Roadbed settlement safety monitoring method and system

Abstract

The invention is suitable for the technical field of road construction, and provides a roadbed settlement safety monitoring method and system, and the method comprises the steps: obtaining measurement equipment, and obtaining equipment information according to the equipment; obtaining basic information according to the measurement equipment information, obtaining a confidence scoring function according to the basic information, and fusing the equipment measurement results according to the fusion weight coefficient to obtain a fusion result; constructing settlement time sequence data according to a fusion result, constructing an anomaly detection model by combining key change characteristics, if the settlement rate is judged to continuously exceed a set threshold value, performing early warning prompt, and marking the stage as a potential mutation interval; and according to the risk label, a cause tracing suggestion is provided, and according to the identified settlement trend state and the risk level, the scheme of each monitoring point is adjusted, so that the problem that the roadbed settlement process cannot be detected in time in the existing roadbed settlement detection process is solved. And when the settlement exceeds a threshold value, timely discovery and processing cannot be carried out.

Description

Technical Field

[0001] The present invention belongs to the technical field of road construction, and in particular relates to a roadbed settlement safety monitoring method and system. Background Art

[0002] During the construction of infrastructure such as roads, railways, and water conservancy projects, the subgrade, as a critical load-bearing layer for structural stability and load transfer, has a significant impact on the safe operation of the project. Particularly in complex geological environments such as soft soil foundations, high fills, and complex construction disturbances, the subsidence process can exhibit risk characteristics such as nonlinear acceleration and sudden subsidence. Failure to identify and intervene promptly can easily lead to pavement cracking, structural deformation, and even overall instability. Therefore, dynamic monitoring of subgrade settlement and risk warnings are crucial technical measures to ensure project quality and operational safety.

[0003] In the existing technology, leveling, GNSS positioning, tilt sensing, fiber optic sensing and other means are often used to observe settlement.

[0004] However, the existing roadbed settlement process cannot be detected in time, and when the settlement exceeds the threshold, it cannot be discovered and handled in time. Summary of the Invention

[0005] The purpose of the embodiment of the present invention is to provide a roadbed settlement safety monitoring method, which aims to solve the problem raised in the third part of the background technology.

[0006] The embodiment of the present invention is implemented as follows: a roadbed settlement safety monitoring method, the method comprising:

[0007] Obtaining a measuring device, the device including a digital level, a micro shock absorbing platform, and a guide stabilizer, and obtaining device information based on the device, the device information including device measurement results and device coordinates;

[0008] Obtain basic information based on the measurement device information, obtain a confidence scoring function based on the basic information, use the confidence of each data source as an input feature, fuse the device measurement results based on the fusion weight coefficient, and obtain a fusion result;

[0009] Based on the fusion results, sedimentation time series data is constructed, and an anomaly detection model is built based on key change characteristics. If the sedimentation rate is determined to exceed the set threshold continuously, an early warning prompt is issued and the stage is marked as a potential mutation interval.

[0010] Determine the risk level based on the potential mutation interval, output structured risk labels based on the risk level, provide cause tracing suggestions based on the risk labels, and adjust the plan for each monitoring point based on the identified settlement trend status and risk level.

[0011] Preferably, the steps of obtaining basic information based on measurement device information, obtaining a confidence scoring function based on the basic information, taking the confidence of each data source as an input feature, fusing the device measurement results based on a fusion weight coefficient, and obtaining a fusion result specifically include:

[0012] Acquire basic information based on measurement equipment information, the basic information including data fluctuation rate within the observation time window, antenna height, shielding factor and weather parameters;

[0013] Obtaining a confidence scoring function based on the basic information. The confidence scoring function is a function model for evaluating the credibility of the basic information and is used to represent the weight priority of the data in the current environment;

[0014] The confidence of each data source is used as input feature to train a set of fusion weight coefficients. The device measurement results are fused according to the fusion weight coefficients to obtain the fusion result.

[0015] Preferably, the step of constructing sedimentation time series data based on the fusion results, constructing an anomaly detection model based on key change characteristics, and issuing an early warning if it is determined that the sedimentation rate continuously exceeds a set threshold, and marking the stage as a potential mutation interval, specifically includes:

[0016] Constructing settlement time series data based on the fusion results, and continuously predicting the settlement process based on the settlement time series data. The prediction adopts the GRU time series neural network model to obtain the prediction results, and outputs the trend change map in real time according to the prediction results;

[0017] Obtain key change features based on the prediction results, including sedimentation rate, acceleration, and curvature, and construct an anomaly detection model based on the key change features. The constructed anomaly detection model is used to identify nonlinear fluctuations or mutations within a short period;

[0018] If the sedimentation rate is determined to exceed the set threshold continuously, an early warning will be issued and the stage will be marked as a potential mutation interval.

[0019] Preferably, the steps of determining the risk level according to the potential mutation interval, outputting a structured risk label according to the risk level, providing a cause tracing suggestion according to the risk label, and adjusting the plan for each monitoring point according to the identified subsidence trend state and risk level specifically include:

[0020] Determine the risk level based on the potential mutation interval, the risk level including high risk and low risk, and output a structured risk label based on the risk level, the output structured risk label including the risk level and treatment method;

[0021] Provide cause tracing suggestions based on risk tags. The tracing suggestions are used to assist managers in emergency response and governance decision-making. The tracing suggestions are sent to the terminal based on the identified subsidence trend status and risk level;

[0022] A plan for adjusting each monitoring point includes changing the observation frequency, which includes enabling a standby station to participate in the observation and switching to a low power consumption mode to lower the measurement frequency.

[0023] Preferably, the fusion weight coefficient is used to automatically adjust the influence of each data source in the fusion output.

[0024] Another object of an embodiment of the present invention is to provide a roadbed settlement safety monitoring system, the system comprising:

[0025] A measuring device module is configured to obtain a measuring device, including a digital level, a micro shock absorbing platform, and a guide stabilizer, and obtain device information based on the device, including the device measurement results and device coordinates.

[0026] The fusion module obtains basic information based on the measurement device information, obtains a confidence scoring function based on the basic information, takes the confidence of each data source as an input feature, and fuses the device measurement results according to the fusion weight coefficient to obtain a fusion result;

[0027] The anomaly detection module constructs sedimentation time series data based on the fusion results and builds an anomaly detection model based on key change characteristics. If the sedimentation rate is determined to exceed the set threshold continuously, an early warning prompt will be issued and the stage will be marked as a potential mutation interval.

[0028] The risk assessment module determines the risk level based on the potential mutation interval, outputs structured risk labels based on the risk level, provides cause tracing suggestions based on the risk labels, and adjusts the plan for each monitoring point based on the identified settlement trend status and risk level.

[0029] Preferably, the fusion module includes:

[0030] A basic information unit, which obtains basic information based on measurement equipment information, wherein the basic information includes data fluctuation rate, antenna height, shielding factor and weather parameters within the observation time window;

[0031] A scoring function model unit, which obtains a confidence scoring function based on the basic information. The confidence scoring function is a function model for evaluating the credibility of the basic information and is used to represent the weight priority of the data in the current environment;

[0032] The fusion unit takes the confidence of each data source as input feature, trains a set of fusion weight coefficients, fuses the device measurement results according to the fusion weight coefficients, and obtains the fusion result.

