A virtual reference line-based method for monitoring lateral displacement of roadbed

Abstract

The application discloses a kind of virtual reference line-based roadbed transverse displacement monitoring method, it is related to monitoring technical field, the stability check of original reference point set is carried out after identifying disturbance moment, and abnormal reference point in response in disturbance is identified in time, and based on the point that non-reference point responds smoothly, it meets the point of preset condition and constructs alternative reference line, and then the replaceable value is calculated by the comparison of original virtual reference line and alternative virtual reference line, quantitatively judge whether it needs to replace reference line, to realize the effective response to the problem of virtual reference point " instantaneous jump " under sudden disturbance, avoid the reference system drift caused by reference point itself participates displacement, prevent the misjudgment or the missed judgment of transverse displacement caused by reference line error, ensure the accuracy of transverse displacement monitoring result and the reliability of system response under critical disturbance condition, effectively improve the stability and practicality of virtual reference line monitoring method in complex dynamic scene.

Description

Technical Field

[0001] This invention relates to the field of monitoring technology, and specifically to a method for monitoring the lateral displacement of roadbed based on a virtual baseline. Background Technology

[0002] In modern infrastructure construction, real-time monitoring of lateral displacement of roadbeds is crucial for preventing disasters such as slope slippage and roadbed instability. Traditional monitoring methods typically rely on external physical benchmarks, such as fixed anchor piles or control points, which are susceptible to construction disturbances, geological deformation, or human-caused damage, leading to instability of the reference system. To address this issue, an increasing number of projects are adopting monitoring methods based on virtual baselines. Within the sensor deployment area, a set of points with the smallest deformation and weakest fluctuations within a historical window are selected as "stable points," and a mathematically meaningful virtual reference line is fitted to these points for calculating the relative displacement of the remaining monitoring points. This method is flexible in deployment, highly automated, and widely used in scenarios such as soft soil, embankments, and slopes where physical benchmarks are unsuitable. It has become an important branch of next-generation intelligent monitoring technology.

[0003] However, the core assumption of the existing methods is that the stable points remain truly stationary during the monitoring period, and their stability can be correctly judged through historical fluctuation characteristics. In reality, during critical disturbance events such as earthquakes, blasting, heavy rainfall, and gradual slope slippage, the stable points that were originally used as the benchmark may experience sudden lateral displacement within a very short time, yet the system still treats them as "unmoved" and participates in the baseline fitting. Such "instantaneous jumps" in stable points will cause the entire virtual baseline to drift, resulting in a systematic deviation in the displacement results of all monitoring points. Because the reference frame itself is subtly skewed, this problem is highly likely to cause abnormal false alarms or misjudgments as an "overall slippage trend," while the risk points where actual displacement has occurred are masked. Currently, mainstream monitoring systems generally lack mechanisms for identifying and handling this phenomenon, making it difficult to guarantee the authenticity of monitoring results and the reliability of safety responses under critical events. Summary of the Invention

[0004] The purpose of this invention is to solve the problems mentioned above and provide a method for monitoring the lateral displacement of roadbed based on a virtual baseline.

[0005] This invention proposes a method for monitoring the lateral displacement of roadbed based on a virtual baseline, the method comprising:

[0006] The system acquires real-time displacement data from multiple monitoring points deployed on the cross section of the roadbed, and identifies the timing of roadbed disturbance based on the real-time displacement data of the monitoring points at adjacent times.

[0007] At the time of disturbance, the stability of the set of reference points used to construct the virtual baseline is checked to identify the reference points with abnormal responses.

[0008] Monitoring points that meet the preset verification conditions are selected from non-benchmark points as candidate benchmark points. An alternative benchmark line for the virtual benchmark line is constructed based on the candidate benchmark points to obtain the alternative virtual benchmark line.

[0009] The replaceable value of the alternative virtual baseline is calculated based on the original virtual baseline and the alternative virtual baseline. The final alternative virtual baseline is determined based on the replaceable value, and the lateral displacement of the roadbed is continuously monitored based on the final alternative virtual baseline.

[0010] Optionally, the steps for identifying the timing of roadbed disturbance based on real-time displacement data of monitoring points at adjacent times are as follows:

[0011] Obtain the lateral displacement value of each monitoring point at the current time and the adjacent previous time, and calculate the displacement change rate of each monitoring point between adjacent time.

[0012] The displacement change rate of each monitoring point is compared with the preset change rate threshold, and the number of monitoring points that meet the condition that the displacement change rate is greater than the disturbance threshold is counted.

[0013] The proportion of monitoring points whose displacement change rate is greater than the disturbance threshold is calculated out of the total number of monitoring points, and this proportion is used as the disturbance ratio value at the current moment.

[0014] The disturbance ratio is compared with a preset ratio threshold. When the disturbance ratio is not less than the ratio threshold, the current time is determined to be the disturbance time.

[0015] Optionally, the step of performing a stability check on the set of reference points used to construct the virtual baseline and identifying reference points with abnormal responses is as follows:

[0016] The lateral displacement increment of each reference point in the reference point set between the time before the disturbance and the time of the disturbance trigger is obtained. The lateral displacement increment is the displacement value of the reference point at the time of the disturbance trigger minus its displacement value at the time before the disturbance.

[0017] The displacement increments of any two reference points are compared in direction. Based on the calculation method of the angle between vectors, the displacement increment direction angle between each reference point and all other reference points is calculated, and the average value of the displacement increment direction angle between each reference point and all other reference points is taken as the perturbation direction difference of each reference point.

[0018] The perturbation direction difference of each reference point is compared with the first preset angle threshold. When the perturbation direction difference of a reference point is not less than the first preset angle threshold, the corresponding reference point is determined to be a response abnormal reference point. If the perturbation direction difference is less than the first preset angle threshold, the corresponding reference point is determined to be a normal abnormal reference point.

[0019] Optionally, the steps for selecting monitoring points that meet preset verification conditions from non-reference points as candidate reference points, and constructing alternative reference lines for the virtual reference line based on the candidate reference points to obtain alternative virtual reference lines are as follows:

[0020] Obtain the lateral displacement increment of each non-reference point between the disturbance trigger time and the reference time before the disturbance. The lateral displacement increment is the displacement value of the non-reference point at the disturbance trigger time minus its displacement value at the reference time before the disturbance.

