High-precision beidou monitoring system for slope displacement

Abstract

The application relates to the technical field of slope engineering monitoring, in particular to a high-precision Beidou monitoring system for slope displacement. The system solves the technical problem that the existing method cannot effectively distinguish different types of data anomalies, which seriously affects the reliability of slope stability early warning. The system comprises a matrix construction unit, which is used for constructing a matching error matrix containing an asymmetric time sequence penalty item according to a surface displacement change sequence and a deep layer equivalent displacement sequence; a matching distribution unit, which is used for distributing a deep layer equivalent displacement for each surface displacement change in the surface displacement change sequence according to the matching error matrix, forming multiple distribution combinations, and outputting a matching error value of each distribution combination; and a filling unit, which is used for generating a filling value and writing the filling value into a data vacancy position according to the deep layer equivalent displacement of the eliminated surface displacement change and a surface layer deformation amplification multiple. The application is used for the scene of slope displacement monitoring.

Description

Technical Field

[0001] This invention relates to the field of slope engineering monitoring technology, specifically to a high-precision BeiDou monitoring system for slope displacement. Background Technology

[0002] In the construction and operation of transportation infrastructure such as highways and railways, geological hazard prevention in slope engineering is a crucial aspect of ensuring the safe operation of the lines. To promptly detect early signs of slope instability, long-term stability monitoring of slopes is necessary in engineering practice. High-precision BeiDou receivers are widely deployed on slope surfaces to acquire three-dimensional displacement information; simultaneously, to understand changes in deeper soil layers, engineers drill boreholes inside the slope to install inclinometers and obtain deep dip angle data. Multi-source sensor collaborative observation has become a commonly adopted slope monitoring technology in the industry.

[0003] In field slope monitoring scenarios, BeiDou receivers and deep inclinometers deployed at the front end collect surface displacement and deep dip angle change data, respectively, and transmit the data back to a cloud server for fusion analysis via a remote communication link. Existing data processing methods typically compare the surface and deep displacement data collected at the same time directly, then discard surface data points with significant discrepancies. However, in dense vegetation environments, tree canopies can obstruct BeiDou signals, causing multipath reflections of electromagnetic waves and resulting in anomalous jumps in the coordinate data output by the receiver. Furthermore, the transmission of deep soil deformation to the surface has an objective time lag of several hours to several days. These factors lead to low accuracy and reliability issues in traditional data fusion processing methods in practical applications.

[0004] Existing slope monitoring data fusion methods are unable to effectively distinguish different types of data anomalies. They are prone to misjudging real deformation information as noise and discarding it, or mistaking occasional electromagnetic interference as valid deformation. The final output monitoring displacement sequence may lose key landslide precursor information or contain a large amount of false abrupt change data, which seriously affects the reliability of slope stability early warning. Summary of the Invention

[0005] To address the technical problems of existing slope monitoring data fusion methods, which struggle to effectively distinguish between different types of data anomalies, easily misclassify genuine deformation information as noise and discard it, or mistake occasional electromagnetic interference for valid deformation, resulting in the loss of crucial landslide precursor information or the inclusion of a large amount of false abrupt changes in the output monitoring displacement sequence, severely impacting the reliability of slope stability early warning, this invention aims to provide a high-precision BeiDou monitoring system for slope displacement. The specific technical solution adopted is as follows: In a first aspect, the present invention provides a high-precision BeiDou monitoring system for slope displacement. The system includes: a data acquisition unit for acquiring a sequence of surface displacement changes and a sequence of deep equivalent displacements within a sliding time window; the deep equivalent displacement is determined based on the tilt angle change of a deep inclinometer and is used to characterize the estimated displacement at the surface caused by deep soil deformation; a matrix construction unit for constructing a matching error matrix containing an asymmetric time-series penalty term based on the surface displacement change sequence and the deep equivalent displacement sequence; a matching allocation unit for allocating a deep equivalent displacement to each surface displacement change in the surface displacement change sequence according to the matching error matrix, forming multiple allocation combinations, and outputting the matching error value of each allocation combination; an anomaly identification unit for comparing each matching error value with a long-term judgment threshold; when the matching error value is greater than the long-term judgment threshold, the corresponding surface displacement change is judged as signal interference abrupt change data and removed; and a filling unit for generating a filling value and writing it into the data gap position based on the deep equivalent displacement of the removed surface displacement change and the surface deformation amplification factor.

[0006] In conjunction with the first aspect mentioned above, in one possible implementation, the data acquisition unit specifically includes: a spatiotemporal alignment subunit, used to perform equal-interval interpolation processing on the BeiDou 3D coordinate data and the deep inclinometer tilt data according to the sampling period of the BeiDou 3D coordinate data and the deep inclinometer tilt data, aligning them to a unified discrete time point sequence on the time axis, and determining the aligned BeiDou 3D coordinate sequence and the aligned deep inclinometer tilt sequence; a surface displacement calculation subunit, used to perform tangential projection calculation based on the aligned BeiDou 3D coordinate sequence and the slope tilt angle, to determine the change in surface displacement at each time point; and a deep displacement conversion subunit, used to perform trigonometric function conversion based on the aligned deep inclinometer tilt sequence and the vertical installation distance, to determine the equivalent deep displacement at each time point.

[0007] In conjunction with the first aspect mentioned above, in one possible implementation, the matrix construction unit specifically includes: a parameter acquisition subunit, used to acquire the satellite positioning accuracy factor sequence and the surface deformation amplification factor; a pairing subunit, used to construct multiple time point pairs based on the first time point corresponding to each surface displacement change within the sliding time window and the second time point corresponding to each deep equivalent displacement; a difference determination subunit, used to determine the difference characterization value of each time point pair based on the degree of difference between the surface displacement change and the product of the deep equivalent displacement and the surface deformation amplification factor in each time point pair; a weight determination subunit, used to acquire the satellite positioning accuracy factor of the first time point in each time point pair; and a penalty term determination subunit. The stator unit is used to determine the asymmetric timing penalty term for each time point pair based on the timing relationship between the first and second time points. When the first time point is later than or equal to the second time point, the asymmetric timing penalty term increases linearly according to the absolute deviation of the difference between the two time points from the preset typical conduction delay. When the first time point is earlier than the second time point, the asymmetric timing penalty term increases exponentially with the increase of the difference between the two time points and has an upper limit cutoff value. The element calculation subunit is used to determine the matching error matrix elements for each time point pair based on the difference characterization value, the satellite positioning accuracy factor, and the asymmetric timing penalty term. The matrix output subunit is used to construct the matching error matrix based on the matching error matrix elements for each time point pair.

[0008] In conjunction with the first aspect mentioned above, in one possible implementation, the matching allocation unit specifically includes: an algorithm execution subunit, used to allocate a unique deep equivalent displacement to each surface displacement change amount based on a preset bipartite graph matching algorithm, using the matching error matrix as the cost matrix, so as to minimize the total cost of all allocation combinations; and an output subunit, used to output the index of the deep equivalent displacement amount allocated to each surface displacement change amount and the matching error value; the matching error value is the element value corresponding to the allocation combination in the matching error matrix.