[0033] Preferably, the anomaly detection module includes:

[0034] The prediction unit constructs the settlement time series data according to the fusion results, and continuously predicts the settlement process according to the settlement time series data. The prediction adopts the GRU time series neural network model to obtain the prediction results, and outputs the trend change map in real time according to the prediction results;

[0035] A change feature unit obtains key change features based on the prediction results. The key change features include sedimentation rate, acceleration, and change curvature. An anomaly detection model is constructed based on the key change features. The constructed anomaly detection model is used to identify nonlinear fluctuations or mutations within a short period.

[0036] The anomaly detection unit will issue an early warning if it determines that the sedimentation rate exceeds the set threshold continuously, and mark the stage as a potential mutation interval.

[0037] Preferably, the risk assessment module includes:

[0038] A risk level unit determines the risk level based on the potential mutation interval, wherein the risk level includes high risk and low risk, and outputs a structured risk label based on the risk level, wherein the output structured risk label includes the risk level and the treatment method;

[0039] The tracing unit provides cause tracing suggestions based on risk tags. The tracing suggestions are used to assist managers in emergency response and governance decisions. The tracing suggestions are sent to the terminal based on the identified subsidence trend status and risk level;

[0040] The adjustment unit adjusts the plan of each monitoring point, the plan including changing the observation frequency, the observation frequency including enabling a standby station to participate in the observation and switching to a low power consumption mode to lower the measurement frequency.

[0041] Preferably, the fusion weight coefficient is used to automatically adjust the influence of each data source in the fusion output.

[0042] A roadbed settlement safety monitoring method provided by an embodiment of the present invention comprises the following steps: obtaining a measuring device, obtaining device information based on the device, obtaining basic information based on the measuring device information, obtaining a confidence scoring function based on the basic information, using the confidence of each data source as an input feature, training a set of fusion weight coefficients, fusing the device measurement results based on the fusion weight coefficients, obtaining a fusion result, constructing settlement time series data based on the fusion result, continuously predicting the settlement process based on the settlement time series data, obtaining a prediction result, outputting a trend change map in real time based on the prediction result, obtaining key change features based on the prediction result, and constructing an anomaly detection model based on the key change features. If the settlement rate is determined to continuously exceed a set threshold, an early warning prompt is issued, and the stage is marked as a potential mutation interval. A risk level is determined based on the potential mutation interval, a structured risk label is output based on the risk level, and a cause tracing suggestion is provided based on the risk label. The tracing suggestion is sent to a terminal. Based on the identified settlement trend state and risk level, the scheme of each monitoring point is adjusted. This solves the problem that existing roadbed settlement detection processes cannot timely detect the roadbed settlement process and cannot timely detect and handle settlement exceeding the threshold. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A flow chart of a roadbed settlement safety monitoring method provided by an embodiment of the present invention;

[0044] Figure 2 A flowchart of the steps of using the confidence of each data source as an input feature and fusing the device measurement results according to the fusion weight coefficient provided in an embodiment of the present invention;

[0045] Figure 3 A flowchart of the steps of constructing an anomaly detection model based on key change characteristics according to an embodiment of the present invention and issuing an early warning if the sedimentation rate is determined to have continuously exceeded a set threshold;

[0046] Figure 4 A flowchart of the steps of providing cause tracing suggestions based on risk tags according to the identified subsidence trend status and risk level provided in an embodiment of the present invention;

[0047] Figure 5 An architectural diagram of a roadbed settlement safety monitoring system provided by an embodiment of the present invention;

[0048] Figure 6 An architectural diagram of the fusion module provided in an embodiment of the present invention;

[0049] Figure 7 An architectural diagram of an anomaly detection module provided in an embodiment of the present invention;

[0050] Figure 8This is an architectural diagram of the risk assessment module provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0052] It is understood that the terms "first," "second," etc., used herein may be used to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, a first xx script may be referred to as a second xx script, and similarly, a second xx script may be referred to as a first xx script without departing from the scope of this application.

[0053] like Figure 1 As shown, a roadbed settlement safety monitoring method provided by an embodiment of the present invention includes:

[0054] S100, obtaining a measuring device, wherein the device includes a digital level, a micro shock absorbing platform, and a guide stabilizer, and obtaining device information based on the device, wherein the device information includes a device measurement result and a device coordinate.

[0055] In this step, obtain the measuring equipment, which includes a digital level, a miniature shock-absorbing platform, and a guide stabilizer. The digital level is a precision optical device with automatic reading and data output capabilities. It uses internal imaging and digital image recognition systems to automatically extract the readings on the standard leveling scale, avoiding the reading errors that may be caused by traditional manual reading.

[0056] The micro-vibration-isolating platform is used to improve measurement stability and is particularly suitable for vibration-prone locations such as construction sites, busy traffic areas, or areas with strong winds. It uses elastic damping structures or passive vibration isolation systems to reduce the impact of ground microseismicity on the level probe, ensuring that the imaging system maintains precise focus and steady-state observation during measurement.

[0057] The guide stabilizer is mainly used to accurately locate the position of the level rod or auxiliary aiming device to prevent the rod from swinging or tilting during observation. It plays an important role especially in areas with wind interference or limited operating space.

[0058] When acquiring device information, in addition to reading the current device measurement results, the precise coordinates of the device on site are also obtained simultaneously. These coordinates can be obtained using a matching GNSS positioning module or static registration, and are uniformly converted to the global reference coordinate system used by the monitoring project. The measurement results and coordinate data are bound and uploaded to the platform.

[0059] S200 , basic information is obtained based on the measurement device information, a confidence scoring function is obtained based on the basic information, the confidence of each data source is used as an input feature, and the device measurement results are fused according to the fusion weight coefficient to obtain a fusion result.

[0060] In this step, basic information is obtained based on the measurement equipment information. Further information related to the measurement environment is extracted, including data volatility within the observation time window, stability of the sensor or antenna installation height, surrounding obstruction, and real-time weather parameters. Each piece of basic information is quantified into standard scoring indicators and input into the confidence scoring function to generate a confidence score that measures the reliability of the measurement. This score reflects the trustworthiness of the measured data in the current environment. The system can assign weights to each indicator through linear combinations, logistic regression, or tree models to achieve accurate scoring.

[0061] The system then uses the confidence levels of data from different measuring devices or measurement points as feature inputs, calls a trained fusion weight model, and automatically calculates the weight proportions of each data source in the fusion calculation. The fusion weight reflects the degree of influence of each device's data on the final settlement results. Based on this, the system dynamically weights the multi-source measurement results and outputs a fused settlement value. This fusion mechanism not only increases the weight of high-quality data but also enhances overall robustness under conditions of data instability or interference, ensuring the accuracy and continuity of settlement monitoring results. It is particularly suitable for engineering scenarios with complex field environments and frequent disturbances.

[0062] S300: Construct sedimentation time series data based on the fusion results, and build an anomaly detection model based on key change characteristics. If it is determined that the sedimentation rate continuously exceeds the set threshold, an early warning prompt is issued and the stage is marked as a potential mutation interval.

