Method and system for processing monitoring data of deep trench base arching

Abstract

The application discloses a deep trench base heave monitoring data processing method and system, relates to the technical field of data processing, and obtains a monitoring point base normal displacement value, a sampling time and a plane coordinate, generates a window displacement median value and a window displacement rate median value; divides historical data into a calibration historical section and a verification historical section, screens historical windows according to the number of missing values, the number of direction reversals and the window end time displacement median absolute deviation, and generates a displacement history boundary, a displacement rate history boundary and a reference duration; extracts a positive continuous change section, combines the overlapping time periods of spatial adjacent monitoring points into a spatiotemporal connected same-direction overlapping section, generates a comprehensive heave discrimination value in combination with the out-of-bounds degree and the overlapping feature, and outputs an early warning result when the discrimination condition is met. The application solves the problems that the discrimination benchmark is unstable due to the difference of historical monitoring data and that it is difficult to identify continuous spatial correlation deformation caused by single-point short-time change, and improves the reliability of the base heave monitoring result.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a method and system for processing monitoring data on the arching of deep roadbed foundations. Background Technology

[0002] After the excavation of a deep road cut is completed, the original stress state of the underlying rock and soil changes. Under the influence of factors such as unloading rebound, rock mass water absorption and expansion, groundwater changes, deformation of local weak structures, and construction disturbance, the foundation may experience continuous displacement along the foundation normal direction. Foundation deformation monitoring typically involves setting up multiple monitoring points to continuously collect displacement changes at different locations, and judging whether there is a deformation process that requires attention based on the changes in monitoring data.

[0003] Existing methods for processing deep road cut foundation deformation monitoring data typically use fixed displacement thresholds or fixed displacement rate thresholds as discrimination criteria. For example, when the cumulative displacement of a monitoring point exceeds a preset value, or the displacement rate within a certain time range exceeds a preset rate of change, an abnormal monitoring result is output. However, different areas of deep road cuts are affected by factors such as lithological conditions, excavation depth, groundwater conditions, the degree of construction disturbance, and the monitoring environment, and their normal displacement variation range may differ. When using a uniform fixed threshold for judgment, for areas with large displacement fluctuations, data within the normal variation range is easily judged as abnormal; for areas with long-term slow changes, the small amplitude of a single change may not reflect the continuous change process in a timely manner.

[0004] Using historical monitoring data to generate regional discrimination boundaries can improve the matching degree between discrimination conditions and the historical change range of the target monitoring area. However, historical monitoring data is not always of the same quality. For example, data may be missing in some time periods due to communication interruptions, displacement direction may change repeatedly due to construction or environmental changes, and some monitoring points may exhibit displacement changes different from surrounding monitoring points due to local installation conditions or environmental factors. If all historical data is directly used to generate discrimination boundaries, missing data, data with frequent directional changes, or data with large local dispersion may alter the historical displacement distribution range, making the generated discrimination boundaries unable to fully reflect the normal change characteristics of the target monitoring area.

[0005] On the other hand, changes related to basal arching typically exhibit a certain degree of persistence and spatial correlation. A large displacement change at a single monitoring point within a short period may be caused by local disturbances, equipment fluctuations, or abnormal data transmission, and does not directly indicate a continuous deformation process in the monitored area. For example, if a monitoring point experiences a sudden change within a single sampling period, while adjacent monitoring points do not show changes in the same direction, this change differs from a situation where multiple monitoring points are undergoing continuous changes together. Therefore, judging solely based on whether a single monitoring point exceeds historical boundaries is insufficient to fully utilize the temporal and spatial correlation information between multiple monitoring points. Summary of the Invention

[0006] This invention provides a method and system for processing monitoring data of arching on the foundation of deep road cuts, which solves the problems of the influence of differences in the quality of historical monitoring data on the establishment of the discrimination benchmark in the existing deep road cut foundation deformation monitoring, and the difficulty of reflecting continuous spatial correlation changes in short-term changes of a single monitoring point.

[0007] A method for processing monitoring data on the arching of a deep road cut foundation includes: acquiring the foundation normal displacement value, sampling time, and plane coordinates of each monitoring point; generating the median displacement value and median displacement rate value of a window according to a time window; setting a calibration historical section and a verification historical section before the current monitoring cycle; filtering candidate historical windows within the two historical sections stepwise according to the number of missing values, the number of direction flips, and the absolute deviation of the median displacement at the end of the window; generating historical boundaries for displacement, historical boundaries for displacement rate, and a reference duration based on the calibration historical windows; extracting the positive continuous change segments that reach the reference duration for each monitoring point; and matching the original values ​​corresponding to the positive continuous change segments of spatially adjacent monitoring points. The initial overlapping time period is merged into a spatiotemporally connected overlapping segment according to the temporal and spatial connectivity relationships. A comprehensive overlapping value is generated based on the participation degree of monitoring points, the degree of temporal overlap, and the degree of migration of the coordinate center of the active monitoring points in the spatiotemporally connected overlapping segment. A boundary value for the continuously changing segment is generated based on the degree of boundary exceedance of the median value of the window displacement and the median value of the window displacement rate. A comprehensive arching discrimination value is obtained based on the comprehensive overlapping value and the boundary value for the continuously changing segment. A comprehensive arching discrimination boundary is generated based on the verification history window. When the current comprehensive arching discrimination value exceeds the comprehensive arching discrimination boundary and the duration of the corresponding spatiotemporally connected overlapping segment reaches the reference duration, the base arching warning result is output.

[0008] Optionally, the median value of the window displacement and the median value of the window displacement rate are generated according to the following steps: The base normal displacement values ​​of each valid monitoring point at the end of the time window are arranged in ascending order, and the median is used as the median value of the window displacement; for each monitoring point that has a valid base normal displacement value at both the beginning and end of the time window, the base normal displacement rate of the corresponding monitoring point is determined based on the base normal displacement value at the end, the base normal displacement value at the beginning, and the duration of the time window; the base normal displacement rates of each monitoring point within the same time window are arranged in ascending order, and the median is used as the median value of the window displacement rate; the median values ​​of the window displacement and the median values ​​of the window displacement rate are arranged according to the order of the end of the time window; wherein, when the number of data involved in the sorting is even, the median is the average of the two data points in the middle of the sorted position.

[0009] Optionally, the candidate historical windows are filtered according to the following steps: the number of missing values ​​is determined based on the difference between the expected number of original effective displacement values ​​and the actual number of original effective displacement values ​​in the candidate historical window; a window displacement sequence is formed based on the median displacement value at each sampling time within the candidate historical window, and after excluding zero differences between adjacent median displacement values, the number of conversions between positive and negative differences is counted to obtain the number of direction reversals; the absolute difference between each effective base normal displacement value at the end of the candidate historical window and the median displacement at that end time is determined, and the median of all absolute differences is used as the absolute deviation of the median displacement at the end of the window; the candidate historical windows are arranged in ascending order according to the number of missing values, the number of direction reversals, and the absolute deviation of the median displacement at the end of the window, and the first half of the candidate historical windows are retained after each sorting to obtain the calibration historical window and the verification historical window respectively; when the number of candidate historical windows participating in the sorting is odd, the number of retentions is determined according to the rounding up result; when multiple candidate historical windows located at the retention boundary have the same sorting data, the candidate historical windows whose end time is closer to the start time of the current monitoring cycle are retained first, and if they are still the same, the retention order is determined according to the ascending order of the candidate historical window numbers.

[0010] Optionally, the displacement history boundary, displacement rate history boundary, and reference duration are generated according to the following steps: The median values ​​of the window displacements of all calibration history windows are arranged in ascending order. The minimum and maximum values ​​are used as the initial lower and upper boundaries of the displacement, respectively. The median difference between adjacent window displacement median values ​​with different values ​​is used as the displacement boundary extension amount. The initial lower and upper boundaries of the displacement are extended downwards and upwards, respectively, to obtain the displacement history lower and upper boundaries. Following the same process, the displacement rate history lower and upper boundaries are generated based on the median values ​​of the window displacement rates of all calibration history windows. When there are no adjacent data with different values, the smallest resolvable unit of the corresponding data is used as the boundary extension amount. All calibration history windows are arranged according to their end times. Adjacent calibration history windows with a sliding step of one time window between their end times are connected into a continuous calibration history window sequence. For each monitoring point within each continuous calibration history window sequence, the time interval covered by the continuous positive displacement difference between the end times of adjacent windows is merged into a historical candidate positive continuous change segment. The median of the duration of all historical candidate positive continuous change segments is used as the reference duration.

[0011] Optionally, the positive continuous change segment is generated according to the following steps: A Deloitte triangulation is established based on the planar coordinates of each monitoring point; two monitoring points connected by the same edge of the triangulation are identified as spatially adjacent monitoring points; when monitoring points cannot form a Deloitte triangulation, they are sorted sequentially according to their coordinates in the direction with the larger planar span, their coordinates in the other direction, and their monitoring point number; two monitoring points with adjacent positions are identified as spatially adjacent monitoring points; for each monitoring point, the displacement difference is calculated only when the time interval between the end times of adjacent time windows is equal to a time window sliding step, and both end times have effective base normal displacement values; the time interval corresponding to the positive displacement difference is recorded as the positive change interval; time-continuous positive change intervals are merged into candidate positive continuous change segments; negative displacement differences, zero displacement differences, missing values, or time gaps interrupt candidate positive continuous change segments; candidate positive continuous change segments with a duration reaching the reference duration are identified as positive continuous change segments.

[0012] Optionally, the spatiotemporally connected overlapping segments and the combined overlapping value are generated according to the following steps: For each pair of spatially adjacent monitoring points, the later start time of the two positive continuous change segments is taken as the start time of the original overlapping time period, and the earlier end time is taken as the end time of the original overlapping time period, and the original overlapping time period is generated when the start time is earlier than the end time; the original overlapping time periods that overlap in time or are adjacent at the boundary, and whose corresponding monitoring point pairs have common monitoring points or can form a continuous adjacent path through other spatially adjacent monitoring point pairs, are merged into spatiotemporally connected overlapping segments; from the monitoring points that have effective base normal displacement values ​​at the end time of all windows of the spatiotemporally connected overlapping segments, the number of different monitoring points in the positive continuous change segments within the overlapping segment is counted to determine the number of monitoring points. The participation rate of positive change monitoring points is determined by: the duration of the longest original overlapping time period and the duration of the spatiotemporally connected overlapping segment in the same direction, with the smaller of the participation rate of positive change monitoring points and the time overlap rate used as the baseline value for the same direction overlap; the average planar coordinate position of monitoring points within the positive continuous change segment in each judgment time interval is determined according to the time sequence to obtain the coordinate center of the active monitoring point; the distance between the coordinate centers of adjacent active monitoring points with continuous time is accumulated to obtain the cumulative migration length of the active center; the migration ratio of the active center is determined based on the cumulative migration length of the active center and the maximum planar distance between monitoring points, with any migration ratio exceeding 1 recorded as 1, and the larger of the baseline value for the same direction overlap and the migration ratio of the active center used as the comprehensive value for the same direction overlap.