[0021] Calculate the magnitude of the displacement increment vector of each non-reference point, and use it as the disturbance amplitude of each non-reference point;

[0022] The target disturbance direction vector is determined based on the normal and abnormal reference points, and the angle between the displacement increment direction of each non-reference point and the target disturbance direction is calculated.

[0023] Non-reference points that simultaneously satisfy the condition that the disturbance amplitude is no greater than a preset amplitude threshold and the angle between the disturbance directions is no greater than a second preset angle threshold are selected as the candidate reference point set.

[0024] Based on the spatial distribution information of the candidate benchmark point set, several candidate benchmark points are selected, and a fitting method is used to generate an alternative virtual benchmark line.

[0025] Optionally, the step of calculating the replaceable value of the alternative virtual baseline based on the original virtual baseline and the alternative virtual baseline is as follows:

[0026] The improved directional entropy and drift response hysteresis values ​​are calculated based on the original virtual baseline and the alternative virtual baseline. The alternative virtual baseline is obtained by subtracting the drift response hysteresis value from the improved directional entropy value.

[0027] Optionally, the calculation steps for the improved directional entropy value are as follows:

[0028] The lateral displacement increment of each monitoring point between the disturbance trigger time and the reference time before the disturbance is obtained. The lateral displacement increment is the lateral displacement value of the corresponding monitoring point at the disturbance trigger time minus the lateral displacement value of the monitoring point at the reference time before the disturbance.

[0029] Calculate the mean and standard deviation of all displacement increments, and subtract the mean from each displacement increment and then divide by the standard deviation to obtain the standardized disturbance value;

[0030] All standardized perturbation values ​​are divided into multiple statistical segments according to fixed intervals. The proportion of standardized perturbation values ​​in each statistical segment to the total number is counted to obtain the probability distribution of the perturbation distribution.

[0031] Based on the probability distribution, calculate the perturbation distribution information entropy corresponding to the original virtual baseline and the alternative virtual baseline, respectively.

[0032] The improved directional entropy value is obtained by subtracting the information entropy under the replacement virtual reference line from the information entropy under the original virtual reference line and dividing the difference by the information entropy under the original virtual reference line.

[0033] Optionally, the calculation steps for the drift response hysteresis value are as follows:

[0034] The lateral displacement values ​​of each monitoring point at the time of disturbance triggering and the adjacent time points before and after it are obtained, and a disturbance response sequence containing three consecutive time points is constructed.

[0035] For the disturbance response sequence of each monitoring point, the displacement difference between adjacent time points is calculated to obtain the first-order difference sequence representing the disturbance trend. The first-order difference sequence is the lateral displacement value at the current time minus the lateral displacement value at the previous time.

[0036] Based on the original virtual baseline and the alternative virtual baseline, the Euclidean distance between the two sets of first-order difference sequences of the same monitoring point is calculated. The Euclidean modulus of the difference sequence under the original virtual baseline is used as the normalized denominator to obtain the disturbance response lag ratio of the monitoring point. The disturbance response lag ratio is the relative rate of change between the original difference sequence and the alternative difference sequence.

[0037] The disturbance response lag ratios of all monitoring points are unified, with the largest ratio as the normalization benchmark, and the average value of the normalized values ​​of all monitoring points is taken as the drift response lag value.

[0038] Optionally, the steps for determining the final alternative virtual baseline based on the replaceable values ​​are as follows:

[0039] Compare the replaceable value with the preset replaceable value threshold. If the replaceable value is not less than the preset replaceable value threshold, it means that the corresponding replacement virtual baseline can be directly used as the final replacement virtual baseline to continuously monitor the lateral displacement of the roadbed.

[0040] If the replaceable value is less than the preset replaceable value threshold, it means that the corresponding alternative virtual baseline cannot be directly used as the final alternative virtual baseline for continuous monitoring of the roadbed lateral displacement; continue to select other candidate baselines from the candidate baseline set to generate a new alternative virtual baseline until the replaceable value of the new alternative virtual baseline is not less than the preset replaceable value threshold, then it means that the corresponding new alternative virtual baseline can be directly used as the final alternative virtual baseline for continuous monitoring of the roadbed lateral displacement.

[0041] The beneficial effects of this invention are:

[0042] This invention proposes a method for monitoring lateral displacement of roadbeds based on virtual baselines. By verifying the stability of the original set of baseline points after identifying the disturbance, it promptly identifies baseline points with abnormal responses during disturbances. A substitute baseline is constructed based on points among the non-baseline points that have stable responses and meet preset conditions. The replaceable value is then calculated by comparing the original virtual baseline with the substitute virtual baseline, quantitatively determining whether the baseline needs to be replaced. This effectively addresses the problem of "instantaneous jumps" in virtual baselines under sudden disturbances, avoiding reference frame drift caused by the baseline points themselves participating in the displacement, and preventing misjudgments or omissions of lateral displacement due to incorrect baselines. It ensures the accuracy of lateral displacement monitoring results and the reliability of the system response under critical disturbance conditions, effectively improving the stability and practicality of the virtual baseline monitoring method in complex dynamic scenarios. Attached Figure Description

[0043] The invention will now be further described with reference to the accompanying drawings.

[0044] Figure 1 This is a flowchart of a method for monitoring the lateral displacement of roadbed based on a virtual baseline. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] This invention provides a method for monitoring lateral displacement of roadbed based on a virtual baseline. See also... Figure 1 , Figure 1 A flowchart illustrating a method for monitoring lateral displacement of roadbed based on a virtual baseline, provided in an embodiment of the present invention. The method includes:

[0047] S1: Acquire real-time displacement data of multiple monitoring points deployed on the cross section of the roadbed, and identify the disturbance time of the roadbed based on the real-time displacement data of the monitoring points at adjacent times.

[0048] S2: At the time of disturbance, perform stability verification on the set of reference points used to construct the virtual baseline and identify the reference points with abnormal responses.