[0063] In this step, the settlement time series data is constructed based on the fusion results. After obtaining the fused settlement results, the system constructs the settlement time series data with the observation time as the index and extracts key change features, including indicators such as settlement rate, acceleration and curvature, to comprehensively describe the settlement evolution trend. The system calculates the settlement change rate per unit time through a sliding window method, and further determines whether its growth trend is continuous, whether it presents an accelerated state or an obvious nonlinear curve shape. These dynamic features are used as input to construct an anomaly detection model. The model can be trained based on rule judgment or time series neural network to accurately identify the risk of sudden settlement.

[0064] If the system determines that the sedimentation rate continuously exceeds a set threshold, or that acceleration or curvature experiences significant sudden changes, it deems the data segment to be abnormal. The system immediately generates an alert and marks the period as a potential sudden change interval. This marking not only supports subsequent refined risk analysis but also triggers automated response mechanisms, such as increasing the sampling frequency at that point, activating backup sensors, or recommending engineering interventions to management personnel, enabling early identification and dynamic management of high-risk areas.

[0065] S400 determines the risk level based on the potential mutation interval, outputs a structured risk label based on the risk level, provides cause tracing suggestions based on the risk label, and adjusts the plan for each monitoring point based on the identified settlement trend status and risk level.

[0066] In this step, the risk level is determined based on the potential mutation interval. Once a potential mutation interval is identified, the risk level is comprehensively assessed based on key characteristics such as settlement rate, acceleration, and duration, generating a structured risk label. This label not only includes the risk level (e.g., Level I, Level II) and the change pattern (e.g., sustained accelerated settlement), but also analyzes abnormal behavior to identify possible causes (e.g., construction disturbance, rainfall-induced soft soil) and corresponding intervention recommendations. For example, if the settlement rate at a particular measuring point consistently exceeds 0.5 mm / h and is accompanied by inclination fluctuations and rainfall impacts, the system will assess it as a Level II risk and recommend manual review and a temporary suspension of construction operations.

[0067] After outputting risk labels, the monitoring plan for each monitoring point is dynamically adjusted based on the identified settlement trend and corresponding risk level. For higher-risk areas, data sampling frequency is increased, backup sensors are activated, or temporary monitoring points are added. This is also linked to the construction scheduling module to implement proactive control measures, such as postponing compaction, enhancing drainage, and initiating manual inspections. By integrating risk assessment, root cause tracing, and strategy adjustment into a closed-loop system, an intelligent monitoring and response mechanism is implemented, improving both the efficiency of response and management capabilities to sudden settlement events.

[0068] like Figure 2 As shown, as a preferred embodiment of the present invention, the steps of obtaining basic information based on measurement device information, obtaining a confidence scoring function based on the basic information, taking the confidence of each data source as an input feature, fusing the device measurement results based on a fusion weight coefficient, and obtaining a fusion result specifically include:

[0069] S201 , acquiring basic information according to measurement equipment information, where the basic information includes data fluctuation rate, antenna height, shielding factor, and weather parameters within an observation time window.

[0070] In this step, basic information is obtained based on the measurement equipment information, and basic information related to the observation environment is automatically extracted to evaluate the reliability of the current measurement data. Specifically, this includes: data volatility within the observation time window, which reflects data stability by calculating the standard deviation or range of the measurement values within a certain period of time; antenna height deviation, which compares the current device installation height with the initial reference height to identify whether the device has moved or is installed abnormally; occlusion factor, which evaluates the surrounding visible sky angle in the 3D scene model based on the device coordinates to determine whether there are any structural obstructions affecting the observation; and weather parameters, which obtain wind speed, rainfall, and other data through a real-time meteorological interface to identify external factors such as wind disturbances or rain and snow interference.

[0071] This basic information is standardized and scored, serving as a key input for constructing the subsequent confidence scoring function. Each basic indicator is assigned a scoring weight, reflecting its actual impact on the measurement's credibility. For example, in strong winds accompanied by rain, low wind speed and weather scores will directly lower the overall confidence level of the measurement result. Similarly, in areas with severe obstruction or significant data fluctuations, the system will also assess low obstruction and stability scores. Through this mechanism, the system achieves a quantitative assessment of measurement data quality, laying the foundation for subsequent fusion and anomaly identification.

[0072] S202: Obtain a confidence scoring function based on the basic information. The confidence scoring function is a function model for evaluating the credibility of the basic information and is used to represent the weight priority of the data in the current environment.

[0073] In this step, a confidence scoring function is derived from the basic information. This function is a functional model used to assess the reliability of measurement data in the current environment. Its input is a number of standardized basic information indicators, including data volatility within the observation time window, antenna or device height deviation, obstruction factor, and weather parameters. The system normalizes these indicators and converts them into a score between 0 and 1, reflecting the degree of influence of each factor on the reliability of the measurement data. For example, if the data at a certain measurement point has low fluctuation, a stable antenna height, no obstructions, and good weather, all scores will be high, indicating that the measurement result is stable under environmental interference.

[0074] Based on the actual impact of each indicator, the system can construct a confidence scoring function using a linear weighted model or machine learning methods, ultimately outputting a unified confidence score (e.g., 0.89) to measure the credibility of the data. This score serves as the weighting basis for subsequent data fusion processing, with high-confidence data taking the lead in the fusion process, while low-confidence data is automatically reduced in influence or marked for review. This mechanism ensures that the system can dynamically assess data quality under various construction interferences or complex environmental conditions, improving the stability and accuracy of overall monitoring results.

[0075] S203: Using the confidence level of each data source as an input feature, a set of fusion weight coefficients is trained. The fusion weight coefficients are used to automatically adjust the influence of each data source in the fusion output. The device measurement results are fused according to the fusion weight coefficients to obtain a fusion result.

[0076] In this step, the confidence level of each data source is used as an input feature. After obtaining the confidence scores for each data source, the system uses these scores as input features and calls a trained fusion weight model to automatically calculate the participation weight of each data source in the current environment. This model can use linear regression, weighted average, or neural network algorithms, and trains by comparing historical data with actual settlement values to derive the optimal weight allocation strategy. Data sources with higher confidence levels receive greater weight in the fusion calculation, while those with lower confidence levels receive less influence. This allows for dynamic weighing and precision optimization of multi-source data in different interference environments.

[0077] The measurement results from each device are then weighted according to the fusion weight coefficient, and a fused settlement value is output. This fusion result integrates reliable data from various devices under different operating conditions, effectively reducing the deviation caused by single measurement errors and improving the overall stability and continuity of monitoring results. For example, in a certain observation, the level, GNSS, and inclinometer each produced different settlement values. The system automatically assigns weights based on their confidence levels and fuses them to produce the final result, which can be used for subsequent trend analysis and early warning assessments, ensuring high-precision monitoring capabilities even in complex environments.

[0078] like Figure 3 As shown, as a preferred embodiment of the present invention, the steps of constructing sedimentation time series data based on the fusion results, constructing an anomaly detection model based on key change characteristics, and issuing an early warning if it is determined that the sedimentation rate continuously exceeds a set threshold, and marking the stage as a potential mutation interval, specifically include:

[0079] S301, constructing settlement time series data based on the fusion results, and performing continuous prediction of the settlement process based on the settlement time series data. The prediction adopts the GRU time series neural network model to obtain the prediction results, and outputs the trend change map in real time according to the prediction results.