[0013] Optionally, the comprehensive arching discrimination value is generated according to the following steps: when the median value of the window displacement does not exceed the historical boundary of the displacement, the displacement over-limit ratio is set to 0; when it exceeds the historical boundary of the displacement, the displacement over-limit ratio is determined based on the displacement amount exceeding the historical boundary of the displacement and the distance between the upper and lower boundaries of the displacement history; when the median value of the window displacement rate does not exceed the historical boundary of the displacement rate, the displacement rate over-limit ratio is set to 0; when it exceeds the historical boundary of the displacement rate, the displacement rate over-limit ratio is determined based on the rate amount exceeding the historical boundary of the displacement rate and the distance between the upper and lower boundaries of the displacement rate history; the larger value between the displacement over-limit ratio and the displacement rate over-limit ratio corresponding to the same time window is used as the window over-limit ratio; the median of the over-limit ratios of all windows covered by a spatiotemporally connected overlapping segment in the same direction is used as the over-limit value of the continuously changing segment; the over-limit value of the continuously changing segment is multiplied by the comprehensive value of the overlapping segment in the same direction to obtain the comprehensive arching discrimination value.

[0014] Optionally, the composite arching discrimination boundary is generated according to the following steps: All verification history windows are arranged according to their end times; adjacent verification history windows with a sliding step of one time window between their end times are connected to form a continuous verification history window sequence; the displacement history boundary, displacement rate history boundary, and reference duration generated by the calibration history windows are used to process each continuous verification history window sequence to obtain the historical composite arching discrimination value; when a continuous verification history window sequence does not form a spatiotemporally connected overlapping segment, the historical composite arching discrimination value corresponding to that continuous verification history window sequence is recorded as... 0; When multiple spatiotemporally connected overlapping segments are formed, each spatiotemporally connected overlapping segment generates a historical comprehensive arch discrimination value; All historical comprehensive arch discrimination values ​​are arranged in ascending order of value, with the maximum value as the initial discrimination boundary, and the median of the difference between adjacent historical comprehensive arch discrimination values ​​with different values ​​as the discrimination boundary expansion amount. The initial discrimination boundary is increased by the discrimination boundary expansion amount to obtain the comprehensive arch discrimination boundary; When there are no adjacent historical comprehensive arch discrimination values ​​with different values, the minimum recording unit of the comprehensive discrimination value set in advance is used as the discrimination boundary expansion amount.

[0015] Optionally, each time a new time window ends and valid monitoring data is obtained, each positive continuous change segment and each spatiotemporally connected overlapping segment is updated; for positive continuous change segments and spatiotemporally connected overlapping segments that have not yet ended, the end time of the latest time window is used as the provisional end time; when there are multiple spatially disconnected spatiotemporally connected overlapping segments within the same time range, the comprehensive arching discrimination value of each spatiotemporally connected overlapping segment is calculated separately; when any current comprehensive arching discrimination value exceeds the comprehensive arching discrimination boundary, and the duration of the corresponding spatiotemporally connected overlapping segment reaches the reference duration, the base arching early warning result containing the monitoring area, the monitoring points involved, the start time, the end time, the comprehensive arching discrimination value, and the comprehensive arching discrimination boundary is output.

[0016] A data processing system for monitoring the arching of deep road cut foundations is also provided, including a data conversion module, a historical calibration module, a spatiotemporal overlap processing module, and a discrimination output module. The data conversion module acquires the foundation normal displacement value, sampling time, and plane coordinates of each monitoring point, and generates the median displacement value and median displacement rate value of each window according to the time window. The historical calibration module divides the historical segments before the current monitoring cycle into calibration historical segments and verification historical segments, and performs step-by-step filtering on candidate historical windows in the two historical segments based on the number of missing windows, the number of direction flips, and the absolute deviation of the median displacement at the end of the window. Based on the calibration historical windows, it generates historical displacement boundaries, historical displacement rate boundaries, and a reference duration. The spatiotemporal overlap processing module generates the positive continuous variation of each monitoring point. The system is divided into three modules: a data conversion module, a historical calibration module, a spatiotemporal overlap module, and a discrimination output module. The data conversion module, the historical calibration module, the spatiotemporal overlap processing module, and the discrimination output module are implemented by the processor executing program instructions in the memory. The data conversion module, the historical calibration module, the spatiotemporal overlap processing module, and the discrimination output module are all implemented by the processor executing program instructions in the memory.

[0017] The beneficial effects of this invention are reflected in:

[0018] This invention divides historical data prior to the current monitoring cycle into calibration and verification historical segments, ensuring that the historical boundary generation and historical discrimination boundary verification processes utilize independent data sources. Candidate historical windows within the calibration historical segment are progressively filtered based on the number of missing records, the number of direction reversals, and the median absolute deviation of displacement at the end of the window. This ensures that the data used to generate displacement historical boundaries, displacement rate historical boundaries, and reference duration have good sampling completeness, stability of change, and consistency in monitoring point distribution. Compared to directly using all historical data to generate boundaries, this invention reduces the impact of missing records, frequent direction changes, and locally discrete data on the historical discrimination benchmark, making the generated discrimination range more consistent with the historical change characteristics of the target monitoring area itself.

[0019] This invention generates a reference duration length based on historical candidate positive continuous change segments in a continuous calibration historical window sequence, and generates positive continuous change segments based on displacement differences in current monitoring data. By setting a continuous change reference scale based on historical data, local changes generated by a single monitoring point within a short period of time will not directly enter the subsequent discrimination process. Simultaneously, this invention forms spatiotemporally connected overlapping segments based on the original overlapping time periods between spatially adjacent monitoring points, combined with temporal and spatial connectivity relationships. This allows changes at spatially unrelated monitoring points to be processed separately, ensuring that the discrimination results maintain a correspondence with the spatial location of the monitoring area.

[0020] This invention further generates a continuous change segment over-limit value based on the degree of over-limitation of the median value of window displacement and the median value of window displacement rate relative to the corresponding historical boundary. It then combines this with the participation ratio of positive change monitoring points in the spatiotemporally connected overlapping segments, the temporal overlap ratio, and the migration ratio of the activity center to generate a comprehensive overlapping value, ultimately forming a comprehensive over-arching discrimination value. This processing method does not rely solely on data changes from a single monitoring point or a single time window for judgment, but simultaneously considers the degree of displacement change, duration, spatial correlation, and the degree of change in activity location, thereby improving the reliability of the over-arching monitoring data processing process for deep road cut foundations. Attached Figure Description

[0021] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0022] Figure 1 This is a schematic diagram illustrating the steps of the deep road cut foundation arch monitoring data processing method of the present invention; Figure 2 This is a schematic diagram of some steps in S1 of the deep road cut foundation arch monitoring data processing method of the present invention; Figure 3 This is a schematic diagram of some steps in step S2 of the deep road cut foundation arch monitoring data processing method of the present invention; Figure 4 This is a schematic diagram of the data flow timing in S3 of the deep road cut foundation arch monitoring data processing method of the present invention; Figure 5 This is a schematic diagram of some steps in S3 of the deep road cut foundation arch monitoring data processing method of the present invention; Figure 6 This is a schematic diagram of the data flow timing in S4 of the deep road cut foundation arch monitoring data processing method of the present invention; Figure 7This is a schematic diagram of some steps in step S4 of the deep road cut foundation arch monitoring data processing method of the present invention; Figure 8 This is a schematic diagram of the module structure and data flow of the deep road cut foundation arch monitoring data processing system of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0024] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0025] It should be noted that similar labels and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0026] This invention provides a method for processing monitoring data on the arching of deep roadbed foundations, such as... Figure 1 As shown, in one specific embodiment, the method includes:

[0027] S1. Obtain the base normal displacement value, sampling time and plane coordinates of each monitoring point, perform unified benchmark processing, validity check and time alignment on the monitoring data, and generate the median value of window displacement and the median value of window displacement rate according to the time window. S2. Before the current monitoring cycle, set the calibration historical section and the verification historical section, and filter the candidate historical windows in the two historical sections step by step. Generate the displacement historical boundary, displacement rate historical boundary and reference duration based on the calibration historical window. S3. Determine spatially adjacent monitoring points based on the plane coordinates of the monitoring points, extract the positive continuous change segments of each monitoring point, merge the original overlapping time periods formed by the positive continuous change segments of spatially adjacent monitoring points into spatiotemporal connected overlapping segments according to the temporal connectivity and spatial connectivity, and generate the same-direction overlapping comprehensive value. S4. Generate the over-limit value of the continuous change segment based on the median value of the window displacement, the median value of the window displacement rate and the corresponding historical boundary. Generate the comprehensive arching discrimination value based on the over-limit value of the continuous change segment and the comprehensive value of the same direction overlap. Generate the comprehensive arching discrimination boundary based on the verification historical window and output the base arching early warning result.

[0028] The current monitoring period refers to the continuous monitoring time range used to perform the current basement arch monitoring data discrimination. Data within the current monitoring period are added gradually in chronological order, and the corresponding intermediate data and discrimination results are updated each time a new time window ends and valid data is obtained.

[0029] Furthermore, in S1, all monitoring points adopt a unified positive displacement direction, a unified displacement reference, and a unified time reference. The raw data undergoes a series of processes, including duplicate record elimination, data quality check, equipment range check, sampling time realignment, missing data record verification, and interpolation within the allowable range, before generating the median value of the window displacement and the median value of the window displacement rate.

[0030] Furthermore, in S2, the calibration history section and the verification history section are responsible for boundary calibration and boundary verification functions, respectively. The calibration history section is only used to generate displacement history boundaries, displacement rate history boundaries, and reference duration lengths; the verification history section is only used to generate historical composite camber discrimination values ​​and composite camber discrimination boundaries.

[0031] Furthermore, in S3, the positive continuous change segment is simultaneously limited by data continuity and reference duration. Missing values, time gaps, zero displacement differences, and negative displacement differences all interrupt candidate positive continuous change segments. Changes from spatially disconnected monitoring points are not merged into the same monitoring event due to temporal overlap.

[0032] Furthermore, in S4, displacement and displacement rate are compared with historical boundaries of the same dimension, and then converted into dimensionless over-boundary ratios. The overall arch discrimination value is simultaneously affected by the degree of numerical over-boundary, the degree of monitoring point participation, the degree of temporal overlap, and the degree of activity location migration.

[0033] In summary, this implementation method first uses calibrated historical segments to generate the historical displacement range, historical displacement rate range, and continuous change scale of the target monitoring area itself, and then uses independent verification historical segments to generate a comprehensive discrimination boundary. After processing for temporal continuity, spatial connectivity, numerical out-of-bounds verification, and activity location migration, the current monitoring data forms a traceable and verifiable early warning result.