[0049] S3: Select monitoring points that meet the preset verification conditions from the non-benchmark points as candidate benchmark points, and construct alternative benchmarks for the virtual benchmark based on the candidate benchmark points to obtain alternative virtual benchmarks;

[0050] S4: Calculate the replaceable value of the alternative virtual baseline based on the original virtual baseline and the alternative virtual baseline, determine the final alternative virtual baseline based on the replaceable value, and continuously monitor the lateral displacement of the roadbed based on the final alternative virtual baseline.

[0051] This invention provides a method for monitoring lateral displacement of roadbed based on a virtual baseline. By performing stability checks on the original set of reference points after identifying the disturbance, it promptly identifies reference points with abnormal responses during disturbances. A substitute baseline is then constructed based on points among the non-reference points that exhibit stable responses and meet preset conditions. The replaceable value is calculated by comparing the original virtual baseline with the substitute virtual baseline, quantifying whether the baseline needs to be replaced. This effectively addresses the problem of "instantaneous jumps" in virtual reference points under sudden disturbances, avoiding reference frame drift caused by the reference points' own displacement, and preventing misjudgments or omissions of lateral displacement due to incorrect baselines. It ensures the accuracy of lateral displacement monitoring results and the reliability of the system response under critical disturbance conditions, effectively improving the stability and practicality of the virtual baseline monitoring method in complex dynamic scenarios.

[0052] In one embodiment, S1: acquire real-time displacement data of multiple monitoring points deployed on the cross section of the roadbed, and identify the disturbance time of the roadbed based on the real-time displacement data of the monitoring points at adjacent times.

[0053] In one implementation, the steps for identifying the timing of roadbed disturbance based on real-time displacement data of monitoring points at adjacent times are as follows:

[0054] Obtain the lateral displacement value of each monitoring point at the current time and the adjacent previous time, and calculate the displacement change rate of each monitoring point between adjacent time.

[0055] The displacement change rate of each monitoring point is compared with the preset change rate threshold, and the number of monitoring points that meet the condition that the displacement change rate is greater than the disturbance threshold is counted.

[0056] The proportion of monitoring points whose displacement change rate is greater than the disturbance threshold is calculated out of the total number of monitoring points, and this proportion is used as the disturbance ratio value at the current moment.

[0057] The disturbance ratio is compared with a preset ratio threshold. When the disturbance ratio is not less than the ratio threshold, the current time is determined to be the disturbance time.

[0058] It should be noted that the lateral displacement values ​​of each monitoring point at the current time and the adjacent previous time point involved in the above steps are usually obtained by displacement sensors deployed at the cross-section of the roadbed or key structural locations. These sensors may include distributed optical fibers, FBG gratings, MEMS displacement gauges, or GNSS displacement acquisition modules, etc. The system continuously collects the displacement data of each sensor at a fixed sampling period (such as every 10 seconds or every minute) and timestamps it, thereby forming a displacement time-series data sequence with time order. Specifically, each sensor periodically uploads its current lateral displacement data to the monitoring platform. The platform calculates the difference between the displacement data corresponding to two adjacent sampling times (such as t1 and t2) to obtain the displacement change rate of each monitoring point within that time interval. For example, if the lateral displacement of a monitoring point is 35.2 mm at time t1 and 39.8 mm at time t2, then its change rate is 4.6 mm. By performing this kind of processing on the data of all monitoring points, complete disturbance monitoring input data can be obtained without manual intervention.

[0059] It should be noted that the above steps calculate the rate of change of lateral displacement of monitoring points between adjacent time points and determine whether a disturbance occurs by combining the abnormal proportion of all monitoring points. The design aims to achieve rapid, real-time, and objective identification of the overall disturbance state of the roadbed, avoiding misjudgments caused by single-point fluctuations, and enhancing the system's response capability to sudden geological activities, structural deformations, or external interference events (such as blasting, slip-start, heavy-load traffic, etc.). Specifically, this method first obtains the lateral displacement value of each monitoring point between the current time and the previous adjacent time point and calculates the rate of change accordingly, because rate reflects the trend of change in a short period of time better than absolute displacement, making it suitable for determining "whether a disturbance is occurring." Then, the rate of each point is compared with a preset threshold to distinguish between small natural fluctuations and abnormally drastic changes. Finally, by statistically analyzing the proportion of abnormal points to the total number and comparing it with a threshold, "false disturbance judgments" caused by individual sensor anomalies, occasional mechanical vibrations, or local construction interference can be effectively avoided, improving the robustness and reliability of the identification. Reliability; for example, if only 3 out of 100 monitoring points have a higher rate, it may only be a local error, but if 30 points change drastically at the same time, it is more likely to be a structural disturbance; therefore, comparing the ratio with a set threshold is to introduce a spatial "common judgment standard"; once a disturbance is identified, the system can immediately start the subsequent benchmark stability analysis and virtual benchmark replacement process to achieve rapid switching of monitoring logic; in summary, this method not only realizes sudden perception of changes in the time dimension, but also combines global response characteristics in the spatial dimension, making disturbance identification more real-time, systematic, and practical in engineering.

[0060] In one embodiment, S2: At the time of disturbance, the stability of the set of reference points used to construct the virtual baseline is checked to determine the reference points that respond abnormally.

[0061] In one implementation, the step of performing a stability check on the set of reference points used to construct the virtual baseline and identifying reference points with abnormal responses is as follows:

[0062] The lateral displacement increment of each reference point in the reference point set between the time before the disturbance and the time of the disturbance trigger is obtained. The lateral displacement increment is the displacement value of the reference point at the time of the disturbance trigger minus its displacement value at the time before the disturbance.

[0063] The displacement increments of any two reference points are compared in direction. Based on the calculation method of the angle between vectors, the angle between the displacement increment direction of each reference point and all other reference points is calculated, and the average of the angles between the displacement increment direction of each reference point and all other reference points is taken as the difference in the disturbance direction of each reference point. The calculation method of the displacement increment direction angle is as follows: In the formula, and The first The and the first Displacement increment of each reference point For the first The and the first The angle between the displacement increment directions of the reference points;

[0064] The perturbation direction difference of each reference point is compared with the first preset angle threshold. When the perturbation direction difference of a reference point is not less than the first preset angle threshold, the corresponding reference point is determined to be a response abnormal reference point. If the perturbation direction difference is less than the first preset angle threshold, the corresponding reference point is determined to be a normal abnormal reference point.