[0080] In this step, settlement time series data is constructed based on the fusion results. After obtaining the fused settlement results, settlement time series data is constructed based on the chronological order, converting the settlement changes at each monitoring point into dynamic input samples for model analysis. To achieve continuous prediction of future settlement trends, the system uses a GRU (Gated Recurrent Unit) temporal neural network model. This model extracts historical settlement sequences as training input using a sliding window approach and performs regression predictions on settlement values at future time steps. The GRU model is capable of capturing data features with strong time dependence and significant nonlinear trends, effectively adapting to both slow and sudden changes in foundation settlement data.

[0081] In actual operation, the trained GRU model uses the latest set of settlement sequences as real-time input to predict settlement trends at several future time points and dynamically outputs them in graphical form, such as a continuous line graph or trend heat map. If the prediction results indicate that settlement is accelerating or approaching a preset safety threshold, the system will automatically issue an alert, indicating areas of potential subsidence risk. By introducing this prediction mechanism, the system is able to identify and proactively judge settlement trends, providing data support for the early development of intervention measures and significantly improving the early warning response efficiency of roadbed settlement monitoring.

[0082] S302, obtaining key change features based on the prediction results, wherein the key change features include sedimentation rate, acceleration, and change curvature, and constructing an anomaly detection model based on the key change features. The constructed anomaly detection model is used to identify nonlinear fluctuations or mutations within a short period.

[0083] In this step, key change features are obtained based on the prediction results. After obtaining the predicted settlement results, the system further extracts key change features, including settlement rate, acceleration, and curvature, to describe the dynamic change process of the settlement trend. The settlement rate reflects the settlement increment per unit time, the acceleration measures the trend of the settlement rate, and the curvature is used to determine whether the trend has nonlinear bending. The system calculates these characteristic indicators using a sliding window. For example, if the settlement rate in the prediction sequence increases from 0.24mm / h to 0.32mm / h, the acceleration reaches 0.12mm / h², and the curvature increases significantly, it indicates that there may be a mutation signal during this period.

[0084] Integrating these characteristics, the system constructs an anomaly detection model, employing a combination of threshold judgment and time-series anomaly learning models to identify nonlinear fluctuations or sudden changes within short periods. When multiple indicators continuously exceed preset thresholds, or when the predicted trend deviates significantly from the normal pattern, the system automatically marks that time period as an abnormal period and issues an early warning. This mechanism not only identifies current abnormal conditions but also provides the ability to proactively detect potential future subsidence risks, significantly improving the monitoring system's response speed and intelligence to sudden subsidence issues.

[0085] S303: If it is determined that the sedimentation rate continuously exceeds the set threshold, an early warning prompt is issued and the stage is marked as a potential mutation interval.

[0086] In this step, if the sedimentation rate is determined to have continuously exceeded the set threshold, when the prediction results identify a situation where the sedimentation rate continuously exceeds the set threshold, it is determined that there is an accelerated sedimentation trend within this period, and the abnormal warning mechanism is triggered. The sedimentation rate between multiple consecutive time points is calculated using a sliding window method and compared with the preset rate threshold. For example, if the threshold is set to 0.30 mm / h, when the rate reaches 0.44, 0.48, and 0.50 mm / h for three consecutive time periods, it is determined to be a continuous over-limit state, triggering a level 1 warning prompt. At the same time, this time period is automatically marked as a potential mutation interval for subsequent key monitoring and response processing.

[0087] To enhance the accuracy of judgments, the system also incorporates auxiliary parameters such as rate duration and rate growth, ensuring that warnings are triggered only when the trend's continuity and intensity meet specific criteria. Once a potential sudden change interval is marked, the sampling frequency at that point is automatically increased, backup monitoring equipment is activated, and the early warning information push module is linked to notify engineering managers or the system platform for intervention. This mechanism effectively identifies and responds to sudden settlement events, enhancing the settlement monitoring system's dynamic perception and intelligent response capabilities under complex working conditions.

[0088] like Figure 4 As shown, as a preferred embodiment of the present invention, the steps of determining the risk level according to the potential mutation interval, outputting a structured risk label according to the risk level, providing a cause tracing suggestion according to the risk label, and adjusting the plan of each monitoring point according to the identified settlement trend state and risk level specifically include:

[0089] S401, determining a risk level according to a potential mutation interval, wherein the risk level includes high risk and low risk, and outputting a structured risk label according to the risk level, wherein the output structured risk label includes the risk level and a handling method.

[0090] In this step, the risk level is determined based on the potential mutation interval. Once a potential mutation interval is identified, a comprehensive assessment of the settlement characteristics within that interval is conducted. The section is then classified as high or low risk based on factors such as settlement rate, acceleration, duration of the change, and the presence of superimposed interference factors. High risk typically manifests when the settlement rate significantly exceeds a threshold, acceleration continues to rise, or is accompanied by other abnormal signals (such as persistent rainfall or structural tilt). Low risk refers to short-term fluctuations with no persistence or the absence of significant interference factors. This risk level determination provides a basis for subsequent response decisions.

[0091] Based on the risk level, structured risk tags are automatically generated, including the "risk level" and the corresponding "handling method." For high-risk situations, the system will output emergency strategies such as "immediately increase sampling frequency, activate backup measurement points, and suspend construction." For low-risk situations, routine monitoring and observation will be maintained. These tags are simultaneously pushed to the platform interface and management end, enabling standardized output of risk information and differentiated dispatch responses, enhancing the monitoring system's emergency response capabilities and project adaptability in the event of sudden changes.

[0092] S402, providing cause tracing suggestions based on risk tags, the tracing suggestions are used to assist managers in emergency response and governance decision-making, and sending tracing suggestions to the terminal based on the identified settlement trend status and risk level.

[0093] In this step, causal tracing suggestions are provided based on risk tags. After generating risk tags, the system automatically provides causal tracing suggestions based on the settlement trend status and risk level to assist management personnel in emergency response and decision-making. Causal inference is performed by analyzing the settlement rate, acceleration, and change trends within the abnormal section, as well as external influencing factors (such as construction logs, rainfall records, and sensor interference). Common causal patterns are matched based on the engineering scenario knowledge graph and rule base. For example, when "continuous accelerated settlement" is identified, accompanied by recent heavy rainfall and compaction operations, it is inferred that "post-rain soil softening + construction disturbance" may be the main cause of the anomaly, and treatment recommendations such as suspending construction and strengthening drainage are proposed.

[0094] The generated tracing recommendations are pushed to the management terminal in a structured manner, including risk level, trend status, possible causes, and recommended operational measures, ensuring that managers can immediately grasp the source of the anomaly and the response strategy. For low-risk situations, continuous observation and follow-up are recommended; for high-risk situations, proactive intervention measures are triggered, such as adjusting the construction plan, strengthening foundation treatment, or arranging manual inspections. This mechanism combines risk identification with cause analysis to establish an intelligently responsive decision-making support system, significantly improving the system's guiding value and application efficiency in subsidence management.

[0095] S403: Adjusting a plan for each monitoring point, the plan including changing the observation frequency, including enabling a standby station to participate in the observation and switching to a low power consumption mode to lower the measurement frequency.

