Municipal road subgrade settlement monitoring system and analysis method

Abstract

The invention relates to the technical field of subgrade settlement monitoring, in particular to a settlement monitoring system for a municipal road subgrade, which comprises a plurality of distributed monitoring nodes arranged at different monitoring points of the subgrade, each monitoring node comprises a sealed rigid main body of which the bottom is fixed on a roadbed rigid base, a high-precision tilt angle sensor fixedly mounted in the main body and a microcontroller unit, and the microcontroller unit is connected with the tilt angle sensor and is configured to execute an initial calibration program after the node is mounted in place and stabilized; recording a reading measured by the tilt angle sensor at the moment as an attitude initial reference value of the node; and in subsequent monitoring; the installation technology is simplified through field self-calibration, high-precision displacement calculation is achieved, real-time output of the settlement amount is guaranteed through edge calculation, multi-dimensional risk assessment and active early warning are achieved through cloud intelligent analysis, and the monitoring efficiency and the intelligent level of road maintenance are improved.

Description

Technical Field

[0001] This invention relates to the field of roadbed settlement monitoring technology, specifically to a settlement monitoring system and analysis method for municipal roadbeds. Background Technology

[0002] The stability of the roadbed is crucial for ensuring traffic safety and extending the service life of the road. After construction, the roadbed is susceptible to uneven settlement due to multiple factors such as geological conditions, construction quality, groundwater fluctuations, and traffic loads. This can lead to problems such as road surface cracking and subsidence, posing serious safety hazards. Therefore, long-term, accurate, and real-time monitoring of roadbed settlement is essential. According to CN117451000B, a machine vision detection method and system for roadbed settlement of intelligent rail transit (IRT) trains is disclosed. This technology includes the following technical solutions: S1. Image acquisition and preprocessing of Schneider-coded markers for IRT train roadbed settlement monitoring points; S2. Obtaining the center point of the preprocessed image based on the Schneider-coded markers; S3. Calculating the three-dimensional coordinates of the center point using a binocular stereo matching algorithm; S4. Decoding the Schneider-coded markers and obtaining the settlement amount of the IRT train roadbed at each monitoring point based on the decoded content and the calculated three-dimensional coordinates of the center point. This method has the technical effect of "using affine transformation to correct the image of the Schneider-coded markers at the monitoring points, thereby correcting the positioning result of the center point of the markers and improving the measurement accuracy of the roadbed settlement monitoring system." In the above scheme, the design of connecting the fiber Bragg grating sensors in series via cascaded lines to a single demodulation system results in the monitoring system's deployment and subsequent maintenance being highly dependent on long-distance, continuous, and intact physical optical cable connections. When this sensor network is applied to the subgrade environment of municipal roads requiring extensive deployment, the installation process itself requires a large amount of drilling, wiring, and welding sealing work, leading to high construction complexity and a long cycle. Furthermore, during long-term service, the subgrade's own settlement and deformation, as well as external activities such as road construction and pipeline excavation, can easily cause the fragile optical cables buried inside the subgrade or laid along the route to be stretched, bent, or even broken. Once the cable is damaged at any cascade node, the data transmission of the entire sensing link is interrupted, forming a single point of failure, and the fault location is difficult to quickly pinpoint. This makes it unsuitable for scenarios like municipal roads, which have a large geographical area, many environmental interferences, and require continuous and stable system operation. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a settlement monitoring system and analysis method for municipal road subgrades. It simplifies the installation process through on-site self-calibration and achieves high-precision displacement calculation. It utilizes edge computing to ensure real-time output of settlement data and achieves multi-dimensional risk assessment and proactive early warning through cloud-based intelligent analysis, thereby improving monitoring efficiency and the level of intelligent road maintenance.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a settlement monitoring system for municipal road subgrade, comprising: Multiple distributed monitoring nodes are deployed at different monitoring points on the roadbed. Each monitoring node includes a sealed rigid body with its bottom fixed to a rigid base on the roadbed, a high-precision tilt sensor fixedly installed inside the body, and a microcontroller unit. The microcontroller unit is connected to the tilt sensor and is configured to: execute an initial calibration program after the node is installed and stabilized, and record the readings measured by the tilt sensor at this time as the initial reference value of the node's attitude; and periodically collect and temporarily store the current attitude readings of the tilt sensor during subsequent monitoring. An edge computing gateway communicates with all distributed monitoring nodes. The edge computing gateway is configured to periodically obtain the current attitude reading and the corresponding initial attitude reference value from each node. Based on a preset geometric kinematic model, the change of the current attitude reading of each node relative to its initial attitude reference value is calculated in real time as the vertical settlement displacement of the node's location. A cloud-based data analytics platform, communicating with an edge computing gateway, receives and stores time-series data on vertical settlement displacement of each node. The platform has built-in analysis models, including: a trend prediction unit, used to model the time-series data on vertical settlement displacement of each node to predict future settlement trends; a spatial field analysis unit, used to construct and visualize the spatial distribution field of roadbed settlement based on the spatial location of all nodes and real-time settlement data; and an intelligent early warning unit, used to automatically identify and provide graded early warnings for abnormal settlement states according to preset rules.