[0065] It should be noted that the lateral displacement increments of each reference point in the reference point set involved in the above steps between the time before the disturbance and the time of the disturbance trigger are usually obtained through high-precision displacement monitoring equipment deployed on the cross-section of the roadbed structure. Such equipment may include fiber optic grating sensors (FBG), distributed optical fiber sensors (DAS / DOFS), GNSS displacement monitoring stations, MEMS inertial measurement units, etc. The system uses a unified time control module to periodically and synchronously sample each sensor to obtain the lateral displacement value of each monitoring point at a set time interval. The specific sampling period can be set to several seconds to several minutes. In practical applications, the system first determines the disturbance trigger time and its corresponding previous reference time in the disturbance identification module, and then... Then, the system retrieves the original displacement data corresponding to the two time points from the database and performs a difference operation on each reference point. This involves subtracting the lateral displacement data at the reference time before the disturbance from the lateral displacement data at the time the disturbance was triggered, thus obtaining the displacement change of that point during the disturbance process. For example, if the displacement of a reference point before the disturbance is 12.4 mm and at the time the disturbance occurs it is 17.6 mm, then its lateral displacement increment is 5.2 mm. Based on this, the system further performs directional consistency analysis and anomaly identification. Therefore, the process of acquiring displacement increment data relies on continuous sampling by the sensor network, time-series archiving by the data management system, and unified execution of the difference algorithm. The entire process is highly automated and requires no manual intervention, ensuring the real-time performance and stability of data processing.

[0066] It should be noted that the above steps calculate the directional angle of the displacement increment of each reference point between the time before the disturbance and the time of the disturbance trigger, and use the average angle between each reference point and all other reference points as the disturbance direction difference, thereby identifying the reference points with abnormal responses. The purpose of this design is to promptly identify those reference points that are no longer stable but are still mistakenly regarded as "reference points" by the system after the occurrence of a critical disturbance event, so as to prevent the abnormal response behavior of these points from causing misleading deviations of the virtual baseline. Unlike traditional methods that rely solely on displacement magnitude to identify anomalies, this method introduces the concept of "disturbance direction consistency." If the disturbance direction of a reference point deviates significantly from that of most reference points (i.e., the average angle between directions is large), it is highly likely that the point experienced a mechanical response different from the main trend during the disturbance event, exhibiting "structural instability." If such anomalies are included in the virtual baseline construction, it will cause the entire reference system to tilt, resulting in a systematic bias in the lateral displacement calculations of all monitoring points. This method effectively solves the problem of implicit reference point instability—where the magnitude is small but the direction deviates—by quantifying direction consistency, thus improving the system's ability to identify minute but structural changes under critical disturbances. For example, if a reference point experiences a displacement in the opposite direction to the overall system during a disturbance, even if the displacement is only a few millimeters, it may cause a change in the tilt of the reference line. This method can identify and eliminate such displacements through angle calculation, ensuring that the reference line relied upon for subsequent monitoring remains truly stable. This improves the robustness, reliability, and engineering practicality of the entire virtual baseline monitoring system under complex geological conditions or sudden loads.

[0067] In one embodiment, S3: Select monitoring points that meet the preset verification conditions from the non-reference points as candidate reference points, and construct an alternative reference line for the virtual reference line based on the candidate reference points to obtain the alternative virtual reference line;

[0068] In one implementation, the steps of selecting monitoring points that meet preset verification conditions from non-reference points as candidate reference points, and constructing alternative reference lines for the virtual reference line based on the candidate reference points to obtain the alternative virtual reference line are as follows:

[0069] Obtain the lateral displacement increment of each non-reference point between the disturbance trigger time and the reference time before the disturbance. The lateral displacement increment is the displacement value of the non-reference point at the disturbance trigger time minus its displacement value at the reference time before the disturbance.

[0070] Calculate the magnitude of the displacement increment vector of each non-reference point as the disturbance amplitude of each non-reference point. If the disturbance amplitude is not greater than the preset amplitude threshold, the corresponding non-reference point is considered to have changed steadily during the disturbance.

[0071] The target disturbance direction vector is determined based on the normal and abnormal reference points, and the angle between the displacement increment direction of each non-reference point and the target disturbance direction is calculated. If the angle is not greater than the second preset angle threshold, the disturbance direction of the non-reference point is considered to be consistent with the main trend.

[0072] Non-reference points that simultaneously satisfy the condition that the disturbance amplitude is no greater than a preset amplitude threshold and the angle between the disturbance directions is no greater than a second preset angle threshold are selected as the candidate reference point set.

[0073] Based on the spatial distribution information of the candidate reference point set, several candidate reference points are selected in ascending order of disturbance amplitude and disturbance direction angle. An alternative virtual reference line is generated by fitting, and the alternative virtual reference line is used to replace the original virtual reference line for subsequent lateral displacement calculation.

[0074] It should be noted that the non-reference points in the above steps refer to the set of monitoring points that were not included in the construction of the virtual baseline before the current disturbance event occurred. That is, the remaining monitoring points after excluding the original reference point set. These points are not used as reference benchmarks but still have real-time displacement monitoring functions and are an important source of candidate benchmarks. The target disturbance direction vector refers to the directional reference quantity that represents the main trend of the overall structural response during the disturbance event. It is usually used to measure whether the displacement direction of each monitoring point is consistent with that of most points during the disturbance. Its determination method is mainly based on the "stable benchmark points" retained after stability verification in the original benchmark point set or the disturbance vector of multiple points verified to have consistent responses. The dynamic displacement increment vector is a unified reference direction obtained by performing directional averaging or principal direction extraction algorithms (such as principal component analysis (PCA) or vector weighted averaging) on ​​these vectors. In practice, the disturbance increment vectors of multiple stable points can be normalized and their directional vectors averaged to obtain a target disturbance direction vector representing the main trend of the disturbance. This vector represents the main motion direction of the entire structure during the disturbance and is used to subsequently determine whether the response direction of non-reference points deviates from the main trend, thereby selecting points with consistent directions to form a candidate reference point set. This method can effectively eliminate outlier points and improve the stability and representativeness of the alternative reference line.