[0009] In conjunction with the first aspect mentioned above, in one possible implementation, the anomaly detection unit specifically includes: a matching error comparison subunit, used to compare each matching error value with a long-term judgment threshold; a statistics subunit, used to count the number of anomaly triggers where the matching error value is greater than the long-term judgment threshold within a preset number of consecutive time points before the current time point when the matching error value is greater than the long-term judgment threshold; and a judgment subunit, used to determine that the current surface displacement change is signal interference abrupt change data and remove it when the number of anomaly triggers is less than a preset number threshold; and to determine that the current surface displacement change is shallow independent sliding data and retain it when the number of anomaly triggers is greater than or equal to the preset number threshold, while simultaneously outputting a shallow anomaly warning signal.

[0010] In conjunction with the first aspect mentioned above, in one possible implementation, the filling unit specifically includes: a matching extraction subunit, used to determine the deep equivalent displacement that matches the surface displacement change based on the time point corresponding to each removed surface displacement change; a filling value generation subunit, used to generate a filling value for each removed surface displacement change based on the product of the deep equivalent displacement and the surface deformation amplification factor; and a sequence output subunit, used to write the filling value for each removed surface displacement change into the data gap position at the corresponding time point and output the time window corrected change sequence.

[0011] In conjunction with the first aspect mentioned above, in one possible implementation, the system further includes a trajectory reconstruction unit, which specifically includes: a data acquisition subunit, used to acquire the time window correction change sequence and the stored historical absolute displacement coordinates; and an accumulation subunit, used to add the change value of the last time point in the time window correction change sequence to the absolute coordinate of the previous adjacent time point to generate a continuous time axis correction displacement sequence.

[0012] In conjunction with the first aspect mentioned above, in one possible implementation, the system further includes an early warning unit, which specifically includes: a rate extraction subunit, used to extract a first average rate within a first preset time period and a second average rate within a second preset time period based on the corrected displacement sequence; the first preset time period is less than the second preset time period; a sliding rate comparison subunit, used to compare the first average rate with the product of the second average rate multiplied by a preset warning multiple and a preset absolute rate lower limit threshold, respectively; and an early warning triggering subunit, used to generate a slope acceleration sliding early warning command when multiple consecutive sliding time windows meet preset conditions; the preset conditions are: the first average rate is greater than the product of the second average rate multiplied by the preset warning multiple and is greater than the absolute rate lower limit threshold.

[0013] In conjunction with the first aspect mentioned above, in one possible implementation, the system further includes a parameter calibration unit, which specifically includes: a condition detection subunit, used to detect whether the average satellite positioning accuracy factor within a continuous preset duration is less than a preset clean signal threshold and the cumulative rainfall is zero; an interference-free period delineation subunit, used to determine the continuous preset duration as an interference-free observation period when the condition detection subunit detects that the condition is correct; a magnification factor calculation subunit, used to determine the surface deformation magnification factor within the interference-free observation period based on the sum of the absolute values ​​of each surface displacement change and the sum of the absolute values ​​of each deep equivalent displacement; a reference sequence acquisition subunit, used to control the operation of the control matrix construction unit and the matching allocation unit within the interference-free observation period to acquire a reference matching error value sequence; and a threshold calculation subunit, used to determine the long-term judgment threshold based on the product of the preset quantile value and the preset tolerance coefficient in the reference matching error value sequence.

[0014] In conjunction with the first aspect mentioned above, in one possible implementation, the system further includes a data acquisition unit, which specifically includes: a BeiDou data acquisition subunit, used to acquire the three-dimensional coordinate data and satellite positioning accuracy factor sequence output by the BeiDou receiver; an inclination data acquisition subunit, used to acquire the inclination change data output by the deep inclinometer; a rainfall data acquisition subunit, used to acquire rainfall pulse data; and a data output subunit, used to output the three-dimensional coordinate data, satellite positioning accuracy factor sequence, inclination change data, and rainfall pulse data to the data acquisition unit and the matrix construction unit.

[0015] The present invention has the following beneficial effects: This invention achieves cross-epoch feature correlation between slope surface displacement and deep calculated displacement through the collaborative work of a data acquisition unit, a matrix construction unit, a matching and allocation unit, an anomaly identification unit, and a filling unit. Specifically, the matching error matrix, which includes an asymmetric time-series penalty term, quantifies the time lag in the transmission of deep deformation to the surface. The matching and allocation unit finds the most relevant historical deep matching term for each surface displacement change, while the anomaly identification unit, combined with a long-term judgment threshold, effectively distinguishes between multi-path interference abrupt changes and true deformation. Breaking away from the conventional thinking of same-epoch alignment filtering, this invention eliminates electromagnetic interference abrupt data while preserving the true lag deformation characteristics, and utilizes the filling unit to reconstruct data gaps based on deep displacement and deformation amplification factor, significantly improving the accuracy and continuity of slope displacement monitoring sequences. This solves the technical problems of existing slope monitoring data fusion methods, which are unable to effectively distinguish different types of data anomalies, easily misjudge real deformation information as noise and discard it, or mistake occasional electromagnetic interference as valid deformation. As a result, the output monitoring displacement sequence may lose key landslide precursor information or contain a large number of false abrupt changes, which seriously affects the reliability of slope stability early warning. Attached Figure Description

[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of a system architecture for high-precision BeiDou monitoring of slope displacement provided in one embodiment of the present invention. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a high-precision BeiDou monitoring system for slope displacement proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0020] The specific scheme of a high-precision BeiDou monitoring system for slope displacement provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0021] Please see Figure 1 The diagram illustrates a system architecture of a high-precision BeiDou monitoring system 100 for slope displacement according to an embodiment of the present invention. The system includes: The data acquisition unit 101 is used to acquire the land surface displacement change sequence and the deep equivalent displacement sequence within the sliding time window.

[0022] Among them, the deep equivalent displacement is determined based on the change in the inclination angle of the deep inclinometer and is used to characterize the estimated displacement at the ground surface caused by the deformation of the deep soil.

[0023] Optionally, a fixed-length sliding window (e.g., 24 hours) is used as the basic data processing unit on a continuous time axis, and raw observation data within the corresponding time segment is extracted at each window position. For the extracted raw data, the system performs preprocessing of multi-source heterogeneous data to obtain the land surface displacement change sequence and the deep equivalent displacement sequence within the sliding time window.

[0024] It should be understood that the preset fixed-length sliding window of 24 hours is based on the following: multipath interference of satellite signals exhibits a periodic change of approximately 24 hours with the Earth-Sun motion, and this window can completely cover one interference cycle, ensuring that interference characteristics are fully sampled; deep soil deformation usually takes several hours to several days to propagate to the surface, and a 24-hour window can accommodate most delayed scenarios without excessively delaying the response; 24 hours aligns with the daily reporting system for geological disaster monitoring, facilitating engineering implementation and data comparison; the length of the sliding window can be adjusted later based on actual results, and this invention does not limit this adjustment.

[0025] Furthermore, in some embodiments of the present invention, the data acquisition unit 101 specifically includes: a spatiotemporal alignment subunit, used to perform equal-interval interpolation processing on the BeiDou three-dimensional coordinate data and the deep inclinometer tilt data according to the sampling period of the BeiDou three-dimensional coordinate data and the deep inclinometer tilt data, aligning them to a unified discrete time point sequence on the time axis, and determining the aligned BeiDou three-dimensional coordinate sequence and the aligned deep inclinometer tilt sequence; a surface displacement calculation subunit, used to perform tangential projection calculation based on the aligned BeiDou three-dimensional coordinate sequence and the slope tilt angle, and determine the surface displacement change at each time point; and a deep displacement conversion subunit, used to perform trigonometric function conversion based on the aligned deep inclinometer tilt sequence and the vertical installation distance, and determine the deep equivalent displacement at each time point.