[0005] Preferably, the tilt sensor is a dual-axis tilt sensor, used to measure the tilt angle of the rigid body relative to the horizontal plane in two mutually orthogonal directions, the X-axis and the Y-axis; the initial attitude reference value includes the initial tilt angle θ_x0, θ_y0, and the current attitude reading includes the current tilt angle θ_x, θ_y.

[0006] Preferably, the geometric kinematic model is: ΔH = L * [sin(θ_x - θ_x0) + sin(θ_y - θ_y0)], where ΔH represents the vertical settlement displacement, and L is the effective conversion length of the rigid body, defined as the vertical distance from the sensing center of the tilt sensor to the bottom fixed point of the rigid body.

[0007] Preferably, the intelligent early warning unit is configured to perform multi-indicator fusion early warning, and the early warning indicators include at least: the instantaneous settlement rate output by the trend prediction unit, the cumulative settlement based on time-series data statistics, and the differential settlement between adjacent monitoring points calculated by the spatial field analysis unit; the intelligent early warning unit is also associated with a knowledge base of disposal measures, which can recommend corresponding engineering verification or disposal measures according to the early warning type, level and location when an early warning is triggered.

[0008] This invention also discloses a settlement analysis method for municipal road subgrade, specifically including the following steps: S1. Distributed monitoring nodes are installed at the planned locations on the roadbed. After the installation is stable, the microcontroller units of each node are executed through remote commands or automatic triggering to complete the calibration and storage of the initial attitude reference values. S2, the tilt sensors of each node continuously measure at a set frequency. The edge computing gateway periodically wakes up and polls each node to obtain its current attitude reading. It calls the geometric kinematics model to calculate the vertical settlement displacement of each monitoring point in real time, and adds a timestamp and node identifier to form structured settlement time series data. S3, the edge computing gateway caches and packages the pre-processed structured settlement time series data locally, and uploads it to the cloud data analysis platform via wireless network; S4, the cloud-based data analysis platform, simultaneously performs time-series trend analysis, spatial field correlation analysis, and intelligent early warning assessment on the aggregated settlement time-series data. It processes the settlement displacement time series of each monitoring node, uses time-series prediction algorithms to fit patterns to predict future trends and calculate real-time settlement rates. Then, combining the geographic coordinates of all nodes, it uses spatial interpolation algorithms to fuse discrete settlement data to generate a continuous digital model of the roadbed settlement field, visualizes it, and calculates the global non-uniform settlement gradient field. Finally, it fuses, compares, and comprehensively assesses the real-time settlement rate, cumulative settlement, and extreme values ​​of the settlement gradient field with preset multi-level early warning thresholds. When any indicator exceeds the threshold, it automatically generates and publishes graded early warning information containing warning locations, related data, and disposal suggestions based on the threshold level.

[0009] Preferably, in step S2, the application of the geometric kinematic model is as follows: For any monitoring node, the edge computing gateway reads its current tilt angle θ_x, θ_y and its calibrated initial tilt angle θ_x0, θ_y0, calculates the angle change Δθ_x = θ_x - θ_x0, Δθ_y = θ_y - θ_y0, and substitutes them into the model ΔH = L * [sin(Δθ_x) + sin(Δθ_y)], and outputs the vertical settlement displacement ΔH of the node in real time.

[0010] Preferably, in step S4, the time series prediction algorithm used is at least one of the following: exponential smoothing, autoregressive integral moving average model, or long short-term memory neural network model.

[0011] Preferably, in the spatial field correlation analysis of step S4, the spatial interpolation algorithm is the inverse distance weighting method, the kriging method, or the spline function method; the calculation of the non-uniform settlement gradient field is achieved by performing spatial differentiation on the digital settlement field model.

[0012] Preferably, in the intelligent early warning assessment of step S4, the multi-level early warning thresholds are divided into at least three levels: monitoring level, early warning level, and alarm level, which correspond to three response states: daily monitoring, risk concern, and emergency response, respectively; the early warning information release channels include prominent display on the platform interface, automatic push via SMS and email.