[0034] like Figure 2 As shown, in one specific embodiment, S1 includes: S11. Obtain the monitoring area number, monitoring point number, sampling time, base normal displacement value, data quality mark and plane coordinates of each monitoring point, and perform unified benchmark processing and validity check on the data; S12. Based on the sampling interval, sampling time tolerance, maximum allowable interpolation time interval, time window length, time window sliding step size, and minimum effective monitoring point ratio, perform time realignment, missing record processing, interpolation, and time window division on the effective data. S13. Generate the median value of window displacement based on the effective base normal displacement value at the end of each time window, generate the median value of window displacement rate based on the effective base normal displacement value at the start and end of each time window, and form a corresponding sequence according to the time order.

[0035] In this embodiment, it should be noted that in S11, the local normal direction of the base refers to the normal direction of the base design surface or as-built measurement surface where the monitoring point is located towards the top of the base. When the location of the monitoring point is defined by multiple base surface elements, the base surface element to which the monitoring point is actually attached is used; if it still cannot be uniquely determined, the base surface element with the smaller number is used in ascending order of the base surface element numbers.

[0036] Specifically, the local normal direction and positive displacement direction of the base are determined based on design surface data, as-built measurement results, monitoring device installation records, and stability control point re-measurement results. If the monitoring device directly outputs the base normal displacement, the positive direction fixed in the device is used; if the device outputs three-dimensional displacement, the three-dimensional displacement is projected onto the local normal direction of the base. For example, the positive local normal direction of a horizontal base surface is the vertically upward direction.

[0037] The base normal displacement value refers to the signed displacement value of the monitoring point relative to the unified reference position along the local normal direction of the base. The displacement along the local normal direction of the base is taken as a positive value, and the displacement in the opposite direction is taken as a negative value. The local normal direction of the base may be different for different monitoring points, but each monitoring point outputs a displacement value according to its own fixed positive normal direction.

[0038] Furthermore, a unified reference position refers to the displacement starting position commonly used by the same monitoring point in the calibration historical section, the verification historical section, and the current monitoring cycle. The same monitoring point must not be zeroed out separately within the entire time frame used for the same data processing; otherwise, the cumulative displacement values ​​across different time frames cannot be directly compared.

[0039] Specifically, the unified benchmark position is determined based on the initial stable remeasurement position after construction is completed or the benchmark time position determined by the engineering monitoring plan. After a monitoring point is replaced, the benchmark position is reset, or the monitoring device is zeroed out, the data before and after the reset cannot be directly used for the same boundary generation and current monitoring judgment. For example, if a monitoring point cannot be converted to its original benchmark position after equipment replacement, historical data must be re-established from the stable remeasurement time after the equipment replacement.

[0040] Furthermore, the target monitoring area is determined based on the segmentation of the deep road cut structure, geological zoning, drainage zoning, construction zoning, and the layout of monitoring points. Monitoring points within the same target monitoring area should have comparable displacement benchmarks and continuous spatial adjacency; areas that are far apart and have no connecting monitoring points should not be forcibly merged.

[0041] Specifically, target monitoring areas can be divided based on limited survey data, construction drawings, and monitoring point layout plans. For example, a deep road cut with a continuous length of 96 m can be divided into three target monitoring areas, each approximately 32 m long, based on geological changes and drainage zones. Historical boundaries and early warning results are generated for each area.

[0042] Among them, the effective base normal displacement value refers to the base normal displacement value that has complete monitoring point number, sampling time and data quality information, does not exceed the effective range of the monitoring device, and is not judged as a duplicate record, equipment failure record, verification failure record or data truncation record.

[0043] Specifically, when multiple records exist at the same monitoring point at the same sampling time, the record with the larger collection sequence number is retained first; if there is no collection sequence number, the record with the later data reception time is retained. Other duplicate records are not included in the actual count of the number of original valid data, nor are they included in the calculation of window feature values.

[0044] Furthermore, records marked with data quality indicating equipment failure, verification failure, data truncation, or exceeding the measurement range are deemed invalid. When a data quality marker is missing, a check is performed based on the effective measurement range of the monitoring device, equipment status records, and the continuity of adjacent sampling; records that cannot be confirmed as valid are not proceeded to further processing.

[0045] Specifically, the effective range and minimum resolvable unit of the monitoring device are determined based on the equipment verification certificate, calibration records, data interface specifications, and the expected maximum deformation on site. The number of decimal places in the data file cannot be directly assumed to represent the measurement accuracy of the equipment. For example, when the expected displacement does not exceed 20 mm, a monitoring device with an effective range of -50 mm to 50 mm and an output resolution of 0.01 mm can be used.

[0046] Furthermore, the horizontal and vertical coordinates of the monitoring points use the same coordinate system and the same unit of length. When the physical location of a monitoring point changes, the changed monitoring point is treated as a new monitoring point, and the horizontal coordinates and displacement records before and after the change are not directly connected.

[0047] In S12, the sampling interval refers to the time interval between two adjacent sampling times at the same monitoring point. The sampling interval is determined based on the expected shortest duration of change, the communication capability of the monitoring device, the data storage capacity, and the short-term fluctuation characteristics in the limited historical data.

[0048] Specifically, the same continuous historical data can be resampled at multiple candidate sampling intervals, and the retention of positive change segments can be compared. For example, 90 days of raw data can be tested using sampling intervals of 5 min, 10 min, and 30 min. If the 5 min data has significant redundancy, and the 30 min interval cannot completely retain the change process of approximately 1 hour, then the 10 min interval can be used. Here, d represents days, h represents hours, and min represents minutes.

[0049] Sampling time tolerance refers to the maximum permissible time deviation from the actual sampling time that should be included in the sampling time. The sampling time tolerance is less than half the sampling interval and remains constant within the same monitoring area.

[0050] Specifically, the sampling time tolerance is determined based on the statistical results of clock deviation during the stable operation period of the monitoring device. The time difference between each record in the limited historical data and the most recent expected sampling time can be statistically analyzed. For example, if statistical analysis of 30 days of data reveals that the clock deviation of normal records does not exceed 40 seconds, then the sampling time tolerance is set to 60 seconds.

[0051] Furthermore, all sampling times are converted to the same time base. When a sampling record falls within the sampling time tolerance range of a given sampling time, it is assigned to that given sampling time. A record can only be assigned to one given sampling time; if it meets the requirements of two times simultaneously, it is assigned to the time with the smaller time difference; if the time differences are the same, it is assigned to the earlier time.

[0052] For example, when the sampling interval is 10 min and the sampling time tolerance is 60 s, the record with a sampling time of 10:09:40 should be included in the 10:10:00 sampling time.

[0053] A time window refers to a time interval divided according to its length and sliding step, encompassing both start and end boundary times. The sliding step is the time interval between the start times of two adjacent time windows, and also the time interval between the end times of two adjacent, consecutive time windows.

[0054] Specifically, the time window length and time window sliding step size are jointly determined based on the sampling interval, the duration of short-term measurement fluctuations, and the response time of the changes to be identified. Multiple candidate windows can be combined to process limited historical data, comparing the fluctuation of the median value within the window and the identification delay.

[0055] For example, test time windows of 30 min, 60 min and 120 min respectively. If the 30 min time window is greatly affected by a single fluctuation and the 120 min time window responds slowly, then use a sliding step size of 60 min time window and 30 min time window.

[0056] Furthermore, each time window includes both a start boundary time and an end boundary time. When the time window length is 60 min and the sampling interval is 10 min, the sampling times that a single monitoring point should have within a time window include the start time, 10 min, 20 min, 30 min, 40 min, 50 min, and the end time, for a total of 7 times.

[0057] The maximum allowed interpolation time interval refers to the maximum time distance between two valid sampling times before and after the missing time point that allows for linear interpolation. Interpolation is only used for window feature value calculation and does not change the already determined original number of missing values.

[0058] Specifically, the maximum allowed interpolation time interval is determined based on the duration of normal short-term packet loss and the smoothness of displacement changes in the limited historical data. For example, after statistical analysis of 60 days of communication records, normal short-term packet loss mainly involves the absence of a single 10-minute sampling time, while the absence of more than three consecutive sampling times is usually related to equipment offline. Therefore, the maximum allowed interpolation time interval is set to 20 minutes.

[0059] Furthermore, when there is valid data before and after the missing time point and the time interval between them does not exceed the maximum allowed interpolation time interval, linear interpolation is performed according to the time position of the missing time point between the valid time points before and after it.

[0060] For example, if the base normal displacement of a monitoring point is 3.1 mm at 10:10 and 3.3 mm at 10:30, then the interpolation value for 10:20, which is the midpoint between the two times, is: .

[0061] Among them, a valid monitoring point refers to a monitoring point that has a valid base normal displacement value at all times required by the corresponding processing step. The median value of the window displacement only requires the monitoring point to have a valid value at the end of the time window; the median value of the window displacement rate requires the monitoring point to have a valid value at both the beginning and end of the time window.

[0062] Furthermore, the minimum effective monitoring point ratio refers to the minimum proportion of the number of effective monitoring points required to generate the corresponding window feature values ​​relative to the total number of monitoring points. The minimum effective monitoring point number is determined by rounding up the product of the total number of monitoring points and the minimum effective monitoring point ratio.

[0063] Specifically, the minimum effective monitoring point ratio is determined based on the total number of monitoring points, their spatial distribution, and the number of disconnections in limited historical data. For example, if 12 monitoring points are tested at 70%, 80%, and 90% respectively, and if at 70% the effective points may be concentrated on one side, at 90% the normal short-term disconnections may cause too many windows to fail, and at 80% the median value remains stable, then 80% is adopted, corresponding to at least 10 effective monitoring points.

[0064] Furthermore, when the number of valid monitoring points with end time data within a time window reaches the minimum number of valid monitoring points, the median value of the window displacement is generated; when the number of valid monitoring points with both start and end time data reaches the minimum number of valid monitoring points, the median value of the window displacement rate is generated.

[0065] The number of consecutive invalid time windows refers to the number of time windows during which the desired feature value cannot be generated consecutively. This number is used to prevent the formation of false continuous change segments across long data gaps.

[0066] Specifically, the number of consecutive invalid time windows is determined based on the duration of normal communication interruptions in limited historical data. For example, if historical communication records show that normal short-term faults typically only affect one or two consecutive time windows, then the stop condition is set to three consecutive invalid time windows.

[0067] Furthermore, if the required feature values ​​cannot be generated within three consecutive time windows, the current continuous sequence processing is stopped and a data re-acquisition prompt is output. After data recovery, a new continuous time window sequence is formed from the recovery time, calculating the displacement difference without crossing invalid time windows.