[0026] Specifically, the process involves acquiring the 3D coordinate sequence output by the BeiDou receiver and the tilt sequence output by the deep inclinometer. Since the sampling periods of the two types of sensors are typically different (e.g., the BeiDou receiver outputs one coordinate point per second, while the inclinometer outputs one tilt value every five minutes), direct joint analysis at the same time point is not possible. First, the least common multiple of the two sampling periods is calculated as a unified interpolation time step. Then, on the discrete time point sequence determined by this step, equally spaced linear interpolation is performed on the BeiDou 3D coordinates and the deep inclinometer tilt, respectively. For BeiDou coordinate data, the coordinate estimates at intermediate moments are determined using the coordinate values ​​of adjacent real sampling points in a time-proportional manner, forming a 3D coordinate sequence uniformly distributed on the time axis. For tilt data, the tilt estimates at intermediate moments are determined using adjacent real tilt values ​​in a time-proportional manner, forming a uniformly distributed tilt sequence. After interpolation, a strictly aligned BeiDou 3D coordinate sequence and deep inclinometer tilt sequence are obtained on the unified discrete time point sequence, with both sequences having the same time index.

[0027] For example, if the BeiDou sampling period is 1 second and the inclinometer sampling period is 300 seconds, then the least common multiple is 300 seconds. Discrete time points are constructed at 300-second intervals (second 0, second 300, second 600, etc.). An intermediate estimate is calculated between every two adjacent real sampling points according to a linear relationship, thereby transforming the original non-uniform discrete data into a dense time series with equal intervals. The interpolation algorithm can use proportional calculation by connecting adjacent points, or it can use higher-order methods such as cubic splines; this invention does not limit this approach.

[0028] Then, the slope inclination constant of the BeiDou receiver installation point is obtained from local storage or a configuration file. This inclination constant represents the fixed tilt angle of the slope surface relative to the horizontal plane. Based on this inclination constant, a unit tangent vector parallel to the slope surface and pointing in the direction of slope descent is calculated through trigonometric geometric transformation. Then, the aligned BeiDou three-dimensional coordinate sequence output by the spatiotemporal alignment subunit is obtained. The three-dimensional coordinate change of each epoch (i.e., the coordinate difference between adjacent time points) is multiplied by the unit tangent vector to calculate the projection component of the three-dimensional coordinate change of each epoch in the tangential direction of the slope surface, thereby obtaining the change in tangential displacement of the ground surface at each time point.

[0029] Finally, the vertical installation distance constant between the deep inclinometer and the BeiDou receiver on the surface is read from local storage. This constant represents the fixed vertical distance between the shallowest inclinometer installation point and the surface installation point. The aligned deep inclinometer tilt angle sequence output by the spatiotemporal alignment subunit is obtained, which records the tilt angle change of the deep soil at each time point (the deflection angle relative to the initial vertical direction). For the tilt angle change at each time point, a trigonometric function sine relationship is used for conversion: the vertical installation distance is multiplied by the sine value of the tilt angle change to calculate the displacement estimate of the deep soil deformation at the surface at that time point. Since the angle change of the shallow soil is usually small, under a small angle approximation, the sine value can be approximately equal to the radian angle value. This system uses a precise sine function to ensure conversion accuracy. After epoch-by-epoch conversion, the system outputs a deep equivalent displacement sequence. Each value in this sequence represents the absolute value of the corresponding displacement that the surface should produce if the rotational deformation of the deep soil is equivalent to the horizontal displacement of the surface.

[0030] The matrix construction unit 102 is used to construct a matching error matrix containing an asymmetric time series penalty term based on the surface displacement change sequence and the deep equivalent displacement sequence.

[0031] Optionally, obtain the surface displacement change sequence and the deep equivalent displacement sequence within the sliding time window. Pair each first time point in the surface displacement change sequence with each second time point in the deep equivalent displacement sequence across epochs to construct a two-dimensional matrix. During construction, an asymmetric temporal penalty term is introduced: for each pair of first and second time points, the penalty intensity is determined based on their temporal relationship. When the first time point is later than or equal to the second time point, it is considered to conform to the physical transmission law of deep displacement preceding surface displacement, and only a small, linearly increasing penalty is applied. When the first time point is earlier than the second time point, it is considered to violate the transmission law, most likely caused by multipath interference, and a larger penalty with an exponentially increasing upper limit is applied. Based on the degree of difference between the surface displacement change and the deep equivalent displacement in each pair and the aforementioned asymmetric temporal penalty term, determine the matching error value corresponding to the pair and fill it into the corresponding position in the matrix. After traversing all time point pairings, output the complete matching error matrix.

[0032] Further, in some embodiments of the present invention, the matrix construction unit 102 specifically includes: a parameter acquisition subunit, used to acquire a satellite positioning accuracy factor sequence and a surface deformation amplification factor; a pairing subunit, used to construct multiple time point pairs based on a first time point corresponding to each surface displacement change within a sliding time window and a second time point corresponding to each deep equivalent displacement; a difference determination subunit, used to determine the difference characterization value of each time point pair based on the degree of difference between the surface displacement change in each time point pair and the product of the deep equivalent displacement and the surface deformation amplification factor; a weight determination subunit, used to acquire the satellite positioning accuracy factor of the first time point in each time point pair; and a penalty term determination subunit. The stator unit is used to determine the asymmetric timing penalty term for each time point pair based on the timing relationship between the first and second time points. When the first time point is later than or equal to the second time point, the asymmetric timing penalty term increases linearly according to the absolute deviation of the difference between the two time points from the preset typical conduction delay. When the first time point is earlier than the second time point, the asymmetric timing penalty term increases exponentially with the increase of the difference between the two time points and has an upper limit cutoff value. The element calculation subunit is used to determine the matching error matrix elements for each time point pair based on the difference characterization value, the satellite positioning accuracy factor, and the asymmetric timing penalty term. The matrix output subunit is used to construct the matching error matrix based on the matching error matrix elements for each time point pair.

[0033] Specifically, the parameter acquisition subunit reads the surface deformation amplification factor calculated and stored by the parameter calibration unit during the pre-calibration stage from local storage. This constant reflects the deformation amplification characteristics of the current slope surface soil relative to the deeper soil. Simultaneously, it obtains the satellite positioning accuracy factor sequence corresponding to the sliding time window from the data acquisition unit. Each accuracy factor value in this sequence is associated with a corresponding time point in the surface displacement change sequence. The parameter acquisition subunit outputs these two parameters to subsequent subunits for use in calculating the matching error matrix elements.

[0034] The pairing sub-unit acquires the first time points (e.g., N in total) corresponding to the surface displacement change sequence within the sliding time window, and the second time points (also N in total) corresponding to the deep equivalent displacement sequence. The pairing sub-unit combines each first time point with each second time point to form N×N time point pairs. The pairing operation is not limited to the same time index; any first time point can be combined with any second time point, breaking the traditional one-to-one correspondence limitation within the same epoch and establishing a complete pairing space for the subsequent construction of the matching error matrix.

