, the standard deviation of noise for the electromagnetic sedimentation tube was 1.2 mm, the standard deviation of noise for the distributed optical fiber was 0.8 mm, and the standard deviation of noise for the surface deformation monitoring prism was 1.5 mm.

[0050] The comparison method is the single-source statistical threshold method recommended in the "Technical Specification for Safety Monitoring of Earth and Rockfill Dams" (SL551-2012). This method independently sets a threshold of 3 times the standard deviation for each monitoring instrument. Data exceeding the threshold is directly discarded and filled with linear interpolation. It does not perform cross-verification of multi-source data, nor does it associate construction event information. The results are shown in Table 1 below. Table 1 shows the phased statistical results of each verification index of the method of this invention during the 180-day construction period, including specific data at 10 time nodes.

[0051] Table 1: Statistical analysis of each verification indicator of the method of the present invention in stages;

[0052] Table 2 shows the comparison of the correction effects of the method of the present invention and the traditional single-source statistical threshold method under the same abnormal injection conditions, including the processing results of 10 abnormal points.

[0053] Table 2: Comparison of correction effects between the method of the present invention and the traditional method;

[0054] Based on the data in Table 1, the heterogeneity index of the method of this invention is... During the normal construction phase, the directional consistency remained above 0.85 on days 20, 40, 60, and 150. The overall reliability score (Q) remained above 0.90, and the overall reliability score (Q) remained stable between 0.87 and 0.91, indicating a high reliability level. At the time of the abrupt jump anomaly, i.e., on day 55, the heterogeneous consistency index... The directional consistency dropped sharply to 0.18. The R-squared value dropped to 0.35, the R-squared value deviated from the reasonable range by 2.80, and the Q-squared value fell to 0.25, causing the confidence level to drop directly to extremely low confidence, thus enabling the anomaly to be effectively identified. During the drift development process, from day 95 to 115, the heterogeneous consistency index... As the confidence level gradually decreased from 0.65 to 0.35, Q dropped from 0.62 to 0.38, indicating that the method of this invention can track slowly changing abnormal trends and provide confidence level indications. In the missing data segment, from day 130 to 135, Q remained at a medium confidence level of 0.72 because heterogeneous data could still provide a reference, and did not drop to low or very low confidence, thus preserving data availability.

[0055] Based on the data comparison in Table 2, the traditional single-source statistical threshold method has a correction deviation of 6.30 mm for sudden jump anomalies, while the correction deviation of the method of this invention is only 1.60 mm, improving the correction accuracy by 74.6%. The traditional method has a correction deviation of -5.30 mm for rebound anomalies, while the correction deviation of the method of this invention is only -1.10 mm, improving the correction accuracy by 79.2%. The traditional method failed to identify drift anomalies, and the correction deviation increased from 1.40 mm to 6.20 mm from day 95 to day 115. The method of this invention controls the deviation at each point within the range of 0.50 to 0.70 mm through heterogeneous consistency verification. The traditional method uses linear interpolation for missing data, with a correction deviation of 2.60 to 3.10 mm. The method of this invention utilizes the spatial coordination of adjacent measuring points and the weighted averaging of heterogeneous data, with a correction deviation of only 0.60 to 0.70 mm. The root mean square error (RMSE) of the 10 outliers was calculated. The traditional method yielded a RMSE of 4.23 mm, while the method of this invention yielded a RMSE of 0.84 mm, resulting in a reduction of 80.1%. This demonstrates that the method of this invention is significantly superior to existing technologies in terms of the comprehensiveness of outlier identification, the accuracy of correction, and the maintenance of data continuity.

[0056] like Figure 4The image shows a comparison between raw observation data and actual settlement from four types of settlement monitoring instruments during the 150-day construction period of the rockfill dam. Traditional methods typically rely on a single monitoring source for trend judgment. When a particular instrument displays abnormal data due to environmental interference or equipment malfunction, human experience is insufficient to determine the trend. Figure 4 The actual construction response and instrument error were directly identified. The water-tube settling meter showed a 15 mm jump on day 55, the electromagnetic settling tube showed a 12.5 mm rebound on day 80, the distributed optical fiber showed continuous drift from day 95 to day 115, and surface deformation monitoring showed missing data segments from day 130 to day 135. These anomalies were... Figure 4 When mixed with normal data, traditional single-source threshold methods may misjudge normal accelerated settlement during the filling period as abnormal, or mistake instrument drift for a true construction response. This invention, through multi-source data cross-verification, can... Figure 4 Based on this, a dual spatial and temporal index is established to provide a data foundation for subsequent anomaly identification. Figure 4 This study reveals the complexity and heterogeneity of multi-source monitoring data during the construction period, proving that it is difficult to achieve systematic verification based on a single data source or human experience alone, thus highlighting the necessity of multi-source fusion and mutual verification in this invention.