[0013] This invention provides a settlement monitoring system and analysis method for municipal road subgrade. Compared with existing technologies, it has the following advantages: 1. The system, by integrating a dual-axis tilt sensor with a rigid body kinematics model, changes the stringent requirement of absolute vertical installation in traditional settlement monitoring. Its key lies in the "on-site self-calibration" mechanism: after the node installation is stable, the sensor readings are set as the sole benchmark, and subsequent measurements are based on this relative comparison, eliminating systematic errors introduced by absolute installation deviations at the source. This simplifies the installation requirement from "millimeter-level" precision adjustment to an engineering operation of "stability is sufficient." The dual-axis sensor fully captures the two-dimensional tilt vector and maps it uniquely and accurately to a one-dimensional vertical displacement through a deterministic physical model. This model is based on physical structural parameters and does not rely on easily drifting electrical signal coefficients, thus achieving traceable and high-precision absolute displacement calculation at the edge, providing a reliable data foundation for subsequent analysis.

[0014] 2. The system adopts a two-tier collaborative architecture of edge computing gateway and cloud platform, achieving optimal allocation of data processing efficiency. The edge gateway undertakes the core calculation tasks, converting raw angle data into engineering displacement in real time, avoiding delays in the transmission of massive amounts of raw data, and enabling settlement measurement to reach near real-time levels at the "minute" level, laying the foundation for rapid early warning. The cloud platform aggregates time-series data from all nodes and performs in-depth intelligent analysis in parallel: predicting single-point trends through time-series algorithms and using spatial interpolation technology to fuse discrete points into a continuous, visualized digital model of the entire road section's settlement field. This allows the system to grasp the spatial distribution of settlement from a macroscopic perspective, identify "settlement basins," and accurately calculate the key mechanical indicator of the non-uniform settlement gradient field, realizing a transformation from "single-point data acquisition" to "full-field situational awareness and mechanism diagnosis."

[0015] 3. The system transforms monitoring data into actionable decision support through multi-indicator fusion analysis and knowledge base linkage. The early warning unit integrates three complementary core indicators with complementary physical meanings: instantaneous settlement rate (short-term risk), cumulative settlement (long-term total), and differential settlement (internal stress of the structure), to construct a three-dimensional risk assessment model, effectively overcoming false alarms and missed alarms caused by single-indicator early warnings. When an early warning is triggered, the system automatically classifies the risk according to preset rules (such as monitoring, early warning, and alarm) and links with the built-in engineering disposal measures knowledge base to automatically recommend targeted verification steps or repair plans (such as key inspections and grouting reinforcement) based on the warning type, level, and location. This mechanism shortens the decision-making chain from risk discovery to disposal initiation, transforming road maintenance from a lagging and passive response relying on manual inspections to a proactive, preventative, and intelligent management system driven by data, predictable, and with precise intervention capabilities, significantly improving the level of infrastructure safety operation and maintenance and management efficiency. Attached Figure Description

[0016] Figure 1 This is a system architecture block diagram of the present invention. Detailed Implementation

[0017] 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 This invention provides a technical solution: a settlement monitoring system for municipal road subgrade, comprising: Multiple distributed monitoring nodes are deployed at different monitoring points on the roadbed. Each monitoring node includes a sealed rigid body with its bottom fixed to a rigid base on the roadbed, a high-precision tilt sensor fixedly installed inside the body, and a microcontroller unit. The microcontroller unit is connected to the tilt sensor and is configured to: execute an initial calibration program after the node is installed and stabilized, and record the readings measured by the tilt sensor at this time as the initial reference value of the node's attitude; and periodically collect and temporarily store the current attitude readings of the tilt sensor during subsequent monitoring. An edge computing gateway communicates with all distributed monitoring nodes. The edge computing gateway is configured to periodically obtain the current attitude reading and the corresponding initial attitude reference value from each node. Based on a preset geometric kinematic model, the change of the current attitude reading of each node relative to its initial attitude reference value is calculated in real time as the vertical settlement displacement of the node's location. A cloud-based data analytics platform, communicating with an edge computing gateway, receives and stores time-series data on vertical settlement displacement of each node. The platform has built-in analysis models, including: a trend prediction unit, used to model the time-series data on vertical settlement displacement of each node to predict future settlement trends; a spatial field analysis unit, used to construct and visualize the spatial distribution field of roadbed settlement based on the spatial location of all nodes and real-time settlement data; and an intelligent early warning unit, used to automatically identify and provide graded early warnings for abnormal settlement states according to preset rules.