[0068] In S13, the median refers to the average of the two median values ​​when the data are arranged in ascending order of numerical value. If the number of data is odd, the median is taken; if the number of data is even, the median is taken.

[0069] The median displacement of a window refers to the median of the base normal displacement values ​​of all valid monitoring points at the end of a time window. The median displacement of a window is used to characterize the middle position of the displacement data within the target monitoring area at the end of the time window.

[0070] For example, if the normal displacements of the two bases at the center of the sorted position at the end of a certain time window are 4.05 mm and 4.09 mm, then the median displacement of the window is: .

[0071] Among them, the base normal displacement rate refers to the rate of change of the difference between the base normal displacement value at the end of the time window and the base normal displacement value at the beginning of the time window at the same monitoring point, relative to the duration of the time window.

[0072] At a certain monitoring point, the base normal displacement value at the start of the time window is 3.60 mm, and the value at the end is 3.84 mm. The time window lasts for 1 hour. Therefore, the base normal displacement rate is: .

[0073] The median displacement rate within a window refers to the median of the base-normal displacement rates of all valid monitoring points within a time window. This data is used to characterize the midpoint of the displacement change rate of the target monitoring area within the time window.

[0074] For example, if the normal displacement velocities of the two bases located at the center of the sorting position within the same time window are 0.20 mm / h and 0.22 mm / h, then the median displacement velocity within the window is: .

[0075] Furthermore, the effective monitoring point sets used for the median window displacement and the median window displacement rate can be different. If 11 out of 12 monitoring points have end time data, and 10 of them have both start and end time data, then the median window displacement is generated from the 11 displacement values, and the median window displacement rate is generated from the 10 rates.

[0076] The median values ​​of all window displacements and the median values ​​of window displacement rates are arranged in chronological order of the end times of the time windows, forming a time series for processing by S2 to S4. Time windows with missing features retain their time position and missing markers, and are not padded across by valid data from before or after them.

[0077] like Figure 3 As shown, in one specific embodiment, S2 includes:

[0078] S21. Determine the calibration historical segment and the verification historical segment, divide the two historical segments into candidate historical windows, and determine the number of missing items, the number of direction flips, and the median absolute deviation of displacement at the end of each candidate historical window.

[0079] S22. Sort and retain the candidate calibration history window and the candidate verification history window level by level to obtain the calibration history window and the verification history window respectively.

[0080] S23. Generate displacement history boundary, displacement rate history boundary and reference duration based on calibration history window, and form a continuous verification history window sequence based on verification history window.

[0081] In this embodiment, it should be noted that in S21, the calibration history segment refers to the historical time range located before the verification history segment, used to generate the displacement history boundary, displacement rate history boundary, and reference duration.

[0082] Among them, the verification historical section refers to the historical time range that is located after the calibration historical section and before the current monitoring cycle, and is used to generate the historical comprehensive arch discrimination value and the comprehensive arch discrimination boundary.

[0083] Furthermore, the calibration of historical sections, the verification of historical sections, and the current monitoring cycle all use the same monitoring point numbering system, the same unified benchmark location, the same sampling interval, the same time window length, and the same time window sliding step size. If the number of monitoring points changes or if a unified benchmark conversion cannot be completed, the data before and after the change will not be used in the same historical calibration.

[0084] Specifically, the initial duration of the current monitoring cycle, the calibration historical section, and the verification historical section are determined based on the engineering monitoring report cycle, seasonal variations, and the amount of available historical data. The two historical sections should be independent of each other and each have a sufficient number of time windows.

[0085] For example, if the current monitoring period is 30 days, an earlier 30 days is selected as the calibration historical period, and the next 30 days is selected as the verification historical period, so that the calibration historical period, the verification historical period and the current monitoring period have the same initial duration.

[0086] Furthermore, known abnormal times in historical sections are identified and verified, and these are eliminated based on construction logs, equipment maintenance records, monitoring benchmark retest records, existing early warning verification records, and on-site inspection records.

[0087] Specifically, the timeframes for monitoring point replacement, unified benchmark location reset, equipment disassembly and assembly, equipment range adjustment, concentrated construction collisions, and confirmed engineering anomalies should be excluded. For example, if a monitoring device was replaced within 2 days in the verification historical section, the data between the device disassembly and assembly and the completion of stable retesting should be deleted, and the remaining 2 days should be added from earlier valid data.

[0088] Among them, the candidate calibration historical window refers to the time window obtained from the calibration historical segment according to the same time window length and time window sliding step as the current monitoring cycle, and which has not yet completed the step-by-step screening.

[0089] Among them, the candidate verification history window refers to the time window obtained from the verification history segment according to the same time window length and time window sliding step as the current monitoring cycle, and which has not yet completed the step-by-step screening.

[0090] The missing quantity refers to the difference between the number of original effective base normal displacement values ​​that the candidate history window should have and the actual number of original effective base normal displacement values. The missing quantity is determined before interpolation, and subsequent interpolation does not change the determined missing quantity.

[0091] Specifically, when the time window lasts for 60 minutes, the sampling interval is 10 minutes, and the number of monitoring points is 12, the number of original data points should be: 7 × 12 = 84. If a candidate historical window actually obtains 82 original valid base normal displacement values, then the number of missing values ​​is: 84 - 82 = 2.

[0092] Furthermore, for each required sampling time within the candidate historical window, when the number of monitoring points with valid data reaches the minimum number of valid monitoring points, the median of the normal displacement values ​​of each valid base at that time is calculated. If the minimum number of valid monitoring points is not met, the corresponding sampling time is marked as invalid.

[0093] Among them, the number of direction reversals refers to the number of times that adjacent positive and negative signs are converted between each other in the sign sequence of the median difference of the sampling time displacement of the candidate historical window after excluding zero difference.

[0094] Specifically, the median displacement values ​​at consecutive effective sampling times are subtracted sequentially. Positive differences are recorded as positive, negative differences as negative, and zero differences are not included in the direction reversal statistics.

[0095] For example, the median displacement values ​​at the seven sampling times of a candidate historical window are 2.10 mm, 2.12 mm, 2.15 mm, 2.17 mm, 2.16 mm, 2.18 mm and 2.19 mm, respectively, with the difference signs being positive, positive, positive, negative, positive and positive, respectively, and the number of direction flips is 2.

[0096] Furthermore, if there is no difference sign or only one difference sign after excluding zero differences, the number of direction reversals is recorded as 0. If a candidate history window cannot form the median displacement value at at least two consecutive valid sampling times, it is marked as having an incalculable number of direction reversals.

[0097] Among them, the absolute deviation of the median displacement at the end of the window refers to the median of the absolute differences between the effective base normal displacement values ​​at the end of the candidate historical window and the median displacement at that end time.

[0098] For example, if the two data points in the middle of the sorted absolute differences at the end of the candidate historical window are 0.06 mm and 0.07 mm, then the median absolute deviation of the displacement at the end of the window is: .

[0099] Furthermore, candidate historical windows for which the number of direction flips or the absolute deviation of the median displacement at the end of the window cannot be calculated are placed after all computable windows in the corresponding sorting. If the number of computable candidate historical windows at any level is less than the number that should be retained at that level, the corresponding historical segment is expanded and reprocessed.

[0100] In S22, the calibration history window refers to the time window retained after the candidate calibration history window has been filtered step by step based on the number of missing values, the number of orientation flips, and the median absolute deviation of the displacement at the end of the window.

[0101] The verification history window refers to the time window retained after the candidate verification history window has undergone the same step-by-step screening. The candidate calibration history window and the candidate verification history window are processed separately, and the windows of the two historical segments are not mixed.

[0102] Specifically, all candidate historical windows are first sorted in ascending order of the number of missing windows, and the first half is retained; then the retained windows are sorted in ascending order of the number of orientation flips, and the first half is retained; finally, they are sorted in ascending order of the median absolute deviation of the displacement at the end of the window, and the first half is determined as calibration historical windows or verification historical windows.

[0103] Furthermore, the first half of each level is retained according to fixed sorting and filtering rules, and is not determined through random training, probability sampling, or manual temporary weighting. The three levels of filtering successively restrict the completeness of data, the repetition of displacement direction, and the width of displacement distribution at monitoring points.

[0104] For example, with an initial pool of 1439 candidate history windows, the number of windows retained at level 1 is: ,in, Rounding up. Level 2 retention. One, Level 3 Reservation indivual.

[0105] Furthermore, when multiple candidate historical windows located at the retention boundary have the same sorted data, the time interval between the end time of each window and the start time of the current monitoring period is compared, and windows with smaller time intervals are retained first; if they are still the same, the order is determined by ascending order of the candidate historical window numbers.

[0106] The candidate historical window numbers are generated sequentially based on the start time of the corresponding historical segment and the window sliding order. Each number within the same historical segment is unique; random or runtime-generated numbers are not used.

[0107] In S23, the displacement history boundary includes the lower boundary and the upper boundary of the displacement history, which is the range of historical values ​​generated based on the median value of the window displacement of the calibrated history window.

[0108] Specifically, the median values ​​of all calibrated historical window displacements are sorted in ascending order, with the minimum value used as the initial lower boundary of the displacement and the maximum value used as the initial upper boundary. The differences between adjacent data with different values ​​are calculated, and the median of all non-zero differences is used as the displacement boundary expansion amount.

[0109] For example, if the minimum median displacement value of the window is 1.80 mm and the maximum is 4.60 mm, and the median of the non-zero adjacent differences is 0.02 mm, then the lower boundary of the displacement history is: 1.80 - 0.02 = 1.78 mm. The upper boundary of the displacement history is: 4.60 + 0.02 = 4.62 mm. The displacement rate history boundary includes both the lower and upper boundaries, and is a range of historical values ​​generated based on the median displacement rate value of the calibrated historical window.

[0110] For example, when the median displacement rate of the window has a minimum value of -0.12 mm / h and a maximum value of 0.18 mm / h, and the median of the non-zero adjacent differences is 0.01 mm / h, the lower boundary of the displacement rate history is: -0.12 - 0.01 = -0.13 mm / h. The upper boundary of the displacement rate history is: 0.18 + 0.01 = 0.19 mm / h.

[0111] Furthermore, the boundary expansion amount is preferentially generated from the median difference between adjacent and different values ​​in the historical data, without pre-setting a fixed percentage. When all corresponding historical data are identical, the smallest resolvable unit of the corresponding data is used as the boundary expansion amount.

[0112] Specifically, the median value of all window displacements is 2.00 mm, and when the device output resolution is 0.01 mm, the displacement boundary extension is 0.01 mm. This substitution rule is used to ensure that there is a non-zero distance between the upper and lower boundaries of the displacement history.

[0113] Among them, the continuous calibration history window sequence refers to a set of calibration history windows arranged according to their end times, where the time interval between adjacent end times is equal to the sliding step size of a time window.