[0096] In this step, the plan for each monitoring point is adjusted. After identifying the risk level and settlement trend of each monitoring point, the corresponding monitoring plan will be automatically adjusted, with a focus on dynamic control of observation frequency. For monitoring points judged to be high-risk or showing accelerated settlement trends, the observation frequency will be increased, such as shortening the sampling interval from 30 minutes to 5 minutes, and automatically activating backup stations that were previously on standby (such as additional GNSS modules or inclination sensors) for collaborative observation, thus forming a redundant monitoring structure and enhancing monitoring density and data continuity. For example, after identifying a sudden change trend in the abutment area, the system will switch the main device to high-frequency mode and activate surrounding auxiliary nodes to participate in the observation, achieving multi-source coverage of key areas.

[0097] For monitoring points with lower risk levels and stable settlement trends, the system switches to low-power monitoring mode to reduce energy consumption and data redundancy. The observation frequency for these points will be lowered, such as adjusting the sampling period from every 30 minutes to every 2 hours. The device will then enter periodic dormancy and automatically wake up for observation only when a specific triggering event (such as rainfall or vibration) is detected. For example, in a stable slope area, if the system detects no significant fluctuations in settlement values for 72 consecutive hours, it will automatically switch to a low-frequency sampling strategy. By dynamically adjusting the observation frequency, the system achieves intelligent allocation and efficient utilization of monitoring resources, ensuring timely responses in key areas and energy-efficient operation in low-risk areas.

[0098] like Figure 5 As shown, a roadbed settlement safety monitoring system provided by an embodiment of the present invention includes:

[0099] The measuring device module 100 is used to obtain the measuring device, which includes a digital level, a micro shock absorbing platform and a guide stabilizer, and obtain device information based on the device, which includes the device measurement results and device coordinates.

[0100] In this system, the measuring device module 100 obtains the measuring device, which includes a digital level, a miniature shock-absorbing platform, and a guide stabilizer. The digital level is a precision optical device with automatic reading and data output capabilities. Through internal imaging and digital image recognition systems, it automatically extracts the readings on the standard leveling rod, avoiding the reading errors that may be caused by traditional manual reading.

[0101] The micro-vibration-isolating platform is used to improve measurement stability and is particularly suitable for vibration-prone locations such as construction sites, busy traffic areas, or areas with strong winds. It uses elastic damping structures or passive vibration isolation systems to reduce the impact of ground microseismicity on the level probe, ensuring that the imaging system maintains precise focus and steady-state observation during measurement.

[0102] The guide stabilizer is mainly used to accurately locate the position of the level rod or auxiliary aiming device to prevent the rod from swinging or tilting during observation. It plays an important role especially in areas with wind interference or limited operating space.

[0103] When acquiring device information, in addition to reading the current device measurement results, the precise coordinates of the device on site are also obtained simultaneously. These coordinates can be obtained using a matching GNSS positioning module or static registration, and are uniformly converted to the global reference coordinate system used by the monitoring project. The measurement results and coordinate data are bound and uploaded to the platform.

[0104] The fusion module 200 is used to obtain basic information based on the measurement device information, obtain a confidence scoring function based on the basic information, use the confidence of each data source as an input feature, fuse the device measurement results according to the fusion weight coefficient, and obtain a fusion result.

[0105] In this system, the fusion module 200 obtains basic information based on the measurement device information and further extracts basic information related to the measurement environment, including data volatility within the observation time window, stability of the sensor or antenna installation height, surrounding obstruction, and real-time weather parameters. Each piece of basic information is quantified into a standard scoring metric and input into a confidence scoring function to generate a confidence score that measures the reliability of the measurement. This score reflects the trustworthiness of the measured data in the current environment. The system can assign weights to each metric through linear combinations, logistic regression, or tree models to achieve accurate scoring.

[0106] The system then uses the confidence levels of data from different measuring devices or measurement points as feature inputs, calls a trained fusion weight model, and automatically calculates the weight proportions of each data source in the fusion calculation. The fusion weight reflects the degree of influence of each device's data on the final settlement results. Based on this, the system dynamically weights the multi-source measurement results and outputs a fused settlement value. This fusion mechanism not only increases the weight of high-quality data but also enhances overall robustness under conditions of data instability or interference, ensuring the accuracy and continuity of settlement monitoring results. It is particularly suitable for engineering scenarios with complex field environments and frequent disturbances.

[0107] The anomaly detection module 300 is used to construct sedimentation time series data based on the fusion results, and to build an anomaly detection model based on key change characteristics. If it is determined that the sedimentation rate continuously exceeds the set threshold, an early warning prompt is issued and the stage is marked as a potential mutation interval.

[0108] In this system, the anomaly detection module 300 constructs settlement time series data based on the fusion results. After obtaining the fused settlement results, the system constructs settlement time series data with the observation time as the index and extracts key change features, including indicators such as settlement rate, acceleration and curvature, to comprehensively describe the settlement evolution trend. The system calculates the settlement change rate per unit time through a sliding window method, and further determines whether its growth trend is continuous, whether it presents an accelerated state or an obvious nonlinear curve shape. These dynamic features are used as input to construct an anomaly detection model. The model can be trained based on rule judgment or time series neural network to accurately identify the risk of sudden settlement.

[0109] If the system determines that the sedimentation rate continuously exceeds a set threshold, or that acceleration or curvature experiences significant sudden changes, it deems the data segment to be abnormal. The system immediately generates an alert and marks the period as a potential sudden change interval. This marking not only supports subsequent refined risk analysis but also triggers automated response mechanisms, such as increasing the sampling frequency at that point, activating backup sensors, or recommending engineering interventions to management personnel, enabling early identification and dynamic management of high-risk areas.

[0110] The risk assessment module 400 is used to determine the risk level based on the potential mutation interval, output structured risk labels based on the risk level, provide cause tracing suggestions based on the risk labels, and adjust the plan for each monitoring point based on the identified settlement trend status and risk level.

[0111] In this system, the risk assessment module 400 determines the risk level based on the potential mutation interval. Once a potential mutation interval is identified, the risk level is comprehensively assessed based on key characteristics such as settlement rate, acceleration, and duration, generating a structured risk label. This label not only includes the risk level (e.g., Level I, Level II) and the change pattern (e.g., sustained accelerated settlement), but also analyzes abnormal behavior to identify possible causes (e.g., construction disturbance, rainfall-induced soft soil) and corresponding intervention recommendations. For example, if the settlement rate at a particular measuring point consistently exceeds 0.5 mm / h and is accompanied by inclination fluctuations and rainfall impacts, the system will assess it as a Level II risk and recommend manual review and a temporary suspension of construction operations.

[0112] After outputting risk labels, the monitoring plan for each monitoring point is dynamically adjusted based on the identified settlement trend and corresponding risk level. For higher-risk areas, data sampling frequency is increased, backup sensors are activated, or temporary monitoring points are added. This is also linked to the construction scheduling module to implement proactive control measures, such as postponing compaction, enhancing drainage, and initiating manual inspections. By integrating risk assessment, root cause tracing, and strategy adjustment into a closed-loop system, an intelligent monitoring and response mechanism is implemented, improving both the efficiency of response and management capabilities to sudden settlement events.

[0113] like Figure 6 As shown, as a preferred embodiment of the present invention, the fusion module 200 includes:

[0114] The basic information unit 201 is used to obtain basic information according to the measurement equipment information. The basic information includes data fluctuation rate, antenna height, shielding factor and weather parameters within the observation time window.