[0075] It should be noted that the above steps involve dual screening of the magnitude and direction of the lateral displacement increments of non-reference points between the time of disturbance and the time of disturbance triggering. The aim is to quickly and accurately identify a set of candidate reference points that still maintain a stable response from the remaining monitoring points when the original reference point set exhibits abnormal response and the virtual reference line becomes unstable. Based on these points, a virtual reference line is refitted and constructed for replacement, thereby restoring the effectiveness and accuracy of the reference system. Specifically, firstly, the displacement increment magnitude of non-reference points is calculated to determine whether the disturbance amplitude is within an acceptable range, and points that fluctuate violently during the disturbance are eliminated. Secondly, a main disturbance direction vector based on a "normal reference point" is introduced as a reference direction, and the angle between the displacement increment direction of each non-reference point and this main trend is calculated, eliminating points with excessively large directional deviations, thereby avoiding structural distortion caused by "abnormal response direction". Finally, non-reference points that simultaneously satisfy the conditions of stable amplitude and consistent direction are selected as the new reference point candidate set. Simultaneously, from the new reference point candidate set, several candidate reference points are selected in ascending order of disturbance amplitude and disturbance direction angle. Specifically, the disturbance amplitude and disturbance direction angle of each reference point in the new reference point candidate set are normalized using MAX-MIN normalization (other methods are also acceptable). The normalized disturbance amplitude and disturbance direction angle of each reference point are summed, and the sum is used as the selection value for each reference point. Several reference points (e.g., the first M) are selected in ascending order of these selection values ​​as the final reference point set for fitting. Based on their spatial distribution, the final reference point set is linearly or curvilinearly fitted to construct a new virtual baseline. The main advantage of this approach is that when the original reference points become unstable due to disturbances, the system can self-repair the reference frame without relying on a physical reference body, maintaining the continuity and accuracy of the monitoring system under extreme events. Simultaneously, through dual constraints on amplitude and direction, the newly generated baseline ensures stronger disturbance resistance and representativeness, significantly reducing overall displacement errors caused by misjudgment of the reference. For example, in a slope sliding or blasting disturbance, most of the original reference points may experience slight movement. If the baseline is not updated, the displacement calculation of all measuring points will be off. However, by dynamically replacing them with new points with consistent direction and smaller fluctuations to reconstruct the reference line using this method, the continuity and reliability of the monitoring curve can be effectively maintained, improving the system's stability and engineering adaptability under sudden conditions.

[0076] It should be noted that, based on the spatial distribution information of the candidate reference point set, the fitting method can take various forms depending on the actual application requirements and data characteristics, including but not limited to linear fitting, curve fitting, principal direction fitting, or spline curve interpolation. Among them, linear fitting is suitable for scenarios where candidate reference points are arranged along straight lines or approximately linearly, and can quickly fit a reference line with a clear overall trend. Curve fitting is suitable for situations where reference points are arranged along curved structures, such as curved roadbeds or slope edges, and can more accurately fit the spatial distribution pattern. Principal direction fitting (e.g., the first principal component direction based on PCA) can extract the overall extension trend direction when there is noise or irregular point layout, and has strong anti-interference ability. Spline interpolation, such as B-splines or cubic splines, is suitable for situations where the point distribution is dense and the fitting requires smooth and continuous lines, and can generate flexible reference lines with good geometric continuity, thereby providing a stable reference for subsequent high-precision calculation of lateral displacement.

[0077] In one embodiment, S4: Calculate the replaceable value of the alternative virtual baseline based on the original virtual baseline and the alternative virtual baseline, determine the final alternative virtual baseline based on the replaceable value, and continuously monitor the lateral displacement of the roadbed based on the final alternative virtual baseline.

[0078] In one implementation, the step of calculating the replaceable value of the alternative virtual baseline based on the original virtual baseline and the alternative virtual baseline is as follows:

[0079] The improved directional entropy and drift response hysteresis values ​​are calculated based on the original virtual baseline and the alternative virtual baseline. The alternative virtual baseline is obtained by subtracting the drift response hysteresis value from the improved directional entropy value.

[0080] In one implementation, the steps for calculating the improved directional entropy value are as follows:

[0081] The lateral displacement increment of each monitoring point between the disturbance trigger time and the reference time before the disturbance is obtained. The lateral displacement increment is the lateral displacement value of the corresponding monitoring point at the disturbance trigger time minus the lateral displacement value of the monitoring point at the reference time before the disturbance.

[0082] Calculate the mean and standard deviation of all displacement increments, and subtract the mean from each displacement increment and then divide by the standard deviation to obtain the standardized disturbance value;

[0083] All standardized perturbation values ​​are divided into multiple statistical segments according to fixed intervals. The proportion of standardized perturbation values ​​in each statistical segment to the total number is counted to obtain the probability distribution of the perturbation distribution.

[0084] Based on the probability distribution, calculate the perturbation distribution information entropy corresponding to the original virtual baseline and the alternative virtual baseline, respectively.

[0085] The information entropy under the original virtual baseline is subtracted from the information entropy under the replacement virtual baseline, and the difference is divided by the information entropy under the original virtual baseline to obtain the improved directional entropy value. The improved directional entropy value is a unitless value between 0 and 1, which represents the degree of improvement in the directional concentration of the disturbance distribution of the replacement virtual baseline relative to the original virtual baseline.

[0086] It should be noted that all data involved in the above calculation of the improved directional entropy value comes from lateral displacement data collected by high-precision monitoring points deployed on the roadbed cross section. These monitoring points can be a monitoring network composed of real-time monitoring devices such as GNSS, fiber optic gratings (FBG), MEMS, laser rangefinders, and total stations. The system synchronously collects the lateral displacement values ​​of each monitoring point by setting a fixed sampling period (such as 1 minute or 10 minutes) and automatically records them to the data management platform in the form of a time series. After identifying the disturbance trigger time and its previous reference time, the system can directly call up the original lateral displacement data of the corresponding two times. The system obtains the lateral displacement increment of each monitoring point through difference calculation. Then, it performs statistical analysis on the collected displacement increment data, calculating the mean, standard deviation, and standardized value. The standardized disturbance value is also automatically calculated by the system without manual intervention. In addition, to construct the probability distribution, the system automatically divides the standardized value into each statistical segment under the preset interval division standard and calculates the proportion of each segment, which is finally used to calculate the disturbance distribution entropy. Therefore, the entire data acquisition and processing process relies on a complete automated acquisition and analysis system, which has the characteristics of strong real-time performance, high data accuracy, and efficient processing, ensuring the objectivity and accuracy of the directional entropy improvement value calculation.