[0057] like Figure 5 As shown, the automatic identification results of the present invention's method for four typical anomalies are presented. Traditional methods rely on fixed thresholds or human experience to judge anomalies. When the settlement rate fluctuates during construction due to changes in the filling rhythm, fixed thresholds can easily misjudge normal accelerated settlement as a sudden jump, or miss slow drift. The present invention's method performs joint verification based on four dimensions: continuity of origin, spatial coordination, consistency of different origins, and rationality of construction response, which can distinguish between real construction response and instrument anomalies. Figure 5 In the data analysis, the red marker indicating a sudden jump in the water-tube settling meter on day 55 and the green marker indicating a rebound in the electromagnetic settling pipe on day 80 were identified because adjacent measuring points did not show synchronous changes and the heterogeneous trend consistency index was observed. Below the 0.6 threshold. Distributed fiber optic drift in days 95-115 was detected by blue markers because its direction of change did not match the construction event label and lacked spatial consistency support. Missing surface deformation monitoring values ​​in days 130-135 were marked as missing data by magenta markers. Traditional methods struggle to handle the four types of anomalies—sudden jumps, rebounds, drifts, and missing values—simultaneously, while the method of this invention achieves classification, identification, and differentiated processing, which helps reduce the false negative and false positive rates.

[0058] like Figure 6 As shown, the heterogeneous trend consistency index in the method of this invention is illustrated. During the 150-day construction period, traditional methods only apply statistical thresholds to a single data source, failing to quantify the degree of trend matching between different monitoring sources. Therefore, when a certain type of instrument exhibits slow drift, traditional methods are difficult to detect until the deviation accumulates to a significant level. The λ index proposed in this invention integrates directional consistency. correlation with amplitude It can quantify the degree of collaboration between multiple data sources in real time; Figure 6 middle, The index remained at a high level between 0.8 and 1.0 for most of the period, indicating that the trends of the water-tube settling meter and the electromagnetic settling pipe were highly consistent under normal operating conditions. When the anomalous jump occurred on day 55, The index plummeted below 0.2, falling below the preset threshold of 0.6 (the red dotted line), triggering an anomaly alarm. During the drift period from day 95 to 115... An index consistently below 0.5 indicates a significant decrease in heterogeneous consistency. An abnormal rebound occurred on day 80. The index also showed a significant decline. Traditional methods cannot provide this real-time, quantitative evaluation of heterogeneous consistency, while this invention... The index enables cross-verification of data from multiple sources, and can promptly detect abnormal changes in a single sensor, providing an objective basis for subsequent credibility grading and anomaly handling.

[0059] like Figure 7 As shown, the correction effects of the multi-source mutual verification method of this invention and the traditional single-source threshold method on the data of water pipe sedimentation meters are compared. The traditional method uses a fixed standard deviation threshold. Outlier removal and linear interpolation resulted in some normal data being misjudged as anomalies and removed during the peak settlement period from day 40 to 45 due to the natural increase in settlement rate. This caused the corrected curve to deviate from the actual construction response. At the abrupt jump anomaly on day 55, although the traditional method removed the outlier, linear interpolation failed to reflect the true physical process, leading to a false smoothing of the corrected curve. During the drift segment from day 95 to 115, the traditional method was ineffective because the drift amplitude did not exceed [a certain threshold]. If the threshold is not met, the drift error is retained for subsequent analysis. The method of this invention utilizes multi-source mutual verification and a heterogeneous consistency index. Anomalies were identified and replaced with weighted averages from other normal sources. The true settlement trend was accurately restored at the abrupt jump on day 55. The drift effect was eliminated through correction using heterogeneous data during the drift segment from day 95 to 115. Precise correction was also achieved at the rebound on day 80. Quantitative calculations show that the root mean square error (RMSE) of the traditional method is 4.23 mm, while the RMSE of the method of this invention is 0.84 mm, a reduction of 80.1%. Figure 7 This intuitively demonstrates that the method of the present invention has good technical effects in terms of anomaly identification accuracy and correction reliability.