[0115] In this module, the basic information unit 201 obtains basic information based on the measurement equipment information and automatically extracts basic information related to the observation environment to evaluate the reliability of the current measurement data. Specifically, this includes: data volatility within the observation time window, which reflects data stability by calculating the standard deviation or range of the measurement values within a certain period of time; antenna height deviation, which identifies whether the device has moved or is installed abnormally by comparing the current device installation height with the initial reference height; occlusion factor, which evaluates the surrounding visible sky angle in the 3D scene model based on the device coordinates to determine whether there are any structural obstructions affecting the observation; and weather parameters, which obtains wind speed, rainfall, and other data through a real-time meteorological interface to identify external factors such as wind disturbances or rain and snow interference.

[0116] This basic information is standardized and scored, serving as a key input for constructing the subsequent confidence scoring function. Each basic indicator is assigned a scoring weight, reflecting its actual impact on the measurement's credibility. For example, in strong winds accompanied by rain, low wind speed and weather scores will directly lower the overall confidence level of the measurement result. Similarly, in areas with severe obstruction or significant data fluctuations, the system will also assess low obstruction and stability scores. Through this mechanism, the system achieves a quantitative assessment of measurement data quality, laying the foundation for subsequent fusion and anomaly identification.

[0117] The scoring function model unit 202 is used to obtain a confidence scoring function based on the basic information. The confidence scoring function is a function model for evaluating the credibility of the basic information and is used to represent the weight priority of the data in the current environment.

[0118] In this module, the scoring function model unit 202 obtains a confidence scoring function based on basic information. The confidence scoring function is a functional model used to assess the credibility of measurement data in the current environment. Its input is multiple standardized basic information indicators, including data volatility within the observation time window, antenna or device height deviation, obstruction factor, and weather parameters. The system normalizes these indicators and converts them into a score between 0 and 1, reflecting the degree of influence of each factor on the reliability of the measurement data. For example, if the data fluctuation at a certain measurement point is small, the antenna height is stable, there is no obstruction, and the weather is good, all scores are high, indicating that the measurement result is stable under environmental interference.

[0119] Based on the actual impact of each indicator, the system can construct a confidence scoring function using a linear weighted model or machine learning methods, ultimately outputting a unified confidence score (e.g., 0.89) to measure the credibility of the data. This score serves as the weighting basis for subsequent data fusion processing, with high-confidence data taking the lead in the fusion process, while low-confidence data is automatically reduced in influence or marked for review. This mechanism ensures that the system can dynamically assess data quality under various construction interferences or complex environmental conditions, improving the stability and accuracy of overall monitoring results.

[0120] The fusion unit 203 is used to use the confidence of each data source as an input feature to train a set of fusion weight coefficients. The fusion weight coefficients are used to automatically adjust the influence of each data source in the fusion output, and the device measurement results are fused according to the fusion weight coefficients to obtain a fusion result.

[0121] In this module, the fusion unit 203 uses the confidence level of each data source as an input feature. After obtaining the confidence scores of each data source, the system uses these scores as input features, calls a trained fusion weight model, and automatically calculates the participation weight of each data source in the current environment. This model can use linear regression, weighted average, or neural network algorithms to derive the optimal weight allocation strategy through comparative training of historical data and actual settlement values. The more confident the data source, the greater its weight in the fusion calculation, and vice versa, its influence is reduced, thereby achieving dynamic weighing and precision optimization of multi-source data in different interference environments.

[0122] The measurement results from each device are then weighted according to the fusion weight coefficient, and a fused settlement value is output. This fusion result integrates reliable data from various devices under different operating conditions, effectively reducing the deviation caused by single measurement errors and improving the overall stability and continuity of monitoring results. For example, in a certain observation, the level, GNSS, and inclinometer each produced different settlement values. The system automatically assigns weights based on their confidence levels and fuses them to produce the final result, which can be used for subsequent trend analysis and early warning assessments, ensuring high-precision monitoring capabilities even in complex environments.

[0123] like Figure 7 As shown, as a preferred embodiment of the present invention, the anomaly detection module 300 includes:

[0124] The prediction unit 301 is used to construct settlement time series data based on the fusion results, and to continuously predict the settlement process based on the settlement time series data. The prediction adopts the GRU time series neural network model to obtain the prediction results, and output the trend change map in real time according to the prediction results.

[0125] In this module, prediction unit 301 constructs settlement time series data based on the fusion results. After obtaining the fused settlement results, the settlement time series data is constructed based on the chronological order, converting the settlement changes at each monitoring point into dynamic input samples for model analysis. To achieve continuous prediction of future settlement trends, the system uses a GRU (Gated Recurrent Unit) temporal neural network model. This model extracts historical settlement sequences as training input using a sliding window approach and performs regression predictions on settlement values at future time steps. The GRU model is capable of capturing data features with strong time dependence and significant nonlinear trends, effectively adapting to both slow and sudden changes in foundation settlement data.

[0126] In actual operation, the trained GRU model uses the latest set of settlement sequences as real-time input to predict settlement trends at several future time points and dynamically outputs them in graphical form, such as a continuous line graph or trend heat map. If the prediction results indicate that settlement is accelerating or approaching a preset safety threshold, the system will automatically issue an alert, indicating areas of potential subsidence risk. By introducing this prediction mechanism, the system is able to identify and proactively judge settlement trends, providing data support for the early development of intervention measures and significantly improving the early warning response efficiency of roadbed settlement monitoring.

[0127] The change feature unit 302 is used to obtain key change features based on the prediction results. The key change features include sedimentation rate, acceleration and change curvature, and to construct an anomaly detection model based on the key change features. The constructed anomaly detection model is used to identify nonlinear fluctuations or mutations within a short period.

[0128] In this module, the change feature unit 302 obtains key change features based on the prediction results. After obtaining the predicted settlement results, the system further extracts key change features, including settlement rate, acceleration, and change curvature, to describe the dynamic change process of the settlement trend. The settlement rate reflects the settlement increment per unit time, the acceleration measures the change trend of the settlement rate, and the change curvature is used to determine whether the trend has nonlinear bending. The system calculates these characteristic indicators through a sliding window. For example, if the settlement rate in the prediction sequence increases from 0.24mm / h to 0.32mm / h, the acceleration reaches 0.12mm / h², and the curvature increases significantly, it means that there may be a mutation signal during this period.

[0129] Integrating these characteristics, the system constructs an anomaly detection model, employing a combination of threshold judgment and time-series anomaly learning models to identify nonlinear fluctuations or sudden changes within short periods. When multiple indicators continuously exceed preset thresholds, or when the predicted trend deviates significantly from the normal pattern, the system automatically marks that time period as an abnormal period and issues an early warning. This mechanism not only identifies current abnormal conditions but also provides the ability to proactively detect potential future subsidence risks, significantly improving the monitoring system's response speed and intelligence to sudden subsidence issues.

[0130] The anomaly detection unit 303 is used to issue an early warning if it determines that the sedimentation rate exceeds the set threshold continuously, and mark the stage as a potential mutation interval.