[0019] In this implementation scheme, distributed nodes utilize high-precision tilt sensors to directly perceive micro-tilt changes in the roadbed. By executing an initial on-site calibration procedure, the sensor readings after stable installation are established as the sole measurement benchmark, thereby eliminating absolute installation errors and ensuring that all subsequent measurements are reliable changes relative to the initial state. The edge computing gateway, as a crucial computing hub, periodically collects the attitude data of each node and, based on a pre-set rigid body kinematics model, calculates these angular changes in real time into intuitive vertical settlement displacement. This edge-side calculation process significantly reduces data transmission volume and achieves near real-time settlement output. The cloud-based data analysis platform aggregates all displacement time-series data and performs in-depth analysis in parallel: the trend prediction unit... Time series algorithms uncover evolution patterns and predict future trends; the spatial field analysis unit uses spatial interpolation technology to fuse discrete monitoring point data to generate a continuous digital model of the subgrade settlement field, intuitively revealing the overall spatial distribution and non-uniformity of settlement; the intelligent early warning unit integrates multiple indicators such as instantaneous rate, cumulative settlement, and differential settlement gradient for analysis and triggers graded early warnings; through a closed-loop design of "sensing-edge computing-cloud intelligence," the traditional single-point, lagging, and manual-dependent settlement monitoring mode is transformed into a new intelligent monitoring mode that is comprehensive, real-time, automated, and predictable, significantly improving monitoring efficiency, data reliability, and the timeliness and accuracy of risk warnings, providing strong decision support for subgrade health diagnosis and preventive maintenance.

[0020] Specifically, the tilt sensor is a dual-axis tilt sensor used to measure the tilt angle of the rigid body relative to the horizontal plane in two mutually orthogonal directions, the X-axis and the Y-axis; the initial attitude reference values ​​include the initial tilt angles θ_x0 and θ_y0, and the current attitude readings include the current tilt angles θ_x and θ_y.

[0021] In this embodiment, the application of a dual-axis tilt sensor forms the physical basis for the system to achieve high-precision displacement calculation. This sensor can simultaneously capture the tilt angle components (θ_x, θ_y) of the rigid body in two mutually orthogonal directions (X-axis and Y-axis), thus completely describing the tilt attitude vector of the roadbed at the monitoring point in a two-dimensional plane. During the initial calibration stage after installation stabilization, the system does not pursue absolute verticality of the sensor or the body, but instead records the stable angle value (θ_x0, θ_y0) measured at this moment as the unique "attitude origin" of that node as the initial reference value. In subsequent monitoring, the sensor continuously outputs the current reading (θ_x, θ_y). In this way, the attitude change at any given time can be accurately obtained by calculating the difference (Δθ_x, Δθ_y) between it and the initial reference value. This fundamentally frees the system from the stringent requirement of absolute verticality during installation, achieving "on-site adaptive and self-calibrated measurement reference," greatly simplifying the on-site installation process and ensuring the consistency of the long-term measurement reference. Furthermore, this complete pair of angle changes (Δθ_x, θ_y)... Δθ_y) provides crucial two-dimensional input for the geometric kinematic model in the subsequent edge computing gateway; through model solving (e.g., ΔH ∝ [sin(Δθ_x) + sin(Δθ_y)]), the two-dimensional angle deflection information can be uniquely and faithfully mapped to a one-dimensional linear displacement (vertical settlement ΔH) along the vertical direction; dual-axis measurement not only provides more comprehensive tilt information, but its seamless integration with the subsequent mathematical model also realizes a precise and reliable conversion from "two-dimensional tilt situation" to "one-dimensional settlement result", avoiding the ambiguity or information loss caused by the inability of single-axis measurement to distinguish the tilt direction, thus ensuring the accuracy and physical authenticity of the final settlement displacement data, which is the core prerequisite for the high precision and high reliability of the entire monitoring system.

[0022] Specifically, the geometric kinematic model is: ΔH = L * [sin(θ_x - θ_x0) + sin(θ_y - θ_y0)], where ΔH represents the vertical settlement displacement, and L is the effective conversion length of the rigid body, defined as the vertical distance from the sensing center of the tilt sensor to the bottom fixed point of the rigid body.