[0035] The difference determination sub-unit is used to obtain the surface displacement change and the deep equivalent displacement for each time point pair. The deep equivalent displacement is multiplied by the surface deformation amplification factor provided by the parameter acquisition sub-unit to obtain the predicted value of the deep displacement, aligned to the surface magnitude. Then, the degree of difference between the surface displacement change and this predicted value is determined. This degree of difference quantifies whether the two match in terms of amplitude: the smaller the difference, the better the amplified deep deformation can explain the actual observed surface displacement change; the larger the difference, the more inconsistent the two are in terms of amplitude, possibly due to multipath interference or independent slip.

[0036] The weighting determination sub-unit involves identifying the time index corresponding to the first time point in each time point pair formed by the paired sub-units, and then extracting the accuracy factor value corresponding to the time index from the acquired satellite positioning accuracy factor sequence. This value serves as the weighting factor for that time point pair, characterizing the degree of interference to the satellite signal at the corresponding time, such as vegetation obstruction and multipath effects.

[0037] The penalty term determines the sub-unit. For each pair of time points, the first and second time point indices are obtained, and their temporal order is compared. If the first time point is later than or equal to the second time point, it indicates that the occurrence of the surface displacement change is not earlier than the occurrence of the deep equivalent displacement, which conforms to the physical law that deep deformation occurs first, is transmitted, and then causes a surface response. Considering that the transmission of deep deformation to the surface requires a certain time lag, the system presets a typical transmission lag L (the unit is the index difference, which can be set according to the soil depth and soil type, for example, 1 / 4 of the total number of epochs in the window). At this time, the penalty term determines the sub-unit by linearly determining a penalty multiplier according to the absolute deviation of the difference between the two time points from this typical transmission lag. This multiplier takes a minimum value of 1 when the lag time is equal to L, and increases slowly as the deviation increases, but generally remains within a small range (maximum not exceeding 2). If the first time point is earlier than the second time point, it indicates that significant changes have occurred on the surface before equivalent displacement has occurred in the deeper layers, violating the normal transmission sequence. This is highly likely a spurious abrupt change caused by external interference such as multipath signal reflection. In this case, the penalty term determines the sub-unit by first calculating the time point difference and pre-setting a safety threshold. If the time point difference is less than or equal to the safety threshold, then the calculation is performed directly. As a penalty multiplier; if the time point difference is greater than the safety threshold, there is no need to calculate the exponent, and the penalty multiplier is directly set to the upper limit. This predictive mechanism avoids numerical overflow that might occur when calculating the exponent first and then taking the minimum value, ensuring the numerical stability of the algorithm within any long time window.

[0038] Preset safety threshold It can be solved To obtain, that is With α=0.1, Taking 100 as an example, In actual calculations, an integer threshold of 46 is used (the unit is the number of sampling intervals, i.e., the index difference). When the index difference at a time point is greater than 46, it is directly assigned the value of 100 without exponential calculation.

[0039] For example, when the first time point is later than or equal to the second time point, the penalty multiplier satisfies the following formula: in, This is the index of the time point where the change in surface displacement occurs (first time point). Index of the time point where the deep equivalent displacement occurs (second time point); The total number of epochs within the sliding time window; This is the difference between the first and second time points; L is the preset typical conduction time lag (unit: index difference, value range: 0 to N-1), representing the theoretical time lag required for deep deformation to conduct to the surface; |ijL| is the absolute deviation between the actual lag time and the typical conduction time lag.

[0040] The preset typical conduction delay L can be estimated in advance based on the slope soil depth, soil type, and empirical formulas, or it can be calibrated through correlation analysis during the initial deployment of the system (e.g., selecting the delay with the largest correlation coefficient during the interference-free period as L). For example, when the total number of time points in the sliding window is 144 (corresponding to 24 hours), if the site soil parameters show a conduction delay of approximately 6 hours, then L can be taken as 36. In actual engineering, L can also be set to half the window length or determined based on field tests; this invention does not limit this.

[0041] For example, when the first time point is earlier than the second time point, the penalty multiplier satisfies the following formula: in, It is an exponential function with the natural constant e as its base, used to realize that the penalty value grows exponentially with the time difference; This is a function that takes the minimum value and is used to truncate the upper limit of the exponential result; Set a preset maximum penalty limit (e.g., 100) to prevent excessively large exponential operation results from causing floating-point overflow. The preset scaling factor (e.g., 0.1) is dimensionless and is used to smooth out the excessive expansion of the exponential function under a long window, so that the penalty value varies within a reasonable range.

[0042] The overall computational logic ensures that: for slightly time-reversed combinations, the exponential function can rapidly amplify the error multiplier, forcing it to expose extremely high matching error values ​​in the matrix; simultaneously, through Scaling and The truncation protection eliminates the problem from a computer engineering perspective when the time window spans ( When it is large The risk of floating-point number out-of-bounds deadlock and crash caused by operations ensures that the two-dimensional matrix has a computable real value under all epoch combinations.

[0043] The preset maximum penalty limit of 100 is set to prevent numerical overflow during exponential operations when the time window is large, and to ensure that a sufficiently large penalty multiplier is applied to combinations that violate the time order (compared to the maximum normal linear penalty of about 2, 100 times is sufficient to suppress them), so that they are effectively excluded from the matching cost matrix. At the same time, it avoids unnecessary loss of calculation precision due to an excessively large upper limit. It can be adjusted based on the actual scenario, and this invention does not limit it.

[0044] The element calculation subunit receives the difference determination subunit output difference representation value, the weight determines the satellite positioning accuracy factor output by the subunit, and the penalty term determines the asymmetric temporal penalty term output by the subunit. The difference representation value is multiplied by the satellite positioning accuracy factor and the asymmetric temporal penalty term, and the product is used as an element of the matching error matrix for that time point pair. This element comprehensively reflects the overall matching cost of surface displacement change and deep equivalent displacement in three dimensions: amplitude difference, satellite signal reliability, and temporal conformity. A larger difference representation value, a higher satellite positioning accuracy factor, and a heavier temporal penalty result in a larger matching error element value, indicating that the surface and deep data corresponding to that time point pair are less likely to be associated with the same physical slip event.

[0045] For example, the elements in the i-th row and j-th column of the matching error matrix satisfy the following formula: in, To match the elements in the i-th row and j-th column of the error matrix (row index i corresponds to the first time point, and column index j corresponds to the second time point), the feature difference between surface epoch i and deep epoch j is characterized. The larger the value, the more mismatched the two are. This represents the change in surface displacement during the i-th epoch within the sliding time window. Let be the deep equivalent displacement in the j-th deep epoch; This represents the magnification factor of the surface soil deformation. The deep displacement is amplified to the surface level to obtain the predicted displacement value of the deep layer on the surface; For numerical difference terms, if the measured displacement at the surface is close to the predicted value after magnification at the deep layer, the smaller this term is, the more matched the two are in terms of amplitude. This is a penalty term for asymmetric time propagation misalignment. is the satellite positioning accuracy factor for the i-th surface epoch. The larger the value, the worse the satellite spatial geometry distribution (the more severe the vegetation obstruction), and the higher the amplification of the positioning error.