[0131] In this module, if the anomaly detection unit 303 determines that the sedimentation rate continuously exceeds the set threshold, when the prediction result identifies that the sedimentation rate continuously exceeds the set threshold, it is determined that there is an accelerated sedimentation trend within the period, and the anomaly warning mechanism is triggered. The sedimentation rate between multiple consecutive time points is calculated using a sliding window method and compared with the preset rate threshold. For example, if the threshold is set to 0.30 mm / h, when the rate reaches 0.44, 0.48 and 0.50 mm / h for three consecutive time periods, it is determined to be a continuous over-limit state, triggering a first-level warning prompt. At the same time, this time period is automatically marked as a potential mutation interval for subsequent key monitoring and response processing.

[0132] To enhance the accuracy of judgments, the system also incorporates auxiliary parameters such as rate duration and rate growth, ensuring that warnings are triggered only when the trend's continuity and intensity meet specific criteria. Once a potential sudden change interval is marked, the sampling frequency at that point is automatically increased, backup monitoring equipment is activated, and the early warning information push module is linked to notify engineering managers or the system platform for intervention. This mechanism effectively identifies and responds to sudden settlement events, enhancing the settlement monitoring system's dynamic perception and intelligent response capabilities under complex working conditions.

[0133] like Figure 8 As shown, as a preferred embodiment of the present invention, the risk assessment module 400 includes:

[0134] The risk level unit 401 is used to determine the risk level according to the potential mutation interval, the risk level includes high risk and low risk, and output a structured risk label according to the risk level, the output structured risk label includes the risk level and processing method.

[0135] In this module, risk level unit 401 determines the risk level based on the potential sudden change interval. Once a potential sudden change interval is identified, a comprehensive assessment of the settlement characteristics within that interval is conducted. Based on factors such as settlement rate, acceleration, duration of change, and the presence of superimposed interference factors, the section is classified as high or low risk. High risk typically manifests when settlement rate significantly exceeds a threshold, acceleration continues to rise, or is accompanied by other abnormal signals (such as persistent rainfall or structural tilt). Low risk refers to short-term fluctuations with no persistence or the absence of significant interference factors. This risk level determination provides a basis for subsequent response decisions.

[0136] Based on the risk level, structured risk tags are automatically generated, including the "risk level" and the corresponding "handling method." For high-risk situations, the system will output emergency strategies such as "immediately increase sampling frequency, activate backup measurement points, and suspend construction." For low-risk situations, routine monitoring and observation will be maintained. These tags are simultaneously pushed to the platform interface and management end, enabling standardized output of risk information and differentiated dispatch responses, enhancing the monitoring system's emergency response capabilities and project adaptability in the event of sudden changes.

[0137] The tracing unit 402 is used to provide cause tracing suggestions based on the risk tags. The tracing suggestions are used to assist managers in emergency response and governance decisions, and send tracing suggestions to the terminal based on the identified settlement trend status and risk level.

[0138] In this module, the traceability unit 402 provides cause tracing suggestions based on risk tags. After generating the risk tags, the system automatically provides cause tracing suggestions based on the settlement trend status and risk level to assist management personnel in emergency response and decision-making. By analyzing the settlement rate, acceleration, change trend, and external influencing factors (such as construction logs, rainfall records, and sensor interference) within the abnormal section, causal inference is performed, and common causal patterns are matched based on the engineering scenario knowledge graph and rule base. For example, when "continuous accelerated settlement" is identified and accompanied by recent heavy rainfall and compaction operations, it is inferred that "post-rain soil softening + construction disturbance" may be the main cause of the anomaly, and treatment suggestions such as suspending construction and strengthening drainage are proposed.

[0139] The generated tracing recommendations are pushed to the management terminal in a structured manner, including risk level, trend status, possible causes, and recommended operational measures, ensuring that managers can immediately grasp the source of the anomaly and the response strategy. For low-risk situations, continuous observation and follow-up are recommended; for high-risk situations, proactive intervention measures are triggered, such as adjusting the construction plan, strengthening foundation treatment, or arranging manual inspections. This mechanism combines risk identification with cause analysis to establish an intelligently responsive decision-making support system, significantly improving the system's guiding value and application efficiency in subsidence management.

[0140] The adjusting unit 403 is configured to adjust a plan for each monitoring point, wherein the plan includes changing the observation frequency, and the observation frequency includes enabling a standby station to participate in the observation and switching to a low power consumption mode to lower the measurement frequency.

[0141] In this module, adjustment unit 403 adjusts the plans for each monitoring point. After identifying the risk level and settlement trend of each monitoring point, it automatically adjusts the corresponding monitoring plan, focusing on the dynamic control of observation frequency. For monitoring points determined to be high-risk or showing accelerated settlement trends, the observation frequency is increased, for example, the sampling interval is shortened from 30 minutes to 5 minutes, and backup stations previously on standby (such as additional GNSS modules or inclination sensors) are automatically activated for collaborative observation, thus forming a redundant monitoring structure and enhancing monitoring density and data continuity. For example, after identifying a sudden change trend in the abutment area, the system switches the main device to high-frequency mode and activates surrounding auxiliary nodes to participate in observation, achieving multi-source coverage of key areas.

[0142] For monitoring points with lower risk levels and stable settlement trends, the system switches to low-power monitoring mode to reduce energy consumption and data redundancy. The observation frequency for these points will be lowered, such as adjusting the sampling period from every 30 minutes to every 2 hours. The device will then enter periodic dormancy and automatically wake up for observation only when a specific triggering event (such as rainfall or vibration) is detected. For example, in a stable slope area, if the system detects no significant fluctuations in settlement values for 72 consecutive hours, it will automatically switch to a low-frequency sampling strategy. By dynamically adjusting the observation frequency, the system achieves intelligent allocation and efficient utilization of monitoring resources, ensuring timely responses in key areas and energy-efficient operation in low-risk areas.

[0143] In one embodiment, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the following steps are performed:

[0144] Obtaining a measuring device, the device including a digital level, a micro shock absorbing platform, and a guide stabilizer, and obtaining device information based on the device, the device information including device measurement results and device coordinates;

[0145] Obtain basic information based on the measurement device information, obtain a confidence scoring function based on the basic information, use the confidence of each data source as an input feature, fuse the device measurement results based on the fusion weight coefficient, and obtain a fusion result;

[0146] Based on the fusion results, sedimentation time series data is constructed, and an anomaly detection model is built based on key change characteristics. If the sedimentation rate is determined to exceed the set threshold continuously, an early warning prompt is issued and the stage is marked as a potential mutation interval.

[0147] Determine the risk level based on the potential mutation interval, output structured risk labels based on the risk level, provide cause tracing suggestions based on the risk labels, and adjust the plan for each monitoring point based on the identified settlement trend status and risk level.

[0148] In one embodiment, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the processor performs the following steps:

[0149] Obtaining a measuring device, the device including a digital level, a micro shock absorbing platform, and a guide stabilizer, and obtaining device information based on the device, the device information including device measurement results and device coordinates;

[0150] Obtain basic information based on the measurement device information, obtain a confidence scoring function based on the basic information, use the confidence of each data source as an input feature, fuse the device measurement results based on the fusion weight coefficient, and obtain a fusion result;

[0151] Based on the fusion results, sedimentation time series data is constructed, and an anomaly detection model is built based on key change characteristics. If the sedimentation rate is determined to exceed the set threshold continuously, an early warning prompt is issued and the stage is marked as a potential mutation interval.