[0023] In this embodiment, the model is based on the kinematic assumption of rigid body rotation at small angles: the rigid monitoring node installed on the roadbed is regarded as a physical pendulum model with a fixed bottom (hinged to a rigid base) and a freely deflectable top. When the roadbed settles and causes the node to tilt, the displacement of the sensor sensing center at its top in space can be decomposed into the sum of the vertical components generated by the angular deflection in two orthogonal directions, the X-axis and the Y-axis. Since the settlement displacement ΔH (i.e., the vertical displacement of the sensing center relative to its initial zero position) is geometrically exactly equal to the vector sum of these two vertical components, and the sine function "sin(Δθ)" can highly approximate the geometric relationship between the angle change and the vertical displacement components within a small angle range, the model uses the angle changes of the X and Y axes (θ_x - θ_x0 and θ_y - θ_x0) to determine the relative displacement of the sensor sensing center in space. By taking the sine of θ_y0 and summing the results, then multiplying by the key physical constant, the effective conversion length L, the precise vertical settlement displacement ΔH is obtained. This provides a clear and definite mathematical conversion path from high-precision, easily measurable angle sensing data to settlement displacement results that are directly usable in engineering, thus enabling the entire technical solution to be closed-loop. The "effective conversion length L" in the model is a structural parameter that can be determined in advance through precise mechanical design and is stable over a long period of time. This ensures that the model's solution results do not depend on volatile circuit gains or software coefficients, but are rooted in a stable physical structure, thereby guaranteeing the long-term stability, consistency, and traceability of the system's measurements. The model is simple in form and has a small computational load, making it very suitable for real-time, efficient, and low-power cyclic calculations in resource-constrained edge computing gateways. It is a key guarantee for achieving "real-time edge" output of settlement displacement data, thereby supporting subsequent real-time early warning.

[0024] Specifically, the intelligent early warning unit is configured to perform multi-indicator fusion early warning. The early warning indicators include at least: the instantaneous settlement rate output by the trend prediction unit, the cumulative settlement based on time-series data statistics, and the differential settlement between adjacent monitoring points calculated by the spatial field analysis unit. The intelligent early warning unit is also associated with a knowledge base of disposal measures, which can recommend corresponding engineering verification or disposal measures based on the early warning type, level, and location when an early warning is triggered.

[0025] In this embodiment, the unit deeply integrates the multi-dimensional outputs of the upstream analysis module: the instantaneous settlement rate reveals the severity of the current deformation and is a sensitive indicator of short-term risk; the cumulative settlement reflects the total amount of deformation and is used to determine whether the long-term deformation has approached or exceeded the structural safety threshold; the differential settlement between adjacent monitoring points characterizes the degree of uneven deformation of the roadbed, which is the most critical factor inducing pavement cracking and structural damage; by conducting parallel monitoring and fusion analysis of these three complementary indicators in terms of physical meaning and early warning timeliness, the system constructs a three-dimensional risk assessment model; when any indicator exceeds the preset multi-level threshold, the system will not only automatically trigger the corresponding early warning level (such as monitoring, early warning, alarm) according to the threshold level, but also activate the associated disposal measures knowledge base. This knowledge base incorporates expert rules developed based on engineering experience, standards, and case studies. It can automatically match and recommend the most relevant engineering verification steps (such as manual review of key areas and increased monitoring frequency) or preliminary handling suggestions (such as load control, grouting reinforcement, and drainage diversion) based on the specific warning type (e.g., excessively rapid rate, excessive total quantity, or uneven distribution), level, and precise GPS location information. This enhances the scientific rigor, accuracy, and action guidance value of the warnings. The multi-indicator fusion effectively overcomes the potential for false alarms (e.g., alarms triggered by short-term normal fluctuations) or missed alarms (e.g., slow but dangerous uneven deformation going undetected) that may occur with single-indicator warnings, making risk assessment more comprehensive and robust. The automatic recommendation of handling measures directly transforms monitoring data into actionable engineering management instructions, significantly shortening the decision-making chain from "detecting anomalies" to "initiating a response." This provides managers with clear action guidelines, upgrading the traditional, lagging, experience-based maintenance model to a forward-looking, data-driven intelligent management model, fundamentally enhancing the safety and operational efficiency of municipal road infrastructure.

[0026] This invention also discloses a settlement analysis method for municipal road subgrade, specifically including the following steps: S1. Distributed monitoring nodes are installed at the planned locations on the roadbed. After the installation is stable, the microcontroller units of each node are executed through remote commands or automatic triggering to complete the calibration and storage of the initial attitude reference values. S2, the tilt sensors of each node continuously measure at a set frequency. The edge computing gateway periodically wakes up and polls each node to obtain its current attitude reading. It calls the geometric kinematics model to calculate the vertical settlement displacement of each monitoring point in real time, and adds a timestamp and node identifier to form structured settlement time series data. S3, the edge computing gateway caches and packages the pre-processed structured settlement time series data locally, and uploads it to the cloud data analysis platform via wireless network; S4, the cloud-based data analysis platform, simultaneously performs time-series trend analysis, spatial field correlation analysis, and intelligent early warning assessment on the aggregated settlement time-series data. It processes the settlement displacement time series of each monitoring node, uses time-series prediction algorithms to fit patterns to predict future trends and calculate real-time settlement rates. Then, combining the geographic coordinates of all nodes, it uses spatial interpolation algorithms to fuse discrete settlement data to generate a continuous digital model of the roadbed settlement field, visualizes it, and calculates the global non-uniform settlement gradient field. Finally, it fuses, compares, and comprehensively assesses the real-time settlement rate, cumulative settlement, and extreme values ​​of the settlement gradient field with preset multi-level early warning thresholds. When any indicator exceeds the threshold, it automatically generates and publishes graded early warning information containing warning locations, related data, and disposal suggestions based on the threshold level.