[0046] Multiplying the three terms together, if any one of them is too large, it will lead to... The multiplicative fusion method significantly increases the cost of matching. Any anomaly in any of the three dimensions—numerical mismatch, timing violation, or poor signal quality—makes the combination a high-cost match, which is easily excluded or forced to match irrelevant deep epochs in bipartite graph assignment.

[0047] The matrix output subunit pre-creates a two-dimensional array of size N rows and N columns, where N is the total number of time points within the sliding time window. For each time point pair generated by the pairing subunit, the matrix output subunit determines the row number based on the index of the first time point in the pair and the column number based on the index of the second time point, then fills the corresponding positions in the array with the elements of the matching error matrix output by the element calculation subunit. After all time point pairs have been traversed, the two-dimensional array is completely filled, forming the matching error matrix.

[0048] The matching and allocation unit 103 is used to allocate deep equivalent displacement to each surface displacement change in the surface displacement change sequence according to the matching error matrix, forming multiple allocation combinations, and outputting the matching error value of each allocation combination.

[0049] Optionally, the matching allocation unit 103 receives a matching error matrix. The rows of this matrix correspond to the surface displacement changes within the sliding time window, and the columns correspond to the deep equivalent displacements. Each element represents the comprehensive matching cost between the corresponding surface time point and the deep time point. Based on the matching error matrix, the matching allocation unit 103 performs a globally optimal allocation calculation: assigning a unique corresponding deep equivalent displacement to each surface displacement change, while ensuring that each deep equivalent displacement is assigned to at most one surface displacement change, and minimizing the overall cost of all allocation combinations. The matching allocation unit 103 outputs the index of the deep equivalent displacement assigned to each surface displacement change, and the matching error value corresponding to that allocation combination in the matching error matrix.

[0050] Furthermore, in some embodiments of the present invention, the matching allocation unit 103 specifically includes: an algorithm running subunit, used to allocate a unique deep equivalent displacement amount to each surface displacement change amount according to a preset bipartite graph matching algorithm, using the matching error matrix as the cost matrix, so as to minimize the total cost of all allocation combinations; and an output subunit, used to output the index of the deep equivalent displacement amount allocated to each surface displacement change amount and the matching error value; the matching error value is the element value corresponding to the allocation combination in the matching error matrix.

[0051] Specifically, the algorithm's running subunit receives the matching error matrix constructed by the matrix output subunit and inputs this matrix as the cost matrix into the preset bipartite graph matching algorithm. The bipartite graph matching algorithm treats the row nodes of the matching error matrix as surface displacement changes and the column nodes as deep equivalent displacements. With the goal of minimizing the total cost of all allocation combinations, it assigns a unique corresponding deep equivalent displacement to each surface displacement change, while ensuring that each deep equivalent displacement is assigned to at most one surface displacement change.

[0052] The output sub-unit receives the globally optimal allocation scheme output by the algorithm. For each surface displacement change, it extracts the index of the deep equivalent displacement to which it is allocated, as well as the matching error value of the allocation combination in the matching error matrix.

[0053] The anomaly identification unit 104 is used to compare each matching error value with the long-term judgment threshold. When the matching error value is greater than the long-term judgment threshold, the corresponding surface displacement change is judged as signal interference abrupt change data and removed.

[0054] Optionally, the anomaly identification unit 104 acquires the matching error value corresponding to each surface displacement change and the long-term determination threshold. Each matching error value is compared with the long-term determination threshold: if the matching error value is less than or equal to the long-term determination threshold, the corresponding surface displacement change is retained; if the matching error value is greater than the long-term determination threshold, the corresponding surface displacement change is determined as signal interference abrupt change data and is removed.

[0055] Furthermore, in some embodiments of the present invention, the anomaly identification unit 104 specifically includes: a matching error comparison subunit, used to compare each matching error value with a long-term determination threshold; a statistics subunit, used to count the number of abnormal triggers where the matching error value is greater than the long-term determination threshold within a preset number of consecutive time points before the current time point when the matching error value is greater than the long-term determination threshold; and a determination subunit, used to determine that the current surface displacement change is signal interference mutation data and perform elimination when the number of abnormal triggers is less than a preset number threshold; and to determine that the current surface displacement change is shallow independent sliding data and retain it when the number of abnormal triggers is greater than or equal to the preset number threshold, while outputting a shallow anomaly warning signal.

[0056] Specifically, the matching error comparison subunit receives the matching error value corresponding to each change in surface displacement output by the matching allocation unit 103, and at the same time obtains the long-term judgment threshold predetermined by the parameter calibration unit. The matching error comparison subunit compares each matching error value with the long-term judgment threshold one by one to determine whether the matching error value of the current change in surface displacement is greater than the long-term judgment threshold, and outputs the comparison result to the statistics subunit.

[0057] For example, a specific example of a long-term judgment threshold is 2.4 mm. This threshold is set based on the statistical distribution of the baseline matching error value sequence during the interference-free observation period. The 95th quantile is taken as the upper limit of noise in a clean environment, and then multiplied by a tolerance coefficient (usually 1.5~3.0) to obtain the judgment threshold for daily monitoring. For example, if the 95th quantile of the clean period is 1.2, and a tolerance of 2.0 is used, then the threshold is 2.4. This logic ensures that normal environmental fluctuations do not trigger rejection, while multipath interference (matching errors usually far exceeding the threshold) is effectively identified. The long-term judgment threshold can be modified according to the actual scenario; this invention does not limit this adjustment.

[0058] The statistics subunit receives the comparison results output by the matching error comparison subunit. When the matching error value at a certain time point is greater than the long-term judgment threshold, the statistics subunit marks that time point as an abnormal trigger point. Then, with the current time point as the endpoint, it traces back a preset number of time points (e.g., 3 consecutive time points) in the historical direction and counts the total number of times that the points are marked as abnormal trigger points within the continuous time period.

[0059] The determination subunit compares the total number of times anomaly trigger points are marked within a continuous time period with a preset threshold. If the number of anomalous triggers is less than the preset threshold (e.g., only one isolated trigger), the determination subunit determines that the current surface displacement change is interference-induced abrupt change data caused by multipath signal reflection and performs a removal operation, removing the surface displacement change at that time point from the sequence. If the number of anomalous triggers is greater than or equal to the preset threshold (e.g., three consecutive triggers), the determination subunit determines that the current surface displacement change is shallow independent sliding data (i.e., no deep layer sliding but localized collapse in the shallow layer), retains it, does not perform a removal operation, and simultaneously outputs a shallow layer anomaly warning signal to the management terminal to remind engineering personnel to pay attention to the risk of shallow layer instability of the slope.

[0060] The filling unit 105 is used to generate a filling value and write it into the data gap position based on the deep equivalent displacement of the removed surface displacement change and the surface deformation amplification factor.

[0061] Optionally, the filling unit 105 receives the location information of the data gaps left after the removal of data. For each time point corresponding to the removed surface displacement change, the filling unit 105 obtains the deep equivalent displacement associated with that surface displacement change, as well as the surface deformation amplification factor. Based on the deep equivalent displacement and the surface deformation amplification factor, a filling value is generated and written into the data gap location of the corresponding time point. By traversing all removed time points, the filling unit outputs a complete and gap-free time window correction change sequence.