Claims

1. A system for cross-verification and verification of multi-source settlement monitoring data during the construction period of a rockfill dam with concrete panel, characterized in that, The dataset includes a multi-source data acquisition module (1) for acquiring data from water-tube settling meters (13), electromagnetic settling tubes (14), array displacement meters (15), distributed optical fiber monitoring (16), surface deformation monitoring (17), and construction process data (18), forming a multi-source raw dataset (2); a spatiotemporal benchmark unification module (3) for constructing spatial and temporal indexes, matching sampling frequencies, and matching construction periods on the multi-source raw dataset (2); and a construction event annotation module (4) for annotating settlement monitoring data in the multi-source raw dataset (2) with construction events based on construction process data (18), forming a spatiotemporal aligned and event-annotated dataset ( ). 5) The multi-level mutual verification module (6) is used to perform source continuity verification, spatial coordination verification, heterogeneous consistency verification and construction response rationality verification on the spatiotemporal alignment and event labeling dataset (5), and form mutual verification results (7); the credibility evaluation module (8) is used to determine the comprehensive credibility level of the monitoring data (49) based on the mutual verification results (7); the abnormal data processing module (9) is used to classify the monitoring data according to the comprehensive credibility level (49) and form a settlement monitoring dataset (10) with credibility labels; the credible settlement dataset output module (11) is used to output the settlement monitoring dataset (10) with credibility labels to the engineering application module (12).

2. The multi-source settlement monitoring data mutual verification system for panel rockfill dams during construction, as described in claim 1, is characterized in that... The unified spatiotemporal reference module (3) includes a spatial index construction unit (19), a dam cross-section mapping unit (20), a station and elevation mapping unit (21), a dam material zoning mapping unit (22), a filling layer mapping unit (23), an adjacent measuring point relationship construction unit (24), a time index construction unit (25), a sampling frequency matching unit (26), and a construction period matching unit (27). The spatial index construction unit (19) is used to establish the correspondence between the monitoring points and the spatial location of the dam. The dam cross-section mapping unit (20) is used to establish the correspondence between the monitoring points and the dam cross-section. The station and elevation mapping unit (21) is used to establish the correspondence between the monitoring points and the station and elevation. The corresponding relationship between monitoring points and dam material zoning mapping unit (22) is used to establish the corresponding relationship between monitoring points and dam material zoning; the filling layer mapping unit (23) is used to establish the corresponding relationship between monitoring points and construction filling layers; the adjacent measuring point relationship construction unit (24) is used to establish the spatial adjacency relationship between monitoring points of the same cross section, same elevation, same dam material zoning or adjacent filling layers; the time index construction unit (25) is used to establish a unified time index for multi-source monitoring data; the sampling frequency matching unit (26) is used to perform time scale matching for data with different sampling frequencies; the construction period matching unit (27) is used to match the monitoring data with the corresponding construction day, construction shift or construction stage.

3. The cross-verification system for multi-source settlement monitoring data during the construction period of a rockfill dam with a concrete panel, as described in claim 1, is characterized in that... The construction event labeling module (4) includes a construction event label set (28), filling event label (29), cessation and refilling event label (30), heavy rainfall event label (31), dam material or compaction parameter change event label (32), and abnormal construction disturbance event label (33). The construction event labeling module (4) is used to identify the type, time of occurrence, affected area and duration of construction events based on the construction process data (18), and associate the corresponding construction event label with the settlement monitoring data within the corresponding time and space range.

4. The multi-source settlement monitoring data mutual verification system for a rockfill dam during construction, as described in claim 1, is characterized in that... The multi-level mutual verification module (6) includes a same-source continuity verification unit (34), a spatial coordination verification unit (35), a heterogeneous consistency verification unit (36), and a construction response rationality verification unit (37). The same-source continuity verification unit (34) is used to verify the continuity of the time series within the same monitoring source. The spatial coordination verification unit (35) is used to verify the spatial coordination between adjacent measuring points, measuring points on the same cross section, measuring points at the same elevation, or measuring points in the same dam material zone. The heterogeneous consistency verification unit (36) is used to verify the consistency of the changing trends of different monitoring sources in the same or similar spatial locations and the same or similar construction stages. The construction response rationality verification unit (37) is used to determine whether the response relationship between the changes in monitoring data and construction events conforms to the settlement evolution law during the construction period.

5. The multi-source settlement monitoring data mutual verification system for a rockfill dam during construction, as described in claim 4, is characterized in that... The co-source continuity verification unit (34) includes a cumulative settlement continuity judgment item (38), a stage settlement change judgment item (39), and a settlement rate mutation judgment item (40). The cumulative settlement continuity judgment item (38) is used to determine whether there is a sudden jump, rebound, or drift in the cumulative settlement curve of the same measuring point without construction reasons. The stage settlement change judgment item (39) is used to determine whether the settlement increment between adjacent construction periods exceeds the preset range. The settlement rate mutation judgment item (40) is used to determine whether the settlement rate has an abnormal mutation in a short period of time.