[0087] It should be noted that the improved directional entropy value is an indicator used to measure whether the lateral displacement response of monitoring points is more concentrated and orderly during a disturbance. Essentially, it reflects the degree to which the alternative virtual baseline optimizes the overall directional stability during the disturbance response by calculating the difference in "information entropy" in the statistical distribution of the lateral displacement increments derived from the original virtual baseline and the alternative virtual baseline. Information entropy, as a measure of the degree of disorder or uncertainty in a system, indicates that the larger the value, the more dispersed and chaotic the response of the monitoring points, lacking a clear cooperative trend; the smaller the value, the more concentrated the disturbance response, and the more stable and consistent the overall behavior. Therefore, the improved directional entropy value calculates the relative reduction in disturbance entropy under the original baseline compared to the disturbance entropy under the alternative baseline. A larger value indicates that after using the alternative baseline, the displacement response of all monitoring points during the disturbance exhibits a stronger concentration trend, and the disturbance behavior is more directionally consistent, thereby enhancing the stability and representativeness of the entire reference system. In this case, it indicates that the alternative baseline more accurately reflects the overall deformation law of the structural system under disturbance conditions and is more suitable as a reference line for long-term use. For example, if the disturbance increment distribution of the monitoring points shows multiple peaks under the original baseline, it indicates that the response directions of each point are inconsistent, which may lead to local misleading. However, under the alternative baseline, all disturbance values ​​tend to be concentrated, with only one main peak and high concentration. Therefore, the alternative baseline can be considered to have a stronger "stable reference capability". Thus, the larger the directional entropy improvement value, the higher the credibility and necessity of the replacement. It can be directly used for subsequent continuous monitoring of lateral displacement to avoid the accumulation and spread of systematic errors.

[0088] The greatest advantage of calculating the improved directional entropy value using the above method is that it does not rely on the directional angle information of the displacement vector, but directly performs statistical analysis based on the one-dimensional distribution characteristics of the lateral displacement increment. This makes it more universal and feasible, especially suitable for scenarios where the data acquisition dimension is limited or where only one-dimensional lateral displacement is obtained in actual monitoring. This method eliminates the influence of displacement scale under different working conditions through standardization, making the data under different disturbance stages comparable. By constructing the distribution probability of the standardized disturbance and calculating its information entropy change, it can intuitively quantify the degree of improvement of the monitoring point response concentration by the replacement baseline. Compared with the traditional method based on mean, variance, or deviation, this method can comprehensively consider the global distribution change of the disturbance response, is not affected by a single outlier, has stronger robustness, and the final improved value is a normalized unitless index, which makes it easy to uniformly set the evaluation threshold in various working conditions and multiple replacement judgments, improving the stability of the algorithm and the consistency of the judgment.

[0089] In one implementation, the calculation steps for the drift response hysteresis value are as follows:

[0090] The lateral displacement values ​​of each monitoring point at the time of disturbance triggering and the adjacent time points before and after it are obtained, and a disturbance response sequence containing three consecutive time points is constructed.

[0091] For the disturbance response sequence of each monitoring point, the displacement difference between adjacent time points is calculated to obtain the first-order difference sequence representing the disturbance trend. The first-order difference sequence is the lateral displacement value at the current time minus the lateral displacement value at the previous time.

[0092] Based on the original virtual baseline and the alternative virtual baseline, the Euclidean distance is calculated for two sets of first-order difference sequences at the same monitoring point. The Euclidean modulus of the difference sequence under the original virtual baseline is used as the normalized denominator to obtain the disturbance response lag ratio of the monitoring point. The disturbance response lag ratio is the relative rate of change between the original difference sequence and the alternative difference sequence.

[0093] The disturbance response lag ratios of all monitoring points are unified, and the maximum ratio is used as the normalization benchmark, normalized to the range between 0 and 1. The average value of the normalized values ​​of all monitoring points is taken as the drift response lag value. The smaller the drift response lag value, the higher the degree to which the alternative virtual baseline maintains the response trend of the monitoring points during the disturbance, and the more suitable it is as the final virtual baseline for continuous monitoring of the lateral displacement of the roadbed.

[0094] It should be noted that all the data involved in the above-mentioned drift response hysteresis calculation process comes from the raw lateral displacement data collected by high-precision displacement monitoring points deployed on the roadbed cross section or key parts. These monitoring points may include monitoring systems composed of equipment such as GNSS, total stations, fiber optic gratings (FBG), and MEMS sensors. They typically collect lateral displacement data from each monitoring point continuously at a preset fixed sampling period (e.g., every 1 minute, every 5 minutes, or other suitable intervals), and upload the data to the central monitoring platform to form a structured time-series database according to the timestamp. After identifying the disturbance trigger time, the system can automatically call the lateral displacement data of that time and its adjacent time intervals (e.g., t'-1, t', t'+1) to construct the disturbance response sequence for each monitoring point. Subsequently, based on the set original virtual baseline and the alternative virtual baseline, the system uses the lateral displacement data under the corresponding reference system as input data, and obtains the first-order difference sequence through time-series difference calculation. This is further used for the Euclidean distance and modulus calculation between the two sets of difference sequences. All data acquisition and processing are completed automatically by the system.