[0062] Furthermore, in some embodiments of the present invention, the filling unit 105 specifically includes: a matching extraction subunit, used to determine the deep equivalent displacement amount matching the surface displacement amount based on the time point corresponding to each removed surface displacement change amount; a filling value generation subunit, used to generate a filling value for each removed surface displacement change amount based on the product of the deep equivalent displacement amount and the surface deformation amplification factor; and a sequence output subunit, used to write the filling value of each removed surface displacement change amount into the data gap position of the corresponding time point and output the time window corrected change amount sequence.

[0063] Specifically, the matching extraction subunit determines, based on the list of removed surface displacement changes, that each item in the list corresponds to a time point identified as a signal interference abrupt change. For each removed time point, the deep equivalent displacement corresponding to that time point is extracted from the deep equivalent displacement sequence according to the deep equivalent displacement index output by the matching allocation unit 103.

[0064] The fill value generation sub-unit obtains the surface deformation amplification factor calculated and stored by the parameter calibration unit. The deep equivalent displacement is multiplied by the surface deformation amplification factor, and the calculated product is used as the fill value for the removed time point. This fill value represents the reasonable displacement increment that the surface should produce after the amplification effect of the loose surface soil, if the deep soil has undergone equivalent displacement.

[0065] The sequence output subunit receives the fill value corresponding to each removed time point from the output of the fill value generation subunit, as well as the surface displacement change sequence output by the anomaly detection unit 104 (which includes valid data that was not removed and the missing positions left after removal). Each fill value is written into the data missing position of the corresponding time point, replacing the original empty marker. After traversing all removed time points, all gaps in the original sequence are filled, forming a complete and uninterrupted time window corrected change sequence.

[0066] Furthermore, in some embodiments of the present invention, the above system further includes a trajectory reconstruction unit, which specifically includes: a data acquisition subunit, used to acquire the time window correction change sequence and the stored historical displacement absolute coordinates; and an accumulation subunit, used to add the change value of the last time point in the time window correction change sequence to the absolute coordinate of the previous adjacent time point to generate a time axis continuous correction displacement sequence.

[0067] Specifically, the data acquisition subunit reads the stored historical displacement absolute coordinate sequence from the cloud server or local database. The historical displacement absolute coordinates record the cumulative absolute displacement values ​​of the slope surface at various time points up to the start time of the current sliding time window.

[0068] The accumulator subunit receives the time window correction change sequence and historical displacement absolute coordinates output by the data acquisition subunit. It extracts the change value of the last time point from the time window correction change sequence and simultaneously obtains the absolute coordinates corresponding to the previous adjacent time point from the historical displacement absolute coordinates. The accumulator subunit adds the extracted change value to the absolute coordinates of the previous adjacent time point, and the sum is used as the absolute coordinate of the last time point of the current window. By performing the above single-step accumulation operation sequentially for each sliding time window, the accumulator subunit generates a continuous and uninterrupted correction displacement sequence from the start of monitoring to the current moment.

[0069] Furthermore, in some embodiments of the present invention, the above system further includes an early warning unit, which specifically includes: a rate extraction subunit, used to extract a first average rate within a first preset time period and a second average rate within a second preset time period based on the corrected displacement sequence; the first preset time period is less than the second preset time period; a sliding rate comparison subunit, used to compare the first average rate with the product of the second average rate multiplied by a preset warning multiple and a preset absolute rate lower limit threshold, respectively; and an early warning triggering subunit, used to generate a slope acceleration sliding early warning command when multiple consecutive sliding time windows meet preset conditions; the preset conditions are: the first average rate is greater than the product of the second average rate multiplied by the preset warning multiple and is greater than the absolute rate lower limit threshold.

[0070] Specifically, the rate extraction subunit receives a corrected displacement sequence, which records the absolute surface displacement values ​​on a continuous time axis from the start of monitoring to the current moment. Based on a preset first duration (e.g., 24 hours) and a second duration (e.g., 72 hours), the rate extraction subunit extracts continuous data segments of corresponding durations from the current moment in the historical direction. For data segments within the first duration, the displacement difference between the end and start moments is calculated and divided by the first duration to obtain the first average rate. For data segments within the second duration, the second average rate is calculated using the same method.

[0071] It should be understood that the rate extraction sub-unit extracts a first average rate reflecting the recent intensity of sliding and a second average rate reflecting the long-term background creep level by extracting historical windows of different lengths. The short-term window (first duration) is sensitive to recent changes, while the long-term window (second duration) smooths out short-term fluctuations. The combination of the two forms the basis for quantitative indicators to determine whether a slope has shifted from uniform creep to accelerated instability.

[0072] The logic behind setting the first duration (24 hours) and the second duration (72 hours) is as follows: The first duration (24 hours) corresponds to a complete day-night cycle, which can smooth out the diurnal variation effects of satellite multipath interference, while also taking into account the short-term response speed of slope deformation. The 24-hour window length can capture rapid sliding trends induced by rainfall, water level changes, etc., without being affected by single-measurement noise due to being too short. The second duration (72 hours) corresponds to three day-night cycles and is used to obtain the long-term background creep rate of the slope under normal conditions. The 72-hour window can effectively filter out short-term fluctuations (such as temporary expansion and contraction caused by diurnal temperature differences), reflecting the stable sliding background value of the slope under conditions without accelerated instability, serving as a benchmark for comparing whether the recent rate has significantly accelerated. By comparing the recent short-window rate with the long-term background rate by a multiple, it can be determined whether the slope has transitioned from uniform creep to an accelerated sliding stage, while avoiding false alarms caused by short-term disturbances or excessively low background rates. The specific duration can be optimized according to the actual scenario, and this invention does not limit this.

[0073] The sliding rate comparison subunit obtains a preset warning multiple (e.g., 3.0) and an absolute rate lower limit threshold (e.g., 1.0 mm / day). First, it calculates the product of the second average rate and the warning multiple, and then determines whether the first average rate is greater than the product and whether the first average rate is greater than the absolute rate lower limit threshold.

[0074] Warning multiples (e.g., 3.0) are used to determine whether recent slip has accelerated significantly: when the short-term rate exceeds three times the long-term background rate, it indicates that the slope has transitioned from uniform creep to a nonlinear acceleration stage, and instability may occur. Absolute rate lower limits (e.g., 1.0 mm / day) are used to eliminate noise interference in static conditions: when the long-term background rate approaches zero, even if the short-term rate is small (e.g., 0.5 mm / day), the multiple condition may be falsely met. Therefore, the rate must reach the engineering-approved millimeter-level slip lower limit.

[0075] It should be understood that the sliding rate comparison sub-unit executes the core logic of slope instability judgment: the first comparison identifies whether the recent rate significantly exceeds the historical background (multiple relationship), thereby excluding normal fluctuations; the second comparison ensures that the warning will not be falsely triggered due to minor noise when the background rate is extremely low.

[0076] The early warning triggering subunit receives two comparison results from the sliding rate comparison subunit. If both comparison results are true within the current sliding time window (i.e., the first average rate is simultaneously greater than the product of the second average rate multiplied by the warning factor and the absolute rate lower limit threshold), the early warning triggering subunit does not immediately issue an early warning. Instead, it continues to monitor multiple subsequent consecutive sliding time windows (e.g., three consecutive windows, each corresponding to one hour). Only when both of the above conditions are continuously true within a preset number of consecutive sliding time windows does the early warning triggering subunit determine that the slope soil has evolved from a slow, uniform sliding state to accelerated unstable sliding, and then generate an accelerated sliding early warning command. This early warning command encapsulates the current cumulative displacement value, the first average rate value, and the occurrence timestamp, and is sent to the on-site audible and visual alarm terminal and the management console via a communication link.