[0114] Furthermore, all calibration history windows are arranged in ascending order of their end times. When there is a time window sliding step between adjacent end times, the two are connected to form a continuous calibration history window sequence; if there are unreserved windows, feature-missing windows, or time gaps, the sequence is interrupted at the gaps.

[0115] Among them, the historical candidate positive continuous change segment refers to the time period covered by the continuous positive displacement difference in the continuous calibration historical window sequence that has not been filtered by the reference duration.

[0116] Specifically, a continuous calibration history window sequence includes at least two calibration history windows. For each monitoring point in the sequence, the displacement difference is calculated only if there is a valid base normal displacement value at the end of adjacent windows. The time intervals corresponding to continuous positive displacement differences are merged into historical candidate positive continuous change segments.

[0117] For example, when the displacement value of a certain monitoring point continuously increases at 08:00, 08:30, 09:00 and 09:30, a historical candidate positive continuous change segment is formed from 08:00 to 09:30, lasting for 1.5 hours.

[0118] The reference duration length refers to the median duration length of all historical candidate positive continuous change segments. Historical candidate positive continuous change segments are not pre-selected using the reference duration length during generation.

[0119] For example, this embodiment generates 256 historical candidate positive continuous change segments, of which 74 last for 0.5 h, 110 last for 1.0 h, 52 last for 1.5 h, and 20 last for 2.0 h. The 128th and 129th data points after sorting are both 1.0 h, with a reference duration of 1.0 h.

[0120] Furthermore, if a continuous historical calibration window sequence cannot be formed after screening, or if a historical candidate positive continuous change segment cannot be formed, the historical calibration segment is extended forward according to the duration of the current monitoring cycle, and data cleaning, candidate window division, and three-level screening are re-executed.

[0121] Specifically, when the current monitoring period is 30 days, it will be extended forward by 30 days each time. If all available historical data still cannot form a reference duration, an insufficient historical data prompt will be output, and the current comprehensive upward arch judgment will not be performed.

[0122] The continuous verification history window sequence refers to a set of verification history windows arranged according to their end times, where the time interval between adjacent end times is equal to the sliding step size of a time window. The continuous verification history window sequence is only used to generate historical composite arch discrimination values ​​and does not regenerate displacement history boundaries, displacement rate history boundaries, or reference duration.

[0123] like Figure 4 and Figure 5 As shown, in one specific embodiment, S3 includes:

[0124] S31. Determine spatially adjacent monitoring points based on the plane coordinates of the monitoring points, and generate a positive continuous change segment based on the difference in the base normal displacement at the end of the adjacent time window of each monitoring point.

[0125] S32. Generate the original overlapping time period based on the common coverage time of the positive continuous change segment of spatially adjacent monitoring points, and merge the original overlapping time period into a spatiotemporally connected overlapping segment according to the temporal connectivity relationship and spatial connectivity relationship.

[0126] S33. Generate a composite value of the same direction of overlap based on the participation rate of positive change monitoring points, the time overlap rate, and the migration rate of activity centers.

[0127] In this embodiment, it should be noted that in S31, spatially adjacent monitoring points refer to two monitoring points connected by the same Deloitte triangulation edge, or two monitoring points whose positions are adjacent after being arranged according to the prescribed coordinate order when a Deloitte triangulation cannot be formed.

[0128] Specifically, before establishing the Deloitte triangulation, all monitoring points are arranged in ascending order according to their x-coordinate, y-coordinate, and monitoring point number. Then, monitoring points are inserted in this fixed order, and all adjacent point pairs corresponding to the edges of the triangulation are saved.

[0129] Furthermore, the insertion order of monitoring points and the rules for handling co-circular conditions in the Deloitte triangulation are fixed when the target monitoring area is first established, and the same rules are used for subsequent inputs of the same coordinates.

[0130] Specifically, when multiple monitoring points share a circle and there are two possible diagonal sides, the combinations of endpoint numbers of the two diagonal sides are compared, and the combination that appears earlier in the lexicographical order after being sorted in ascending order is retained. For example, when the endpoint numbers of the candidate diagonal sides are (3,6) and (4,5) respectively, (3,6) is used.

[0131] Furthermore, among multiple monitoring points with the same planar coordinates, only the monitoring point with the smallest monitoring point number is retained to participate in the construction of spatial adjacency relationships, while other monitoring points with overlapping coordinates still participate in the calculation of the median value of window displacement and the median value of window displacement rate.

[0132] Furthermore, when there are fewer than 3 monitoring points, all monitoring points are collinear, or a Deloitte triangulation cannot be formed for other reasons, the horizontal and vertical spans are calculated separately, and the points are prioritized for sorting along the direction with the larger span. When the spans are the same, the points are prioritized for sorting along the horizontal coordinate direction, and then sorted according to the coordinates of the other direction and the monitoring point number. Two monitoring points with adjacent positions are determined as spatially adjacent monitoring points.

[0133] The judgment time interval refers to the time interval between the end of the previous time window and the end of the next time window. A judgment time interval is formed only when the time interval between the two end times is equal to the sliding step size of a time window.

[0134] Among them, the positive change interval refers to the judgment time interval when the base normal displacement value at the end of the next time window of the same monitoring point is greater than the base normal displacement value at the end of the previous time window.

[0135] Specifically, the displacement difference is calculated only when the time interval between the end times of two adjacent time windows is equal to the sliding step size of a time window, and there are effective base normal displacement values ​​at both end times.

[0136] Furthermore, when the displacement difference is positive, the corresponding time interval is determined as a positive change interval. When the displacement difference is negative, equal to 0, data is missing at either end, or there is a time gap, no positive change interval is generated, and the continuity between the preceding and following intervals is interrupted.

[0137] Among them, the candidate positive continuous change segment refers to the time period formed by merging multiple consecutive positive change intervals at the same monitoring point.

[0138] Among them, the positive continuous change segment refers to the candidate positive continuous change segment whose duration reaches the reference duration.

[0139] Specifically, the product of the number of positive change intervals contained in the candidate positive continuous change segment and the sliding step size of the time window is the duration of the candidate positive continuous change segment.

[0140] For example, the base normal displacement values ​​at the end of six consecutive time windows at a certain monitoring point were 4.30 mm, 4.42 mm, 4.58 mm, 4.77 mm, 4.96 mm, and 5.14 mm, respectively, forming five consecutive positive variation intervals. When the time window sliding step size is 0.5 h, the duration is 5 × 0.5 = 2.5 h. This duration reaches the reference duration of 1.0 h, therefore it is determined to be a positive continuous variation segment.

[0141] In S32, the original overlapping time period refers to the time interval covered by the positive continuous change segments of a pair of spatially adjacent monitoring points and whose duration is greater than 0.

[0142] Specifically, the later start time of the two positive continuous change segments is taken as the start time of the original overlapping time period, and the earlier end time is taken as the end time of the original overlapping time period. When the start time is earlier than the end time, the original overlapping time period is generated; when the start time is equal to or later than the end time, the original overlapping time period is not generated.

[0143] Furthermore, each original overlapping time period simultaneously saves the start time, end time, and the corresponding two monitoring point numbers, so as to simultaneously determine the temporal connectivity and spatial connectivity in subsequent tests.

[0144] Among them, the spatiotemporally connected overlapping segment refers to a time period formed by merging multiple original overlapping time periods that overlap in time or have adjacent boundaries and are interconnected in terms of the adjacency relationship of monitoring points.

[0145] Specifically, all original overlapping time periods are arranged in ascending order by their start time, end time, and monitoring point pair number. Two original overlapping time periods are merged into the same spatiotemporally connected overlapping segment only if both temporal connectivity and spatial connectivity conditions are met simultaneously.

[0146] Furthermore, the temporal connectivity condition is that the two original overlapping time periods overlap in time or their boundaries are adjacent. The spatial connectivity condition is that the monitoring point pairs corresponding to the two original overlapping time periods have common monitoring points, or can form a continuous adjacent path through other spatially adjacent monitoring point pairs that have been merged.

[0147] Specifically, during merging, the first sorted original overlapping time period is taken as the current merged segment. If the next original overlapping time period satisfies both the temporal connectivity condition and the spatial connectivity condition, the start time, end time, and monitoring point pair set of the current merged segment are updated; otherwise, the current merged segment is output and a new current merged segment is established.

[0148] For example, the 4th monitoring point and the 5th monitoring point form an original overlapping time period from 08:30 to 10:00, the 5th monitoring point and the 6th monitoring point form an original overlapping time period from 09:00 to 10:30, and the 6th monitoring point and the 7th monitoring point form an original overlapping time period from 10:00 to 10:30.

[0149] The aforementioned monitoring points are continuously connected through a common monitoring point, and they overlap in time or are adjacent in boundary, thus forming a spatiotemporally connected overlapping segment from 08:30 to 10:30.

[0150] If the 11th and 12th monitoring points form another original overlapping time period from 09:00 to 10:00, but cannot form a continuous adjacent path with the 4th to 7th monitoring points, then another spatiotemporally connected overlapping segment is formed separately.

[0151] In S33, a data integrity monitoring point refers to a monitoring point that has a valid base normal displacement value at the end of all time windows contained in a spatiotemporally connected overlapping segment.

[0152] Among them, the participation ratio of positive change monitoring points refers to the ratio of the number of data complete monitoring points participating in positive continuous change within the spatiotemporally connected overlapping segment to the total number of data complete monitoring points.

[0153] Specifically, first count the number of monitoring points that have effective base normal displacement values ​​at the end of all time windows in a spatiotemporally connected overlapping segment, and then count the number of different monitoring points that are in a positive continuous change segment within at least one judgment time interval.

[0154] For example, when the number of positive change monitoring points is 9 and the number of monitoring points with complete data is 12, the participation ratio of positive change monitoring points is 9 / 12 = 0.75. Furthermore, when the number of monitoring points with complete data is 0, the corresponding spatiotemporal connectivity overlapping segment is marked as insufficient data, the overlapping value is not calculated, and it is not included in the comprehensive arch discrimination.

[0155] The temporal overlap ratio refers to the ratio of the longest original overlapping time period within a spatiotemporally connected overlapping segment to the length of that spatiotemporally connected overlapping segment.

[0156] For example, if the longest original overlap period lasts 1.5 hours and the spatiotemporal connectivity overlapping segment in the same direction lasts 2.0 hours, the time overlap ratio is 1.5 / 2 = 0.75. Here, the baseline value for the same-direction overlap refers to the smaller of the participation ratio of the positive change monitoring points and the time overlap ratio. In this embodiment, both are 0.75, therefore the baseline value for the same-direction overlap is 0.75.

[0157] Among them, the coordinate center of the activity monitoring point refers to the average position of the plane coordinates of the monitoring point in a positive continuous change segment within a judgment time interval.