[0027] In this embodiment, based on the collaborative architecture of "edge computing-cloud fusion", self-calibration is performed after the node installation is stable to eliminate absolute installation errors and establish a reliable relative measurement benchmark. The edge gateway uses a geometric model to calculate the original tilt angle into displacement in real time, achieving data lightweighting and real-time performance. The cloud platform simultaneously performs time series prediction and spatial field reconstruction. The former explores the evolution patterns of each point, while the latter generates a continuous gradient field reflecting the overall deformation situation by interpolating and fusing discrete data. Finally, multiple indicators such as fusion rate, cumulative amount, and gradient are used for comprehensive risk assessment, and graded early warning is automatically triggered. This transforms settlement monitoring from a traditional mode that relies on manual and single-point interpretation to an automated, multi-dimensional, and predictable intelligent process, significantly improving monitoring efficiency and the accuracy of risk identification, and realizing a shift from passive response to proactive prevention.

[0028] Specifically, in step S2, the application of the geometric kinematics model is as follows: For any monitoring node, the edge computing gateway reads its current tilt angle θ_x, θ_y and its calibrated initial tilt angle θ_x0, θ_y0, calculates the angle change Δθ_x = θ_x - θ_x0, Δθ_y = θ_y - θ_y0, and substitutes them into the model ΔH = L * [sin(Δθ_x) + sin(Δθ_y)], and outputs the vertical settlement displacement ΔH of the node in real time.

[0029] In this embodiment, the geometric kinematics model application process detailed in step S2 is the core algorithm execution link that drives the entire method to achieve real-time and accurate settlement calculation. When the edge computing gateway obtains the current dual-axis tilt angle (θ_x, θ_y) and its unique reference (θ_x0, θ_y0) of a certain monitoring node, it first calculates the pure angle change (Δθ_x, Δθ_y) through subtraction. This step essentially removes the absolute angle offset caused by the initial installation posture and extracts the relative posture change caused purely by the roadbed settlement. Subsequently, the gateway substitutes these two changes into the deterministic model ΔH = L * [sin(Δθ_x) + sin(Δθ_y)]. The physical essence of this model lies in its decomposition of the spatial displacement caused by the tilt of the node top (sensor sensing center) into vertical displacement components in two orthogonal planes, XOZ and YOZ. It utilizes the mathematical property that the sine function is highly approximated to the vertical component under small angle conditions to calculate the two components separately. Finally, the combined vertical displacement ΔH is obtained by summing and multiplying by the physical constant L (effective conversion length).

[0030] Specifically, in step S4, the time series prediction algorithm used is at least one of the following: exponential smoothing, autoregressive integral moving average model, or long short-term memory neural network model.

[0031] In this embodiment, the various time series prediction algorithms mentioned in step S4 provide the system's time series trend analysis module with a configurable intelligent toolset adapted to different monitoring scenarios and data characteristics. Mathematical models are constructed using the hidden time series dependencies (such as trends, seasonality, autocorrelation, or more complex nonlinear patterns) in historical settlement displacement data, and future changes are extrapolated from these models. The exponential smoothing method emphasizes assigning higher weights to recent data, capturing and predicting the basic trend of the data by smoothing historical observations. Its advantages lie in its simple model, high computational efficiency, and robust prediction of data with stable trends. The autoregressive integral moving average model analyzes the lag values ​​(autoregressive) of the time series itself and the historical prediction errors (sliding average). The system utilizes the statistical characteristics of average data and introduces differencing to stabilize non-stationary sequences, thereby more rigorously revealing and utilizing the inherent statistical regularities of the data for prediction. This approach is suitable for scenarios requiring quantitative analysis of the random characteristics of sequences. Meanwhile, the Long Short-Term Memory (LSTM) neural network, as a deep learning model, possesses an internal gating mechanism that makes it particularly adept at learning and memorizing complex nonlinear relationships and long-term dependency patterns in long sequences. It can automatically extract deep features from massive historical data to address complex and non-stationary dynamic changes that may occur during the settlement process. The system can flexibly select or combine the most suitable prediction tools based on the specific geological conditions of the road section, the amount of monitoring data, data quality (noise level, stationarity), and the different requirements of the prediction task for real-time performance and accuracy.