[0077] Furthermore, in some embodiments of the present invention, the above system further includes a parameter calibration unit, which specifically includes: a condition detection subunit, used to detect whether the average value of the satellite positioning accuracy factor within a continuous preset duration is less than a preset pure signal threshold and the cumulative rainfall is zero; an interference-free period delineation subunit, used to determine the continuous preset duration as an interference-free observation period when the detection result of the condition detection subunit is yes; a magnification factor calculation subunit, used to determine the surface deformation magnification factor within the interference-free observation period based on the sum of the absolute values ​​of each surface displacement change and the sum of the absolute values ​​of each deep equivalent displacement; a reference sequence acquisition subunit, used to control the matrix construction unit 102 and the matching allocation unit 103 to operate within the interference-free observation period and acquire a reference matching error value sequence; and a threshold calculation subunit, used to determine the long-term judgment threshold based on the product of the preset quantile value and the preset tolerance coefficient in the reference matching error value sequence.

[0078] Specifically, during the initial deployment of the system, the condition detection subunit continuously monitors the real-time data stream. For a continuous preset duration (e.g., 72 hours), the condition detection subunit calculates the average satellite positioning accuracy factor at all time points within that duration and compares it with a preset clean signal threshold (e.g., 2.0). Simultaneously, it checks whether the cumulative rainfall during that duration is zero. When both conditions are met—that is, the average satellite positioning accuracy factor is less than the clean signal threshold and the cumulative rainfall is zero—the condition detection subunit marks the continuous preset duration as meeting the interference-free condition and outputs the detection results to the interference-free period delineation subunit.

[0079] The preset clean signal threshold is set according to common standards in the satellite navigation field, and this threshold is set to 2.0. When the satellite positioning accuracy factor (PDOP) is less than 2.0, it indicates that the satellite's spatial geometry is well distributed, the positioning accuracy is high, and the multipath error is small; when the PDOP is greater than 2.0, there may be obstruction or poor geometry. Therefore, 2.0 is used as the dividing line for judging whether the signal is clean or not, to ensure that the BeiDou data quality meets the high-precision calibration requirements during the interference-free observation period.

[0080] The interference-free period delineation subunit defines a continuously preset duration as the interference-free observation period, records the start and end times of this observation period, and notifies the magnification calculation subunit and the reference sequence acquisition subunit to perform subsequent parameter calibration operations within this observation period. During the interference-free observation period, the magnification calculation subunit acquires all surface displacement change sequences and deep equivalent displacement sequences within that time period, calculates the sum of the absolute values ​​of all surface displacement changes, and simultaneously calculates the sum of the absolute values ​​of all deep equivalent displacements, adding a very small positive number (e.g., 0.001) to the denominator to prevent division by zero. The magnification calculation subunit divides the sum of the former by the sum of the latter, and the resulting ratio is the surface deformation magnification factor. The surface deformation magnification factor reflects the inherent amplification characteristics of the loose surface soil to the deep layers under the current slope site conditions. The magnification calculation subunit stores this magnification factor locally for subsequent use by the matrix construction unit 102 and the filling unit 105.

[0081] During the non-interference observation period, the control matrix construction unit 102 and the matching allocation unit 103 of the baseline sequence acquisition subunit operate according to the normal daily monitoring process, that is, to construct the matching error matrix and perform the bipartite optimal matching allocation. Since the data during this period are all pure and interference-free data, the matching error values ​​obtained represent the normal matching cost level of the slope under the condition of no external interference. The baseline sequence acquisition subunit collects all the matching error values ​​output by the matching allocation unit 103 to form a baseline matching error value sequence and outputs it to the threshold calculation subunit.

[0082] It should be understood that the benchmark sequence acquisition subunit obtains the benchmark matching error distribution of the slope site under undisturbed conditions by running the complete matrix construction and matching allocation process during a clean environment period. This sequence reflects the typical level of matching error caused by factors such as normal environmental noise and sensor background error.

[0083] The threshold calculation subunit sorts all values ​​in the baseline matching error value sequence in ascending order and extracts the value located at a preset quantile (e.g., the 95th quantile). This value represents the upper bound of most normal matching errors in a clean environment. The threshold calculation subunit multiplies this quantile value by a preset tolerance coefficient (e.g., a value ranging from 1.5 to 3.0), and the resulting product is the long-term judgment threshold.

[0084] Furthermore, in some embodiments of the present invention, the above system further includes a data acquisition unit, which specifically includes: a BeiDou data acquisition subunit, used to acquire three-dimensional coordinate data and satellite positioning accuracy factor sequence output by the BeiDou receiver; an inclination data acquisition subunit, used to acquire inclination change data output by the deep inclinometer; a rainfall data acquisition subunit, used to acquire rainfall pulse data; and a data output subunit, used to output the three-dimensional coordinate data, satellite positioning accuracy factor sequence, inclination change data, and rainfall pulse data to the data acquisition unit 101 and the matrix construction unit 102.

[0085] Specifically, the BeiDou data acquisition subunit establishes a communication connection with the BeiDou receiver deployed on site, and reads the three-dimensional coordinate data (including east, north, and celestial coordinate values) and the synchronously output satellite positioning accuracy factor sequence in real time according to the receiver's inherent sampling frequency.

[0086] The tilt data acquisition subunit establishes a communication connection with the shallowest inclinometer buried inside the slope, and reads the tilt change data (reflecting the deflection angle of the deep soil relative to the vertical direction) in real time according to the inherent sampling frequency of the inclinometer.

[0087] The rainfall data acquisition subunit establishes a communication connection with the rain gauges deployed on-site, and reads the rainfall pulse signals output by them in real time. Rainfall pulse data is usually provided in the form of cumulative rainfall per unit time or switching signals, used to characterize whether the slope is affected by rainwater infiltration.

[0088] After the data output subunit performs preliminary processing of the above multi-source data according to the timestamp, it outputs the data to the data acquisition unit 101 (for spatiotemporal alignment and tangential projection calculation) and the matrix construction unit 102 (for acquiring the satellite positioning accuracy factor sequence for matching error matrix construction).

[0089] It should be understood that the data output subunit, as the convergence outlet of the data acquisition section, distributes the three types of raw data to the downstream processing unit in a unified manner, thereby decoupling the front-end sensor from the back-end data processing module and ensuring the clarity and scalability of the data flow.