[0158] Specifically, for each judgment time interval within the spatiotemporally connected overlapping segment, the monitoring points that completely cover the judgment time interval by the positive continuous change segment are counted, and the average values ​​of the horizontal and vertical coordinates of these monitoring points are calculated to obtain the coordinate center of the active monitoring point.

[0159] The cumulative migration length of the activity center refers to the length obtained by accumulating the planar distance between the coordinate centers of adjacent activity monitoring points that are consecutive in time, according to the time sequence.

[0160] Furthermore, the planar distance between adjacent activity monitoring point coordinate centers is calculated only if their corresponding judgment time intervals are temporally continuous. If there is no activity monitoring point coordinate center in the intermediate judgment time interval, the gap between the preceding and following coordinate centers is not crossed. When there is only one activity monitoring point coordinate center, the cumulative migration length of the activity center is recorded as 0.

[0161] Among them, the plane span of the monitoring point refers to the maximum plane distance between any two monitoring points with different plane coordinates within the target monitoring area.

[0162] The activity center migration ratio refers to the ratio of the cumulative migration length of the activity center to the plane span of the monitoring point. When the calculated result is greater than 1, it is recorded as 1.

[0163] For example, when the cumulative relocation length of the activity center is 10.7581 m and the horizontal span of the monitoring point is 32.0 m, the relocation ratio of the activity center is: Furthermore, when the plane span of the monitoring point is 0, the migration ratio of the activity center is not calculated, and the basic value of the same-direction overlap is directly used as the comprehensive value of the same-direction overlap. The comprehensive value of the same-direction overlap refers to the larger of the basic value of the same-direction overlap and the migration ratio of the activity center. In this embodiment, the basic value of the same-direction overlap is 0.75, and the migration ratio of the activity center is 0.3362; therefore, the comprehensive value of the same-direction overlap is 0.75.

[0164] The cumulative migration length of the activity center only indicates the degree to which the average position of the monitoring points participating in the positive continuous change has migrated over time. It does not directly indicate that the base entity has undergone planar movement, nor does it directly indicate that the area of ​​the arched region has increased.

[0165] like Figure 6 and Figure 7 As shown, in one specific embodiment, S4 includes: S41. Generate the displacement over-limit ratio based on the median value of the window displacement and the historical boundary of the displacement; generate the displacement rate over-limit ratio based on the median value of the window displacement rate and the historical boundary of the displacement rate; and generate the window over-limit ratio and the over-limit value of the continuously changing segment. S42. Generate a comprehensive upward arch discrimination value based on the over-limit value of the continuous change segment and the comprehensive value of the same direction overlap. Verify the historical window by using the historical boundary generated by the calibration historical segment and the reference duration length. Generate historical comprehensive upward arch discrimination value and comprehensive upward arch discrimination boundary. S43. Update the current positive continuous change segment and the spatiotemporal connected overlapping segment in the same direction in a time window, and output the base arching early warning result based on the current comprehensive arching discrimination value, comprehensive arching discrimination boundary and reference duration.

[0166] In S41, the displacement overshoot ratio refers to the proportion of the portion of the window displacement median value that exceeds the boundary of the displacement history relative to the distance between the upper and lower boundaries of the displacement history. When the window displacement median value does not exceed the boundary of the displacement history, the displacement overshoot ratio is 0.

[0167] Specifically, when the lower boundary of the displacement history is 1.78 mm and the upper boundary is 4.62 mm, the distance between the upper and lower boundaries is 4.62 - 1.78 = 2.84 mm. When the median displacement of the window is 4.90 mm, the displacement exceedance ratio is: The displacement rate exceeding the limit ratio refers to the proportion of the portion of the window displacement rate median value that exceeds the historical boundary of the displacement rate relative to the distance between the upper and lower boundaries of the historical displacement rate. When the window displacement rate median value does not exceed the historical boundary of the displacement rate, the displacement rate exceeding the limit ratio is 0.

[0168] Specifically, when the lower boundary of the displacement rate history is -0.13 mm / h and the upper boundary of the displacement rate history is 0.19 mm / h, the distance between the upper and lower boundaries is: When the median displacement rate of the window is 0.31 mm / h, the displacement rate exceeding the limit is: .

[0169] Furthermore, when the historical upper and lower boundary distances used for calculating the over-limit ratio are less than the smallest resolvable unit of the corresponding data, the smallest resolvable unit of the corresponding data is used as the calculation distance.

[0170] Among them, the window over-limit ratio refers to the larger value between the displacement over-limit ratio and the displacement rate over-limit ratio corresponding to the same time window.

[0171] The out-of-bounds value of a continuously varying segment refers to the median proportion of all windows covered by a spatiotemporally connected overlapping segment in the same direction that exceed the bounds. Using the median for aggregation can reduce the impact of large instantaneous data in a single time window on the entire overlapping segment.

[0172] In S42, the comprehensive arch discrimination value refers to the discrimination data obtained by multiplying the over-limit value of the continuous change segment with the comprehensive value of the same direction overlap.

[0173] For example, when the over-limit value of the continuously changing segment is 0.3750 and the comprehensive value of the same direction overlap is 0.75, the comprehensive upward arch discrimination value is: 0.3750 × 0.75 = 0.2813.

[0174] Among them, the historical comprehensive arch discrimination value refers to the comprehensive arch discrimination value obtained after verifying the historical section by using the displacement historical boundary, displacement rate historical boundary and reference duration generated by the calibration historical section.

[0175] Specifically, all verification history windows are arranged according to their end times. Adjacent verification history windows with a sliding step of one time window between their end times are connected into a continuous verification history window sequence and processed according to the same S3 and S4 processing steps as the current monitoring cycle.

[0176] Furthermore, if a continuous verification history window sequence does not form a spatiotemporally connected overlapping segment, the corresponding historical comprehensive upward arch discriminant value is recorded as 0. When multiple spatiotemporally connected overlapping segments are formed, a historical comprehensive upward arch discriminant value is generated for each overlapping segment.

[0177] For example, if the window overshoot ratio of a certain historical overlapping segment is 0.10, 0.22, 0.32, 0.38 and 0.44, the overshoot value of the continuously changing segment is 0.32, and the comprehensive value of the same-direction overlapping segment is 0.50, then the historical comprehensive upward arch discrimination value is: 0.32 × 0.50 = 0.16.

[0178] Among them, the comprehensive upward arch discrimination boundary refers to the discrimination data obtained by adding the discrimination boundary expansion amount to the maximum value of the historical comprehensive upward arch discrimination value.

[0179] Specifically, all historical composite upward arch discrimination values ​​are arranged in ascending order of numerical value, with the maximum value as the initial discrimination boundary, and the median of the difference between adjacent historical composite upward arch discrimination values ​​with different values ​​as the discrimination boundary expansion amount.

[0180] For example, the total historical composite upward arch discriminant values ​​include 0, 0.04, 0.06, 0.08, 0.10, 0.12, 0.14 and 0.16, with a maximum value of 0.16. The median difference between adjacent different data is 0.02, and the composite upward arch discriminant boundary is: 0.16 + 0.02 = 0.18.

[0181] Furthermore, when there are no adjacent historical comprehensive discriminant values ​​with different values, the smallest recording unit of the uniformly set comprehensive discriminant value is used as the discriminant boundary expansion amount.

[0182] Specifically, the minimum recording unit for the comprehensive discrimination value is determined uniformly based on the precision of the preceding data and the number of decimal places to be saved during the deployment of the processing program, and remains unchanged in the calibration historical segment, the verification historical segment, and the current monitoring cycle. For example, when the comprehensive discrimination value is uniformly saved to 4 decimal places, the minimum recording unit for the comprehensive discrimination value is set to 0.0001.

[0183] In S43, the early warning result of the base arch refers to the monitoring data processing result output when the current comprehensive arch discrimination value exceeds the comprehensive arch discrimination boundary and the duration of the corresponding spatiotemporal connectivity overlapping segment reaches the reference duration.

[0184] Specifically, whenever a new time window ends and valid monitoring data is obtained, the monitoring point displacement difference, candidate positive continuous change segment, positive continuous change segment, original overlapping time period, spatiotemporal connectivity same-direction overlapping segment, and comprehensive upward arch discrimination value are updated.

[0185] Furthermore, for continuous positive change segments that are ongoing and have not yet exhibited negative displacement differences, zero displacement differences, missing data, or time gaps, the latest time window's end time is taken as the provisional end time.

[0186] Furthermore, the end time of the latest time window is also used as the provisional end time for the ongoing spatiotemporal connectivity overlapping segments. Once the provisional duration reaches the reference duration, the current comprehensive upward arch discrimination value can be calculated without waiting for the event to end.

[0187] Furthermore, when there are multiple spatially disconnected spatiotemporally connected overlapping segments in the same direction within the same time range, the participation ratio of positive change monitoring points, the temporal overlap ratio, the migration ratio of activity centers, the comprehensive value of overlapping segments in the same direction, the boundary value of continuous change segments, and the comprehensive upward arch discrimination value are calculated for each overlapping segment.

[0188] When any current composite arching discrimination value exceeds the composite arching discrimination boundary, and the duration of the corresponding spatiotemporal connectivity overlapping segment reaches the reference duration, the base arching warning result is output.

[0189] The early warning results should include at least the target monitoring area number, the number of spatiotemporal connectivity overlapping segments, the number of monitoring points involved, the start time, the current end time or the final end time, the participation rate of positive change monitoring points, the time overlap rate, the migration rate of the activity center, the comprehensive value of overlapping segments, the boundary value of continuous change segments, the comprehensive upward arching discrimination value, and the comprehensive upward arching discrimination boundary.

[0190] Furthermore, when the warning conditions are met for the first time, the event status is set to "New Warning". As the event continues to develop, the current end time and discrimination data are updated. When the event ends due to negative displacement difference, zero displacement difference, missing data, time gap, or interruption of spatial connectivity, the final end time and final discrimination data are saved.

[0191] This invention also provides a data processing system for monitoring the arching of deep roadbed foundations, such as... Figure 8 As shown, it includes a data conversion module, a history calibration module, a spatiotemporal overlap processing module, a discrimination output module, a processor, and a memory.

[0192] The data conversion module is used to obtain the base normal displacement value, sampling time, monitoring point number, data quality mark and plane coordinates of each monitoring point, perform unified benchmark processing, validity check, time realignment, missing records, interpolation within the allowable range and time window division, and generate the median value of window displacement and the median value of window displacement rate.

[0193] The historical calibration module is used to set the calibration historical segment and the verification historical segment, determine the number of missing candidate historical windows, the number of direction reversals, and the absolute deviation of the median displacement at the end of the window, filter the candidate historical windows step by step, generate the displacement historical boundary, displacement rate historical boundary, and reference duration based on the calibration historical window, and form a continuous verification historical window sequence.