6. The multi-source settlement monitoring data mutual verification system for a rockfill dam during construction, as described in claim 4, is characterized in that... The spatial coordination verification unit (35) includes an adjacent measuring point coordination judgment item (41), which is used to determine whether the settlement change direction, settlement magnitude and settlement rate between monitoring points of the same dam cross section, the same elevation, the same dam material zone or adjacent filling layer have spatial coordination.

7. A cross-verification system for multi-source settlement monitoring data during the construction period of a rockfill dam with a concrete panel, as described in claim 4, is characterized in that... The heterogeneous consistency verification unit (36) includes a heterogeneous trend consistency judgment item (42), which is used to judge whether the direction, magnitude and rate of settlement change are consistent among the water pipe settlement meter data (13), electromagnetic settlement pipe data (14), array displacement meter data (15), distributed optical fiber monitoring data (16) and surface deformation monitoring data (17); the construction response rationality verification unit (37) includes a construction event response judgment item (43), which is used to judge whether the response relationship between settlement change and filling event label (29), stop filling and refilling event label (30), heavy rainfall event label (31), dam material or compaction parameter change event label (32) and abnormal construction disturbance event label (33) is reasonable.

8. The multi-source settlement monitoring data mutual verification system for a rockfill dam during construction, as described in claim 1, is characterized in that... The credibility evaluation module (8) includes data integrity evaluation items (44), time continuity evaluation items (45), spatial coordination evaluation items (46), heterogeneous consistency evaluation items (47), construction response rationality evaluation items (48), and comprehensive credibility level (49). The credibility evaluation module (8) comprehensively determines the comprehensive credibility level (49) of the monitoring data based on the data integrity evaluation items (44), time continuity evaluation items (45), spatial coordination evaluation items (46), heterogeneous consistency evaluation items (47), and construction response rationality evaluation items (48). The abnormal data processing module (9) divides the monitoring data into normal data (50), suspicious data (51), and abnormal or missing data (52) based on the comprehensive credibility level (49). Normal data (50) is directly retained and used for subsequent analysis and modeling. Suspicious data (51) is retained with reduced weight, manually reviewed, or checked twice. In abnormal or missing data (52), abnormal data is corrected or removed, and missing data is marked as missing or reasonably completed.

9. A cross-verification system for multi-source settlement monitoring data during the construction period of a rockfill dam with a concrete panel, as described in claim 1, is characterized in that... The settlement monitoring dataset (10) with credibility labels includes data source labels, construction event labels, credibility level labels and processing status labels; the data source label is used to record the source of the monitoring data; the construction event label is used to record the corresponding construction event type; the credibility level label is used to record the overall credibility level; the processing status label is used to record the status of retention, downgrading, correction, removal, missing or completion.

10. A method for cross-verification of multi-source settlement monitoring data during the construction period of a rockfill dam with a concrete panel, as described in any one of claims 1-9, characterized in that, Includes the following steps: S1, collects data from water pipe type settlement meter (13), electromagnetic settlement pipe (14), array type displacement meter (15), distributed optical fiber monitoring (16), surface deformation monitoring (17) and construction process data (18) through multi-source data acquisition module (1) to form multi-source raw dataset (2). S2, through the unified spatiotemporal reference module (3), performs dam body cross-section mapping, station number and elevation mapping, dam material zoning mapping, filling layer mapping, adjacent measuring point relationship construction, time index construction, sampling frequency matching and construction period matching on the multi-source original dataset (2); S3, the construction event labeling module (4) identifies the type, time of occurrence, area of ​​impact and duration of construction events based on the construction process data (18), and labels the settlement monitoring data in the multi-source original dataset (2) with construction events to form a spatiotemporal alignment and event labeling dataset (5). S4, through the multi-level mutual verification module (6), the spatiotemporal alignment and event labeling dataset (5) is checked for the continuity of the same source, the spatial coordination, the consistency of different sources and the rationality of construction response, and the mutual verification result (7) is formed. S5, the overall credibility level (49) is determined by the credibility evaluation module (8) based on the mutual verification results (7); S6, the abnormal data processing module (9) classifies the monitoring data as normal, suspicious, abnormal or missing according to the comprehensive credibility level (49); S7. Based on the processing results of the abnormal data processing module (9), a settlement monitoring dataset with credibility labels is formed (10). S8, the settlement monitoring dataset (10) with credibility labels is output to the engineering application module (12) through the credible settlement dataset output module (11).

Images