[0090] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0091] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A high-precision BeiDou monitoring system for slope displacement, characterized in that, The system includes: The data acquisition unit is used to acquire the sequence of surface displacement changes and the sequence of deep equivalent displacements within the sliding time window; the deep equivalent displacement is determined based on the dip angle change of the deep inclinometer and is used to characterize the estimated displacement at the surface caused by deep soil deformation. The matrix construction unit is used to construct a matching error matrix containing an asymmetric temporal penalty term based on the surface displacement change sequence and the deep equivalent displacement sequence. The matching and allocation unit is used to allocate deep equivalent displacement to each surface displacement change in the surface displacement change sequence according to the matching error matrix, forming multiple allocation combinations, and outputting the matching error value of each allocation combination. The anomaly identification unit is used to compare each matching error value with the long-term judgment threshold. When the matching error value is greater than the long-term judgment threshold, the corresponding surface displacement change is judged as signal interference abrupt change data and removed. The filling unit is used to generate filling values ​​and write them into the data gap positions based on the deep equivalent displacement of the removed surface displacement changes and the surface deformation amplification factor. The matrix construction unit specifically includes: The parameter acquisition subunit is used to acquire the satellite positioning accuracy factor sequence and the surface deformation magnification factor; Paired sub-units are used to form multiple time point pairs based on the first time point corresponding to each change in surface displacement within the sliding time window and the second time point corresponding to each equivalent deep displacement. The difference determination subunit is used to determine the difference characterization value of each time point pair based on the degree of difference between the change in surface displacement in each time point pair and the product of the deep equivalent displacement and the surface deformation amplification factor. The weight determination subunit is used to obtain the satellite positioning accuracy factor of the first time point in each time point pair; The penalty term determination subunit is used to determine the asymmetric timing penalty term for each time point pair based on the timing relationship between the first time point and the second time point; when the first time point is later than or equal to the second time point, the asymmetric timing penalty term increases linearly according to the absolute deviation of the difference between the two time points from the preset typical conduction delay; when the first time point is earlier than the second time point, the asymmetric timing penalty term increases exponentially with the increase of the difference between the two time points and has an upper limit cutoff value. The element calculation subunit is used to determine the matching error matrix elements of each time point pair based on the difference characterization value, the satellite positioning accuracy factor, and the asymmetric time series penalty term. The matrix output sub-unit is used to construct the matching error matrix based on the matching error matrix elements of each time point pair.

2. The high-precision BeiDou monitoring system for slope displacement according to claim 1, characterized in that, The data acquisition unit specifically includes: The spatiotemporal alignment subunit is used to perform equal-interval interpolation processing on the BeiDou three-dimensional coordinate data and the deep inclinometer tilt data according to the sampling period of the BeiDou three-dimensional coordinate data and the deep inclinometer tilt data, align them to a unified discrete time point sequence on the time axis, and determine the aligned BeiDou three-dimensional coordinate sequence and the aligned deep inclinometer tilt sequence. The surface displacement calculation subunit is used to perform tangential projection calculation based on the aligned BeiDou three-dimensional coordinate sequence and the slope inclination angle to determine the change in surface displacement at each time point. The deep displacement conversion subunit is used to perform trigonometric function conversion based on the aligned deep inclinometer tilt sequence and vertical installation distance to determine the deep equivalent displacement at each time point.

3. The high-precision BeiDou monitoring system for slope displacement according to claim 1, characterized in that, The matching and allocation unit specifically includes: The algorithm execution subunit is used to assign a unique deep equivalent displacement to each surface displacement change according to the preset bipartite graph matching algorithm, using the matching error matrix as the cost matrix, so as to minimize the total cost of all allocation combinations. The output sub-unit is used to output the index of the deep equivalent displacement allocated to each surface displacement change and the matching error value; the matching error value is the element value corresponding to the allocation combination in the matching error matrix.

4. The high-precision BeiDou monitoring system for slope displacement according to claim 1, characterized in that, The anomaly detection unit specifically includes: The matching error comparison subunit is used to compare each matching error value with the long-term judgment threshold. The statistics subunit is used to count the number of abnormal triggers where the matching error value is greater than the long-term judgment threshold within a preset number of consecutive time points before the current time point when the matching error value is greater than the long-term judgment threshold. The determination subunit is used to determine the current surface displacement change as signal interference abrupt change data and remove it when the number of abnormal triggers is less than a preset threshold number; and to determine the current surface displacement change as shallow independent sliding data and retain it when the number of abnormal triggers is greater than or equal to the preset threshold number, while outputting a shallow anomaly warning signal.

5. A high-precision BeiDou monitoring system for slope displacement according to claim 1, characterized in that, The filling unit specifically includes: The matching extraction sub-unit is used to determine the deep equivalent displacement that matches the surface displacement change based on the time point corresponding to each removed surface displacement change. The fill value generation sub-unit is used to generate a fill value for each removed surface displacement change based on the product of the deep equivalent displacement and the surface deformation amplification factor. The sequence output subunit is used to write the fill value of each removed surface displacement change into the data gap position at the corresponding time point, and output the time window corrected change sequence.

6. A high-precision BeiDou monitoring system for slope displacement according to claim 5, characterized in that, It also includes a trajectory reconstruction unit, which specifically includes: The data acquisition subunit is used to acquire the time window correction change sequence and the stored historical absolute displacement coordinates; The accumulator sub-unit is used to add the change value of the last time point in the time window correction change sequence to the absolute coordinate of the previous adjacent time point to generate a continuous correction displacement sequence along the time axis.

7. A high-precision BeiDou monitoring system for slope displacement according to claim 6, characterized in that, It also includes an early warning unit, which specifically includes: A rate extraction subunit is used to extract a first average rate within a first preset time period and a second average rate within a second preset time period based on the corrected displacement sequence; the first preset time period is less than the second preset time period. The sliding rate comparison subunit is used to compare the first average rate with the product of the second average rate multiplied by a preset warning multiple and a preset absolute rate lower limit threshold. The early warning triggering subunit is used to generate an early warning command for accelerated slope sliding when multiple consecutive sliding time windows meet preset conditions; the preset conditions are: the first average rate is greater than the product of the second average rate and a preset warning multiple, and is greater than the absolute rate lower limit threshold.

8. A high-precision BeiDou monitoring system for slope displacement according to claim 1, characterized in that, It also includes a parameter calibration unit, which specifically includes: The condition detection subunit is used to detect whether the average satellite positioning accuracy factor is less than a preset pure signal threshold and the cumulative rainfall is zero within a continuous preset time period. The interference-free period delineation subunit is used to determine a continuous preset duration as the interference-free observation period when the detection result of the condition detection subunit is yes. The magnification factor calculation subunit is used to determine the surface deformation magnification factor based on the sum of the absolute values ​​of each surface displacement change and the sum of the absolute values ​​of each deep equivalent displacement during the interference-free observation period. The benchmark sequence acquisition subunit is used to control the operation of the matrix construction unit and the matching allocation unit during the interference-free observation period to acquire the benchmark matching error value sequence. The threshold calculation subunit is used to determine the long-term judgment threshold based on the product of the preset quantile value and the preset tolerance coefficient in the benchmark matching error value sequence.

9. A high-precision BeiDou monitoring system for slope displacement according to claim 1, characterized in that, It also includes a data acquisition unit, which specifically includes: The BeiDou data acquisition subunit is used to acquire the three-dimensional coordinate data and satellite positioning accuracy factor sequence output by the BeiDou receiver. The tilt data acquisition subunit is used to acquire the tilt change data output by the deep inclinometer. The rainfall data acquisition subunit is used to acquire rainfall pulse data; The data output subunit is used to output the three-dimensional coordinate data, satellite positioning accuracy factor sequence, tilt change data, and rainfall pulse data to the data acquisition unit and the matrix construction unit.

Images