[0095] It should be noted that the drift response hysteresis value is an indicator used to measure the degree to which the alternative virtual baseline maintains the trend of the lateral displacement response of the monitoring point relative to the original virtual baseline during a disturbance event. Essentially, it reflects the degree of difference in the "rhythm of change" or "synchronicity of response" between the disturbance behavior of the monitoring point under the alternative baseline and the behavior under the original baseline. This indicator is obtained by performing first-order difference analysis on the continuous lateral displacement data of the monitoring point during the disturbance period, calculating the displacement change rate under both the original and alternative baselines, and then comparing the normalized distances of the two to derive the relative rate of change of the disturbance trend. The smaller the drift response lag value, the more synchronized or highly consistent the disturbance response speed and trend of each monitoring point under the replacement baseline are with those under the original baseline. This means the replacement has not introduced new "response delays" or "trend shifts," and it can be considered to have sufficient inheritance and reference stability in its time-series response. Therefore, it can be directly used as a new virtual baseline for subsequent continuous monitoring. Conversely, if the drift response lag value is large, it means the replacement baseline has introduced significant response differences, possibly due to lag, advancement, or directional changes. In this case, the replacement reference system lacks a reliable basis for continuous continuation, and hasty replacement may introduce systematic misjudgments. For example, if, during a blasting disturbance, the response of each monitoring point under the original baseline rises rapidly within one minute, while the response under the replacement baseline shows a "delayed increase" or "amplitude fluctuation," the lag value will increase significantly, reflecting that the replacement system is insensitive to or asynchronous in its disturbance response, making it unsuitable as a reference baseline for subsequent displacement calculations. Therefore, the smaller the drift response hysteresis value, the better the inheritance of the alternative virtual baseline to the structural disturbance response, and the more suitable it is as the final baseline for continuous monitoring.

[0096] The advantage of calculating the drift response hysteresis value using the above method is that it uses the first-order difference of the disturbance response sequence as the measurement basis, which can directly reflect the rate and rhythm difference of change of the lateral displacement trend of the monitoring point during the disturbance. It does not depend on the absolute displacement magnitude, nor is it affected by the offset of the reference origin of the baseline itself. Therefore, it has stronger objectivity and robustness in assessing the ability of the substitute virtual baseline to maintain the temporal sequence of the disturbance response. Compared with the traditional method of judging the stability of the substitute baseline only by the difference in displacement increment or static deviation value, this method pays more attention to the dynamic change trend, can capture small hysteresis effects or trend deviations, and avoids the influence of disturbance. In the initial response phase, errors in judgment may occur due to delays or reversals in the response of the replacement baseline. Furthermore, this method does not involve complex vector analysis or direction determination; it only relies on one-dimensional lateral displacement data for calculation. It is suitable for engineering environments in most practical scenarios where monitoring points are deployed in a single direction and data dimensions are limited. It has advantages such as simple implementation, low data requirements, and strong portability. The final calculation results are normalized and averaged using Euclidean distance to generate a stable, unitless index in the 0-1 interval. This facilitates a unified assessment of the feasibility of the replacement baseline under different monitoring batches or operating conditions, improving the accuracy and practicality of the monitoring system's judgment on reference frame updates after sudden disturbances.

[0097] In one embodiment, the step of determining the final alternative virtual baseline based on the replaceable value is as follows:

[0098] Compare the replaceable value with the preset replaceable value threshold. If the replaceable value is not less than the preset replaceable value threshold, it means that the corresponding replacement virtual baseline can be directly used as the final replacement virtual baseline to continuously monitor the lateral displacement of the roadbed.

[0099] If the replaceable value is less than the preset replaceable value threshold, it means that the corresponding alternative virtual baseline cannot be directly used as the final alternative virtual baseline for continuous monitoring of the roadbed lateral displacement; continue to select other candidate baselines from the candidate baseline set to generate a new alternative virtual baseline until the replaceable value of the new alternative virtual baseline is not less than the preset replaceable value threshold, then it means that the corresponding new alternative virtual baseline can be directly used as the final alternative virtual baseline for continuous monitoring of the roadbed lateral displacement.

[0100] It should be noted that by comparing the replaceable value of the current alternative virtual baseline with the system's preset replaceable value threshold, the system determines whether the alternative baseline possesses sufficient stability and representativeness. If the replaceable value is not less than the threshold, it indicates that it can effectively maintain the continuity of the overall monitoring network's reference system after disturbances and can be directly used as the final virtual baseline. If it is lower than the threshold, it indicates that the baseline has a significant deviation in the consistency of the disturbance direction or the response timing, and it is not suitable for direct use. To avoid misjudgment or misleading the monitoring curves, it is necessary to change the combination method from the candidate baseline set, select a new set of candidate points, generate a new alternative baseline, and recalculate its replaceable value until a new set of reference baselines with sufficient stability is selected. This method not only enhances the controllability of the generation of alternative baselines but also realizes iterative updates based on indicator feedback, avoiding distortion of the alternative reference system due to subjective human judgment. For example, if the initially generated alternative baseline consists of 3 candidate reference points, but its replaceable value is 0.21, which is lower than the threshold of 0.25, the system can try to add a fourth stable candidate point to regenerate a new baseline and calculate its new replaceable value. If it reaches 0.28, it can be considered to meet the replacement condition, thereby completing the switch of the reference system and ensuring the continuity and accuracy of subsequent roadbed lateral displacement monitoring.

[0101] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the scope of the claims.