[0194] The spatiotemporal overlap processing module is used to establish spatial adjacency relationships based on the planar coordinates of monitoring points, generate positive continuous change segments based on the difference in base normal displacement at the end of adjacent time windows, generate original overlapping time periods based on the positive continuous change segments of spatially adjacent monitoring points, generate spatiotemporal connected overlapping segments in the same direction according to temporal and spatial connectivity relationships, and generate a comprehensive value of overlapping segments in the same direction.

[0195] The discrimination output module is used to generate displacement over-limit ratio, displacement rate over-limit ratio, window over-limit ratio, continuous change segment over-limit value and comprehensive over-arch discrimination value. It uses the verification history window to generate historical comprehensive over-arch discrimination value and comprehensive over-arch discrimination boundary, and outputs the base over-arch warning result according to the current comprehensive over-arch discrimination value, comprehensive over-arch discrimination boundary and reference duration length.

[0196] Furthermore, the memory is used to store raw monitoring data, unified benchmark information, data quality markers, sampling intervals, sampling time tolerance, maximum allowable interpolation time intervals, time window parameters, minimum effective monitoring point ratio, calibration historical segments, verification historical segments, candidate historical windows, calibration historical windows, verification historical windows, historical boundaries, reference duration, spatial adjacency relationships, positive continuous change segments, original overlapping time periods, spatiotemporally connected overlapping segments, overlapping composite values, composite upward arching discrimination values, composite upward arching discrimination boundaries, and early warning results.

[0197] Furthermore, the processor reads program instructions from memory and executes data processing according to the processing order defined by S1 to S4. The data conversion module, history calibration module, spatiotemporal overlap processing module, and discrimination output module are functional units formed by the processor executing different program instructions, and do not require separate physical processors.

[0198] Specifically, raw monitoring data and early warning results can be stored on field terminals, local servers, or remote servers. Regardless of the deployment method, the accuracy of monitoring point numbers, time bases, displacement units, coordinate units, unified reference positions, historical boundaries, and comprehensive discrimination values ​​remains consistent.

[0199] To further illustrate the operation of this invention, the calculation steps in S1 to S4 are described below using a specific application scenario. The data provided is for demonstrating the processing logic and is not intended as a fixed monitoring threshold for other projects.

[0200] The target monitoring area in a deep road cut has a longitudinal length of 32.0 m and a transverse width of 8.0 m, with 12 monitoring points set up. The planar coordinates of each monitoring point are (0,4) m, (3,0) m, (6,8) m, (9,0) m, (12,8) m, (15,0) m, (17,8) m, (20,0) m, (23,8) m, (26,0) m, (29,8) m, and (32,4) m, respectively. The horizontal span of the monitoring points is 32.0 m.

[0201] In S1, the sampling interval is 10 min, the sampling time tolerance is 60 s, the maximum allowed interpolation time interval is 20 min, the time window length is 60 min, the time window sliding step is 30 min, and the minimum effective monitoring point ratio is 80%. The current monitoring cycle lasts for 30 days, and each monitoring point has a total of 4321 required sampling times, including the start and end times.

[0202] The total number of displacement values ​​that should exist for all monitoring points is 4321 × 12 = 51852. There are actually 48 missing original data points, with a missing percentage of 48 / 51852 = 9.2571 × 10⁻⁶. -4 Of the missing data, 36 met the 20-minute interpolation criteria. The remaining 12 missing data were distributed across different monitoring points and time windows, ensuring that the number of valid monitoring points in any time window did not fall below 10. The current monitoring period comprises 1439 time windows.

[0203] After sorting the 12 base normal displacement values ​​at the end of a certain time window, the two middle values ​​are 4.05 mm and 4.09 mm, and the median value of the window displacement is 4.07 mm. Within the same time window, the two middle values ​​of the base normal displacement rate at each monitoring point are 0.20 mm / h and 0.22 mm / h, and the median value of the window displacement rate is 0.21 mm / h.

[0204] In S2, the calibration historical period and the verification historical period each lasted 30 days, forming 1439 candidate historical windows each. A single time window contains 7 expected sampling times, with 12 monitoring points corresponding to 84 expected raw data points. In one candidate historical window, 82 valid raw data points were actually obtained, with 2 missing data points.

[0205] The median displacement values ​​at the seven sampling times of this candidate historical window are 2.10 mm, 2.12 mm, 2.15 mm, 2.17 mm, 2.16 mm, 2.18 mm, and 2.19 mm, respectively, with a direction flip of 2. The two data points in the middle of the absolute difference at the end of the window are 0.06 mm and 0.07 mm, and the median absolute deviation of the displacement at the end of the window is 0.065 mm.

[0206] The historical segments for calibration and verification underwent three levels of screening, reducing the number of candidate windows from 1,439 to 720, 360, and 180 respectively.

[0207] The median displacement of the calibration history window has a minimum value of 1.80 mm and a maximum value of 4.60 mm. The median difference between non-zero adjacent values ​​is 0.02 mm. Therefore, the displacement history boundary ranges from 1.78 mm to 4.62 mm.

[0208] The median displacement rate of the window has a minimum value of -0.12 mm / h and a maximum value of 0.18 mm / h. The median difference between non-zero adjacent values ​​is 0.01 mm / h. Therefore, the historical boundary of the displacement rate is from -0.13 mm / h to 0.19 mm / h.

[0209] The median duration of all historical candidate positive continuous change segments is 1.0 h, therefore the reference duration is 1.0 h.

[0210] In S3, the base normal displacement values ​​at the end of the 6 consecutive time windows of the 4th monitoring point were 4.30 mm, 4.42 mm, 4.58 mm, 4.77 mm, 4.96 mm and 5.14 mm, respectively, forming a continuous positive change segment lasting 2.5 h.

[0211] The original overlapping time periods between monitoring points 4 and 5, 5 and 6, and 6 and 7 are merged through spatiotemporal connectivity to form a spatiotemporal connectivity overlapping segment from 08:30 to 10:30.

[0212] Within the four consecutive judgment time intervals of the spatiotemporal interconnected overlapping segment, the set of monitoring points in the positive continuous change segment are, in order, the 2nd to 6th monitoring points, the 3rd to 7th monitoring points, the 4th to 8th monitoring points, and the 5th to 10th monitoring points.

[0213] The corresponding activity monitoring point coordinates are (9.0, 3.2) m, (11.8, 4.8) m, (14.6, 3.2) m, and (18.8333, 4.0) m, respectively. The adjacent migration distances are 3.2249 m, 3.2249 m, and 4.3083 m, respectively, with a cumulative migration length of 10.7581 m for the activity center. The activity center migration ratio is 10.7581 / 32 = 0.3362. A total of 9 different monitoring points participated in the positive continuous change during this overlapping segment. All 12 monitoring points had valid data at the end of all windows in the overlapping segment; therefore, the positive change monitoring point participation ratio is 0.75. The longest original overlap period lasted 1.5 h, and the overlapping segment lasted 2.0 h, with a time overlap ratio of 0.75. The base value for same-direction overlap is 0.75, which is greater than the activity center migration ratio of 0.3362; therefore, the comprehensive value for same-direction overlap is 0.75.

[0214] In S4, the median window displacements of the five time windows covered by the overlapping segment are 4.50 mm, 4.72 mm, 4.90 mm, 5.12 mm, and 5.36 mm, respectively. The corresponding median window displacement rates are 0.16 mm / h, 0.23 mm / h, 0.31 mm / h, 0.40 mm / h, and 0.46 mm / h, respectively. The displacement exceeding the limit percentages are 0, 0.0352, 0.0986, 0.1761, and 0.2606, respectively, and the displacement rate exceeding the limit percentages are 0, 0.1250, 0.3750, 0.6563, and 0.8438, respectively.

[0215] The window out-of-bounds ratios were 0, 0.1250, 0.3750, 0.6563, and 0.8438, respectively. The out-of-bounds value for the continuously varying segment was 0.3750. The comprehensive upward arch discrimination value was 0.3750 × 0.75 = 0.2813. Verification showed that the maximum historical comprehensive upward arch discrimination value generated from historical data was 0.16, the median difference between adjacent historical comprehensive upward arch discrimination values ​​was 0.02, and the comprehensive upward arch discrimination boundary was 0.18.

[0216] The current comprehensive arching discrimination value of 0.2813 exceeds the comprehensive arching discrimination boundary of 0.18, corresponding to a spatiotemporal connectivity overlapping segment lasting for 2.0 h, reaching the reference duration of 1.0 h. Therefore, the base arching warning result is output.

[0217] The preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the specific details of the above embodiments. Within the scope of the technical concept of the present invention, various simple modifications can be made to the technical solution of the present invention, and these simple modifications all fall within the protection scope of the present invention.

[0218] It should also be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the present invention will not describe the various possible combinations separately.

[0219] Furthermore, various different embodiments of the present invention can be combined in any way, as long as they do not violate the spirit of the present invention, they should also be regarded as the content disclosed by the present invention.

[0220] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for processing monitoring data on the arching of a deep roadbed foundation, characterized in that, The methods include: Obtain the base normal displacement value, sampling time and plane coordinates of each monitoring point, and generate the median value of window displacement and the median value of window displacement rate according to the time window. Before the current monitoring cycle, set up calibration historical segments and verification historical segments, and filter candidate historical windows in the two historical segments according to the number of missing windows, the number of direction flips, and the median absolute deviation of displacement at the end of the window. Generate displacement history boundaries, displacement rate history boundaries, and reference duration based on the calibration history window; Extract the positive continuous change segments that reach the reference duration for each monitoring point, and merge the original overlapping time periods corresponding to the positive continuous change segments of spatially adjacent monitoring points into spatiotemporally connected overlapping segments according to temporal and spatial connectivity. Based on the participation degree of monitoring points, time overlap degree, and coordinate center migration degree of the active monitoring points in the spatiotemporal interconnected overlapping segments, a comprehensive value of overlapping in the same direction is generated. Based on the degree of exceeding the limits of the median value of window displacement and the median value of window displacement rate, a limit value of continuously changing segments is generated. Based on the comprehensive value of overlapping in the same direction and the limit value of continuously changing segments, a comprehensive upward arching discrimination value is obtained. A comprehensive arching discrimination boundary is generated based on the verification history window. When the current comprehensive arching discrimination value exceeds the comprehensive arching discrimination boundary and the duration of the corresponding spatiotemporal connectivity overlapping segment reaches the reference duration, an early warning result for the base arching is output.