[0152] Determine the risk level based on the potential mutation interval, output structured risk labels based on the risk level, provide cause tracing suggestions based on the risk labels, and adjust the plan for each monitoring point based on the identified settlement trend status and risk level.

[0153] It should be understood that, although the various steps in the flow chart of each embodiment of the present invention are shown in sequence according to the indication of the arrows, these steps are not necessarily performed in sequence according to the order indicated by the arrows. Unless otherwise specified herein, the execution of these steps is not strictly limited in order, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0154] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When executed, the program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media used in the various embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0155] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0156] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

[0157] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A roadbed settlement safety monitoring method, characterized in that: The method comprises: Obtaining a measuring device, the device including a digital level, a micro shock absorbing platform, and a guide stabilizer, and obtaining device information based on the device, the device information including device measurement results and device coordinates; Obtain basic information based on the measurement device information, obtain a confidence scoring function based on the basic information, use the confidence of each data source as an input feature, fuse the device measurement results based on the fusion weight coefficient, and obtain a fusion result; Based on the fusion results, sedimentation time series data is constructed, and an anomaly detection model is built based on key change characteristics. If the sedimentation rate is determined to exceed the set threshold continuously, an early warning prompt is issued and the stage is marked as a potential mutation interval. Determine the risk level based on the potential mutation interval, output structured risk labels based on the risk level, provide cause tracing suggestions based on the risk labels, and adjust the plan for each monitoring point based on the identified settlement trend status and risk level.

2. A roadbed settlement safety monitoring method according to claim 1, characterized in that: The steps of obtaining basic information based on the measurement device information, obtaining a confidence scoring function based on the basic information, using the confidence of each data source as an input feature, fusing the device measurement results based on a fusion weight coefficient, and obtaining a fusion result specifically include: Acquire basic information based on measurement equipment information, the basic information including data fluctuation rate within the observation time window, antenna height, shielding factor and weather parameters; Obtaining a confidence scoring function based on the basic information. The confidence scoring function is a function model for evaluating the credibility of the basic information and is used to represent the weight priority of the data in the current environment; The confidence of each data source is used as input feature to train a set of fusion weight coefficients. The device measurement results are fused according to the fusion weight coefficients to obtain the fusion result.

3. A roadbed settlement safety monitoring method according to claim 1, characterized in that: The steps of constructing sedimentation time series data based on the fusion results, building an anomaly detection model based on key change characteristics, and issuing an early warning if it is determined that the sedimentation rate continuously exceeds a set threshold, and marking the stage as a potential mutation interval, specifically include: Constructing settlement time series data based on the fusion results, and continuously predicting the settlement process based on the settlement time series data. The prediction adopts the GRU time series neural network model to obtain the prediction results, and outputs the trend change map in real time according to the prediction results; Obtain key change features based on the prediction results, including sedimentation rate, acceleration, and curvature, and construct an anomaly detection model based on the key change features. The constructed anomaly detection model is used to identify nonlinear fluctuations or mutations within a short period; If the sedimentation rate is determined to exceed the set threshold continuously, an early warning will be issued and the stage will be marked as a potential mutation interval.

4. A roadbed settlement safety monitoring method according to claim 1, characterized in that: The steps of determining the risk level according to the potential mutation interval, outputting a structured risk label according to the risk level, providing cause tracing suggestions according to the risk label, and adjusting the plan for each monitoring point according to the identified subsidence trend state and risk level specifically include: Determine the risk level based on the potential mutation interval, the risk level including high risk and low risk, and output a structured risk label based on the risk level, the output structured risk label including the risk level and treatment method; Provide cause tracing suggestions based on risk tags. The tracing suggestions are used to assist managers in emergency response and governance decision-making. The tracing suggestions are sent to the terminal based on the identified subsidence trend status and risk level; A plan for adjusting each monitoring point includes changing the observation frequency, which includes enabling a standby station to participate in the observation and switching to a low power consumption mode to lower the measurement frequency.

5. A roadbed settlement safety monitoring method according to claim 2, characterized in that: The fusion weight coefficient is used to automatically adjust the influence of each data source in the fusion output.

6. A roadbed settlement safety monitoring system, characterized in that: The system comprises: A measuring device module is configured to obtain a measuring device, including a digital level, a micro shock absorbing platform, and a guide stabilizer, and obtain device information based on the device, including the device measurement results and device coordinates. The fusion module obtains basic information based on the measurement device information, obtains a confidence scoring function based on the basic information, takes the confidence of each data source as an input feature, and fuses the device measurement results according to the fusion weight coefficient to obtain a fusion result; The anomaly detection module constructs sedimentation time series data based on the fusion results and builds an anomaly detection model based on key change characteristics. If the sedimentation rate is determined to exceed the set threshold continuously, an early warning prompt will be issued and the stage will be marked as a potential mutation interval. The risk assessment module determines the risk level based on the potential mutation interval, outputs structured risk labels based on the risk level, provides cause tracing suggestions based on the risk labels, and adjusts the plan for each monitoring point based on the identified settlement trend status and risk level.

7. A roadbed settlement safety monitoring system according to claim 6, characterized in that: The fusion module includes: A basic information unit, which obtains basic information based on measurement equipment information, wherein the basic information includes data fluctuation rate, antenna height, shielding factor and weather parameters within the observation time window; A scoring function model unit, which obtains a confidence scoring function based on the basic information. The confidence scoring function is a function model for evaluating the credibility of the basic information and is used to represent the weight priority of the data in the current environment; The fusion unit takes the confidence of each data source as input feature, trains a set of fusion weight coefficients, fuses the device measurement results according to the fusion weight coefficients, and obtains the fusion result.

8. A roadbed settlement safety monitoring system according to claim 7, characterized in that: The anomaly detection module includes: The prediction unit constructs the settlement time series data according to the fusion results, and continuously predicts the settlement process according to the settlement time series data. The prediction adopts the GRU time series neural network model to obtain the prediction results, and outputs the trend change map in real time according to the prediction results; A change feature unit obtains key change features based on the prediction results. The key change features include sedimentation rate, acceleration, and change curvature. An anomaly detection model is constructed based on the key change features. The constructed anomaly detection model is used to identify nonlinear fluctuations or mutations within a short period. The anomaly detection unit will issue an early warning if it determines that the sedimentation rate exceeds the set threshold continuously, and mark the stage as a potential mutation interval.

9. A roadbed settlement safety monitoring system according to claim 8, characterized in that: The risk assessment module includes: A risk level unit determines the risk level based on the potential mutation interval, wherein the risk level includes high risk and low risk, and outputs a structured risk label based on the risk level, wherein the output structured risk label includes the risk level and the treatment method; The tracing unit provides cause tracing suggestions based on risk tags. The tracing suggestions are used to assist managers in emergency response and governance decisions. The tracing suggestions are sent to the terminal based on the identified subsidence trend status and risk level; The adjustment unit adjusts the plan of each monitoring point, the plan including changing the observation frequency, the observation frequency including enabling a standby station to participate in the observation and switching to a low power consumption mode to lower the measurement frequency.

10. A roadbed settlement safety monitoring system according to claim 9, characterized in that: The fusion weight coefficient is used to automatically adjust the influence of each data source in the fusion output.

Images