[0032] Specifically, in the spatial field correlation analysis of step S4, the spatial interpolation algorithm is the inverse distance weighting method, the kriging method, or the spline function method; the calculation of the non-uniform settlement gradient field is achieved by performing spatial differentiation on the digital settlement field model.

[0033] In this embodiment, the various spatial interpolation algorithms detailed in step S4, spatial field correlation analysis, and the subsequent gradient field calculation methods together constitute the core technical means for upsizing discrete point data into a continuous spatial field and performing refined feature extraction. Roadbed settlement is essentially a continuous spatial variable, but monitoring can only obtain data from a limited number of discrete points. Spatial interpolation algorithms are designed to resolve this contradiction. They estimate the settlement value for each unknown location point based on the "first law of geography" (things that are spatially close are more similar) or more complex spatial statistical assumptions. The inverse distance weighting method is intuitive, assuming that the value of the point to be estimated is influenced by nearby known points, and the influence weight is inversely proportional to its distance, making it computationally efficient. Kriging's rule is a more advanced geostatistical method that considers not only distance but also the analysis of known points. The spatial autocorrelation structure (variance function) between data is used to perform optimal unbiased estimation and can provide the estimation variance, quantifying the uncertainty of the interpolation results; the spline function rule uses a mathematically smooth surface (spline) to fit all known points, striving to minimize the overall curvature of the surface, thereby generating visually very smooth settlement contour lines; after generating a continuous digital settlement field model (a two-dimensional scalar field) through any of the above algorithms, the principle of calculating the non-uniform settlement gradient field turns to vector field analysis: by performing spatial differentiation on the digital model (e.g., calculating its partial derivatives in the east-west and north-south directions), the settlement change rate vector at each location can be obtained. The magnitude of this vector represents the severity of the non-uniform settlement at that point, and its direction indicates the direction of the fastest settlement change.

[0034] Specifically, in the intelligent early warning assessment in step S4, the multi-level early warning thresholds are divided into at least three levels: monitoring level, early warning level, and alarm level, which correspond to three response states: daily monitoring, risk attention, and emergency response, respectively; the early warning information release channels include prominent display on the platform interface, automatic push via SMS and email.

[0035] In this embodiment, by setting three progressive threshold levels—monitoring, early warning, and alarm—the system discretizes continuous risk metrics (such as settlement rate and cumulative amount) into explicit action instructions. The monitoring threshold aims to identify early deviations or trends requiring attention. Once triggered, it primarily serves daily monitoring and data recording, prompting management personnel to remain vigilant without immediate intervention. The early warning threshold corresponds to confirmed risk states exceeding normal fluctuations. Once triggered, the system switches to a risk monitoring mode, potentially requiring increased monitoring frequency or preparation of verification plans. The alarm threshold points to emergency situations that may endanger structural safety or operations, requiring immediate initiation of emergency response procedures upon triggering. Matching the tiered mechanism, multi-channel information dissemination (platform interface, SMS, email) adheres to the principles of reliability and timeliness in information transmission: the platform interface prominently displays the richest and most intuitive overall situation and detailed information, suitable for routine monitoring; SMS pushes utilize their high delivery rate and strong reminder characteristics to ensure that critical early warnings overcome spatial limitations and reach relevant responsible persons instantly; email pushes provide formal, archived, and complete records, facilitating post-event traceability and process management.

[0036] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0037] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A settlement monitoring system for municipal road subgrade, characterized in that: include: Multiple distributed monitoring nodes are deployed at different monitoring points on the roadbed. Each monitoring node includes a sealed rigid body with its bottom fixed to a rigid base on the roadbed, a high-precision tilt sensor fixedly installed inside the body, and a microcontroller unit. The microcontroller unit is connected to the tilt sensor and is configured to: execute an initial calibration program after the node is installed and stabilized, and record the readings measured by the tilt sensor at this time as the initial reference value of the node's attitude; and periodically collect and temporarily store the current attitude readings of the tilt sensor during subsequent monitoring. An edge computing gateway communicates with all distributed monitoring nodes. The edge computing gateway is configured to periodically obtain the current attitude reading and the corresponding initial attitude reference value from each node. Based on a preset geometric kinematic model, the change of the current attitude reading of each node relative to its initial attitude reference value is calculated in real time as the vertical settlement displacement of the node's location. A cloud-based data analytics platform, communicating with an edge computing gateway, receives and stores time-series data on vertical settlement displacement of each node. The platform has built-in analysis models, including: a trend prediction unit, used to model the time-series data on vertical settlement displacement of each node to predict future settlement trends; a spatial field analysis unit, used to construct and visualize the spatial distribution field of roadbed settlement based on the spatial location of all nodes and real-time settlement data; and an intelligent early warning unit, used to automatically identify and provide graded early warnings for abnormal settlement states according to preset rules.