Claims

1. A method for monitoring lateral displacement of roadbed based on a virtual baseline, characterized in that, Includes the following steps: The system acquires real-time displacement data from multiple monitoring points deployed on the cross section of the roadbed, and identifies the timing of roadbed disturbance based on the real-time displacement data of the monitoring points at adjacent times. At the time of disturbance, the stability of the set of reference points used to construct the virtual baseline is checked to identify the reference points with abnormal responses. Monitoring points that meet the preset verification conditions are selected from non-benchmark points as candidate benchmark points. An alternative benchmark line for the virtual benchmark line is constructed based on the candidate benchmark points to obtain the alternative virtual benchmark line. The replaceable value of the alternative virtual baseline is calculated based on the original virtual baseline and the alternative virtual baseline. The final alternative virtual baseline is determined based on the replaceable value. The lateral displacement of the roadbed is continuously monitored based on the final alternative virtual baseline. The steps for calculating the replaceable value of the alternative virtual baseline based on the original virtual baseline and the alternative virtual baseline are as follows: The improved directional entropy and drift response hysteresis values ​​are calculated based on the original virtual baseline and the alternative virtual baseline. The alternative virtual baseline is obtained by subtracting the drift response hysteresis value from the improved directional entropy value. The steps for calculating the improved directional entropy value are as follows: The lateral displacement increment of each monitoring point between the disturbance trigger time and the reference time before the disturbance is obtained. The lateral displacement increment is the lateral displacement value of the corresponding monitoring point at the disturbance trigger time minus the lateral displacement value of the monitoring point at the reference time before the disturbance. Calculate the mean and standard deviation of all displacement increments, and subtract the mean from each displacement increment and then divide by the standard deviation to obtain the standardized disturbance value; All standardized perturbation values ​​are divided into multiple statistical segments according to fixed intervals. The proportion of standardized perturbation values ​​in each statistical segment to the total number is counted to obtain the probability distribution of the perturbation distribution. Based on the probability distribution, calculate the perturbation distribution information entropy corresponding to the original virtual baseline and the alternative virtual baseline, respectively. Subtract the information entropy under the original virtual reference line from the information entropy under the replacement virtual reference line to obtain the difference. Divide the difference by the information entropy under the original virtual reference line to obtain the improved directional entropy value. The calculation steps for the drift response hysteresis value are as follows: The lateral displacement values ​​of each monitoring point at the time of disturbance triggering and the adjacent time points before and after it are obtained, and a disturbance response sequence containing three consecutive time points is constructed. For the disturbance response sequence of each monitoring point, the displacement difference between adjacent time points is calculated to obtain the first-order difference sequence representing the disturbance trend. The first-order difference sequence is the lateral displacement value at the current time minus the lateral displacement value at the previous time. Based on the original virtual baseline and the alternative virtual baseline, the Euclidean distance between the two sets of first-order difference sequences of the same monitoring point is calculated. The Euclidean modulus of the difference sequence under the original virtual baseline is used as the normalized denominator to obtain the disturbance response lag ratio of the monitoring point. The disturbance response lag ratio is the relative rate of change between the original difference sequence and the alternative difference sequence. The disturbance response lag ratios of all monitoring points are unified, with the largest ratio as the normalization benchmark, and the average value of the normalized values ​​of all monitoring points is taken as the drift response lag value.

2. The method for monitoring lateral displacement of roadbed based on a virtual baseline according to claim 1, characterized in that, The steps for identifying the timing of roadbed disturbance based on real-time displacement data of monitoring points at adjacent times are as follows: Obtain the lateral displacement value of each monitoring point at the current time and the adjacent previous time, and calculate the displacement change rate of each monitoring point between adjacent time. The displacement change rate of each monitoring point is compared with the preset change rate threshold, and the number of monitoring points that meet the condition that the displacement change rate is greater than the preset change rate threshold is counted. The proportion of monitoring points whose displacement change rate is greater than the preset change rate threshold is calculated as the disturbance ratio value at the current moment. The disturbance ratio is compared with a preset ratio threshold. When the disturbance ratio is not less than the ratio threshold, the current time is determined to be the disturbance time.

3. The method for monitoring lateral displacement of roadbed based on a virtual baseline according to claim 1, characterized in that, The steps for performing stability checks on the set of reference points used to construct the virtual baseline and identifying reference points with abnormal responses are as follows: The lateral displacement increment of each reference point in the reference point set between the time before the disturbance and the time of the disturbance trigger is obtained. The lateral displacement increment is the displacement value of the reference point at the time of the disturbance trigger minus its displacement value at the time before the disturbance. The displacement increments of any two reference points are compared in direction. Based on the calculation method of the angle between vectors, the displacement increment direction angle between each reference point and all other reference points is calculated, and the average value of the displacement increment direction angle between each reference point and all other reference points is taken as the perturbation direction difference of each reference point. The perturbation direction difference of each reference point is compared with the first preset angle threshold. When the perturbation direction difference of a reference point is not less than the first preset angle threshold, the corresponding reference point is determined to be an abnormal response reference point. If the perturbation direction difference is less than the first preset angle threshold, the corresponding reference point is determined to be a normal reference point.

4. The method for monitoring lateral displacement of roadbed based on a virtual baseline according to claim 3, characterized in that, The steps for selecting monitoring points that meet preset verification conditions from non-reference points as candidate reference points, and constructing alternative reference lines for the virtual reference line based on the candidate reference points, are as follows: Obtain the lateral displacement increment of each non-reference point between the disturbance trigger time and the reference time before the disturbance. The lateral displacement increment is the displacement value of the non-reference point at the disturbance trigger time minus its displacement value at the reference time before the disturbance. Calculate the magnitude of the displacement increment vector of each non-reference point, and use it as the disturbance amplitude of each non-reference point; The target disturbance direction vector is determined based on the normal reference point, and the angle between the displacement increment direction of each non-reference point and the target disturbance direction is calculated. Non-reference points that simultaneously satisfy the condition that the disturbance amplitude is no greater than a preset amplitude threshold and the angle between the disturbance directions is no greater than a second preset angle threshold are selected as the candidate reference point set. Based on the spatial distribution information of the candidate benchmark point set, several candidate benchmark points are selected, and a fitting method is used to generate an alternative virtual benchmark line.

5. The method for monitoring lateral displacement of roadbed based on a virtual baseline according to claim 1, characterized in that, The steps for determining the final alternative virtual baseline based on the replaceable values ​​are as follows: Compare the replaceable value with the preset replaceable value threshold. If the replaceable value is not less than the preset replaceable value threshold, it means that the corresponding replacement virtual baseline can be directly used as the final replacement virtual baseline to continuously monitor the lateral displacement of the roadbed. If the replaceable value is less than the preset replaceable value threshold, it means that the corresponding replacement virtual baseline cannot be directly used as the final replacement virtual baseline for continuous monitoring of the roadbed lateral displacement. Continue to select other candidate reference points from the candidate reference point set to generate new alternative virtual reference lines until the replaceable value of the new alternative virtual reference line is not less than the preset replaceable value threshold. Then, the corresponding new alternative virtual reference line can be directly used as the final alternative virtual reference line to continuously monitor the lateral displacement of the roadbed.

Images