2. The data processing method for monitoring arching of deep roadbed foundation according to claim 1, characterized in that, The median value of the window displacement and the median value of the window displacement rate are generated according to the following steps: The base normal displacement values ​​of each effective monitoring point at the end of the time window are arranged in ascending order of numerical value, and the median value is used as the median value of the window displacement. For each monitoring point that has a valid base normal displacement value at both the start and end of the time window, the base normal displacement rate of the corresponding monitoring point is determined based on the base normal displacement value at the end of the time window, the base normal displacement value at the start of the time window, and the duration of the time window. The base normal displacement rate of each monitoring point within the same time window is arranged in ascending order of numerical value, and the median value is used as the median value of the displacement rate of the window. The median displacement and median displacement rate of each window are arranged in chronological order of the end time of the time window. When the number of data points involved in the sorting is even, the median is the average of the two data points in the middle of the sorted position.

3. The data processing method for monitoring arching of deep roadbed foundation according to claim 2, characterized in that, The candidate history window is filtered according to the following steps: The number of missing values ​​is determined based on the difference between the expected number of original valid displacement values ​​in the candidate history window and the actual number of original valid displacement values. A window displacement sequence is formed based on the median displacement value at each sampling time within the candidate historical window. After excluding zero differences between adjacent median displacement values, the number of conversions between positive and negative differences is counted to obtain the number of direction reversals. Determine the absolute difference between the effective base normal displacement value at the end of the candidate historical window and the median displacement at that end time, and use the median of all absolute differences as the absolute deviation of the median displacement at the end of the window. Candidate historical windows are sorted in ascending order by the number of missing windows, the number of orientation flips, and the median absolute deviation of displacement at the end of the window. After each sorting, the first half of the candidate historical windows are retained to obtain calibration historical windows and verification historical windows respectively. When the number of candidate historical windows participating in the sorting is odd, the number to be retained is determined by rounding up. When multiple candidate historical windows located at the retention boundary have the same sorting data, the candidate historical window whose end time is closer to the start time of the current monitoring period is retained first. If they are still the same, the retention order is determined by the ascending order of the candidate historical window numbers.

4. The data processing method for monitoring arching of deep roadbed foundation according to claim 3, characterized in that, The displacement history boundary, the displacement rate history boundary, and the reference duration are generated according to the following steps: Arrange the median values ​​of all calibrated historical window displacements in ascending order of numerical value. Use the minimum and maximum values ​​as the initial lower boundary and initial upper boundary of displacement, respectively. Use the median of the difference between adjacent and different median values ​​of window displacement as the displacement boundary expansion amount. Extend the initial lower boundary and the initial upper boundary of displacement downward and upward, respectively, to obtain the historical lower boundary and historical upper boundary of displacement. Following the same processing procedure, the lower boundary and upper boundary of the displacement rate history are generated based on the median value of the displacement rate of all calibration history windows. When there are no adjacent data with different values, the smallest distinguishable unit of the corresponding data is used as the boundary expansion amount. Arrange all calibration history windows according to their end times, and connect adjacent calibration history windows with a sliding step of one time window between their end times to form a continuous calibration history window sequence; For each monitoring point within each continuous calibration historical window sequence, the time intervals covered by the continuous positive displacement differences between the end times of adjacent windows are merged into historical candidate positive continuous change segments, and the median of the duration of all historical candidate positive continuous change segments is used as the reference duration.

5. The data processing method for monitoring arching of deep roadbed foundation according to claim 4, characterized in that, The positive continuous change segment is generated according to the following steps: A Deloitte triangulation network is established based on the planar coordinates of each monitoring point, and two monitoring points connected by the same edge of the triangulation network are identified as spatially adjacent monitoring points. When monitoring points cannot form a Deloitte triangulation, they are arranged in order of their coordinates in the direction with the larger plane span, their coordinates in the other direction, and their monitoring point numbers. Two monitoring points with adjacent positions are then identified as spatially adjacent monitoring points. For each monitoring point, the displacement difference is calculated only when the time interval between the end times of adjacent time windows is equal to the sliding step size of a time window, and both end times have effective base normal displacement values. The time interval corresponding to the positive displacement difference is recorded as the positive change interval. The time-continuous positive change intervals are merged into candidate positive continuous change segments. Negative displacement difference, zero displacement difference, missing value, or time gap interrupts the candidate positive continuous change segment. Candidate positive continuous change segments whose duration reaches the reference duration are identified as positive continuous change segments.

6. The data processing method for monitoring arching of deep roadbed foundation according to claim 5, characterized in that, The spatiotemporal connectivity overlapping segments and the overall overlapping value are generated according to the following steps: For each pair of spatially adjacent monitoring points, the later start time of the two positive continuous change segments is taken as the start time of the original overlapping time period, and the earlier end time is taken as the end time of the original overlapping time period. The original overlapping time period is generated when the start time is earlier than the end time. The original overlapping time periods that overlap in time or are adjacent at the boundary, and whose corresponding monitoring point pairs have common monitoring points or can form a continuous adjacent path through other spatially adjacent monitoring point pairs, are merged into spatiotemporally connected overlapping segments. From the monitoring points that have effective base normal displacement values ​​at the end of all windows of the spatiotemporal interconnected overlapping segment, the number of different monitoring points in the positive continuous change segment within the overlapping segment is counted to determine the participation ratio of positive change monitoring points. The time overlap ratio is determined based on the longest original overlapping time period and the longest spatiotemporal connectivity co-directional overlapping segment, and the smaller value between the participation ratio of the positive change monitoring point and the time overlap ratio is used as the co-directional overlap base value. The average position of the plane coordinates of the monitoring points in the positive continuous change segment within each judgment time interval is determined according to the time sequence to obtain the coordinate center of the active monitoring point. The distance between the coordinate centers of adjacent active monitoring points with continuous time is accumulated to obtain the cumulative migration length of the active center. The migration ratio of the activity center is determined based on the cumulative migration length of the activity center and the maximum planar distance between monitoring points. The migration ratio of the activity center that exceeds 1 is recorded as 1, and the larger value between the basic value of the same-direction overlap and the migration ratio of the activity center is used as the comprehensive value of the same-direction overlap.

7. The method for processing monitoring data of deep roadbed arching according to claim 6, characterized in that, The comprehensive arch discrimination value is generated according to the following steps: When the median value of the window displacement does not exceed the historical boundary of the displacement, the displacement over-limit ratio is set to 0; when it exceeds the historical boundary of the displacement, the displacement over-limit ratio is determined based on the amount of displacement exceeding the historical boundary of the displacement and the distance between the upper and lower boundaries of the displacement history. When the median value of the window displacement rate does not exceed the historical boundary of the displacement rate, the displacement rate over-limit ratio is set to 0; when it exceeds the historical boundary of the displacement rate, the displacement rate over-limit ratio is determined based on the amount of the displacement rate exceeding the historical boundary and the distance between the upper and lower boundaries of the displacement rate history. The larger of the displacement over-limit ratio and the displacement rate over-limit ratio corresponding to the same time window is used as the window over-limit ratio. The median of the proportion of all windows that cross the boundary covered by a spatiotemporally connected overlapping segment in the same direction is taken as the boundary value of the continuously changing segment. Multiply the super-boundary value of the continuously changing segment by the comprehensive value of the same-direction overlapping segment to obtain the comprehensive upward arch discrimination value.

8. The data processing method for monitoring arching of deep roadbed foundation according to claim 7, characterized in that, The comprehensive upper arch discrimination boundary is generated according to the following steps: Arrange all verification history windows according to their end times, and connect adjacent verification history windows with a sliding step of one time window between their end times to form a continuous verification history window sequence. The displacement history boundary, displacement rate history boundary and reference duration generated by the calibration history window are used to process each continuous verification history window sequence to obtain the historical comprehensive arching discrimination value. When a continuous verification history window sequence does not form a spatiotemporally connected overlapping segment, the historical comprehensive upward arch discriminant value corresponding to the continuous verification history window sequence is recorded as 0; when multiple spatiotemporally connected overlapping segments are formed, each spatiotemporally connected overlapping segment generates a historical comprehensive upward arch discriminant value. All historical composite arch discrimination values ​​are arranged in ascending order of numerical value. The maximum value is used as the initial discrimination boundary. The median of the difference between adjacent historical composite arch discrimination values ​​with different values ​​is used as the discrimination boundary expansion amount. The initial discrimination boundary is increased by the discrimination boundary expansion amount to obtain the composite arch discrimination boundary. When there are no adjacent historical comprehensive discriminant values ​​with different values, the minimum recording unit of the comprehensive discriminant value set in advance is used as the discriminant boundary expansion amount.

9. The data processing method for monitoring arching of deep roadbed foundation according to claim 8, characterized in that: Whenever a new time window ends and valid monitoring data is obtained, update each positive continuous change segment and each spatiotemporally connected overlapping segment in the same direction; For the unfinished positive continuous change segment and the spatiotemporally connected overlapping segment, the latest time window end time is taken as the provisional end time. When there are multiple spatially disconnected spatiotemporally connected overlapping segments in the same direction within the same time range, calculate the comprehensive upward arch discrimination value of each spatiotemporally connected overlapping segment in the same direction. When any current comprehensive arching discrimination value exceeds the comprehensive arching discrimination boundary, and the duration of the corresponding spatiotemporal interconnected overlapping segment reaches the reference duration, the base arching early warning result is output, including the monitoring area, the monitoring points involved, the start time, the end time, the comprehensive arching discrimination value, and the comprehensive arching discrimination boundary.

10. A data processing system for monitoring the arching of a deep roadbed foundation, characterized in that, It includes a data conversion module, a historical calibration module, a spatiotemporal overlap processing module, and a discrimination output module; The data conversion module is used to obtain the base normal displacement value, sampling time and plane coordinates of each monitoring point, and generate the median value of window displacement and the median value of window displacement rate according to the time window. The historical calibration module is used to divide the historical segment before the current monitoring cycle into calibration historical segment and verification historical segment. The candidate historical windows in the two historical segments are filtered step by step according to the number of missing windows, the number of direction flips, and the median absolute deviation of displacement at the end of the window. Based on the calibration historical window, the displacement historical boundary, the displacement rate historical boundary, and the reference duration are generated. The spatiotemporal overlap processing module is used to generate positive continuous change segments for each monitoring point, form spatiotemporal connected overlapping segments based on the original overlapping time periods of the positive continuous change segments of spatially adjacent monitoring points and their spatiotemporal connectivity, and generate a comprehensive overlapping value based on the participation ratio of positive change monitoring points, the time overlap ratio, and the migration ratio of the activity center. The discrimination output module is used to generate a comprehensive upward arch discrimination value based on the median value of window displacement, the median value of window displacement rate, the corresponding historical boundary and the comprehensive value of the same direction overlap, generate a comprehensive upward arch discrimination boundary based on the verification historical window, and output a base upward arch warning result when the current comprehensive upward arch discrimination value exceeds the comprehensive upward arch discrimination boundary and the corresponding spatiotemporal connectivity same direction overlap segment reaches the reference duration. The data conversion module, the history calibration module, the spatiotemporal overlap processing module, and the discrimination output module are implemented by the processor executing program instructions in the memory.

Images