2. The settlement monitoring system for municipal road subgrade according to claim 1, characterized in that: The tilt sensor is a dual-axis tilt sensor used to measure the tilt angle of a rigid body relative to the horizontal plane in two mutually orthogonal directions, the X-axis and the Y-axis. The initial attitude reference values ​​include the initial tilt angles θ_x0 and θ_y0, and the current attitude readings include the current tilt angles θ_x and θ_y.

3. The settlement monitoring system for municipal road subgrade according to claim 2, characterized in that: The geometric kinematic model is: ΔH = L * [sin(θ_x - θ_x0) + sin(θ_y - θ_y0)], where ΔH represents the vertical settlement displacement, and L is the effective conversion length of the rigid body, defined as the vertical distance from the sensing center of the tilt sensor to the bottom fixed point of the rigid body.

4. The settlement monitoring system for municipal road subgrade according to claim 1, characterized in that: The intelligent early warning unit is configured to perform multi-indicator fusion early warning. The early warning indicators include at least: the instantaneous settlement rate output by the trend prediction unit, the cumulative settlement based on time-series data statistics, and the differential settlement between adjacent monitoring points calculated by the spatial field analysis unit. The intelligent early warning unit is also associated with a knowledge base of disposal measures, which can recommend corresponding engineering verification or disposal measures based on the early warning type, level, and location when an early warning is triggered.

5. A method for settlement analysis of municipal road subgrade, characterized in that: The settlement monitoring system for municipal road subgrade according to any one of claims 1-4 specifically includes the following steps: S1. Distributed monitoring nodes are installed at the planned locations on the roadbed. After the installation is stable, the microcontroller units of each node are executed through remote commands or automatic triggering to complete the calibration and storage of the initial attitude reference values. S2, the tilt sensors of each node continuously measure at a set frequency. The edge computing gateway periodically wakes up and polls each node to obtain its current attitude reading. It calls the geometric kinematics model to calculate the vertical settlement displacement of each monitoring point in real time, and adds a timestamp and node identifier to form structured settlement time series data. S3, the edge computing gateway caches and packages the pre-processed structured settlement time series data locally, and uploads it to the cloud data analysis platform via wireless network; S4, the cloud-based data analysis platform, simultaneously performs time-series trend analysis, spatial field correlation analysis, and intelligent early warning assessment on the aggregated settlement time-series data. It processes the settlement displacement time series of each monitoring node, uses time-series prediction algorithms to fit patterns to predict future trends and calculate real-time settlement rates. Then, combining the geographic coordinates of all nodes, it uses spatial interpolation algorithms to fuse discrete settlement data to generate a continuous digital model of the roadbed settlement field, visualizes it, and calculates the global non-uniform settlement gradient field. Finally, it fuses, compares, and comprehensively assesses the real-time settlement rate, cumulative settlement, and extreme values ​​of the settlement gradient field with preset multi-level early warning thresholds. When any indicator exceeds the threshold, it automatically generates and publishes graded early warning information containing warning locations, related data, and disposal suggestions based on the threshold level.

6. The settlement analysis method for municipal road subgrade according to claim 5, characterized in that: In step S2, the application of the geometric kinematic model is as follows: For any monitoring node, the edge computing gateway reads its current tilt angle θ_x, θ_y and its calibrated initial tilt angle θ_x0, θ_y0, calculates the angle change Δθ_x = θ_x - θ_x0, Δθ_y = θ_y - θ_y0, and substitutes them into the model ΔH = L * [sin(Δθ_x) + sin(Δθ_y)], and outputs the vertical settlement displacement ΔH of the node in real time.

7. The settlement analysis method for municipal road subgrade according to claim 5, characterized in that: In step S4, the time series prediction algorithm used is at least one of the following: exponential smoothing, autoregressive integral moving average model, or long short-term memory neural network model.

8. The settlement analysis method for municipal road subgrade according to claim 5, characterized in that: In the spatial field correlation analysis of step S4, the spatial interpolation algorithm is the inverse distance weighting method, the kriging method, or the spline function method; the calculation of the non-uniform settlement gradient field is achieved by performing spatial differentiation on the digital settlement field model.

9. The settlement analysis method for municipal road subgrade according to claim 5, characterized in that: In the intelligent early warning analysis in step S4, the multi-level early warning thresholds are divided into at least three levels: monitoring level, early warning level, and alarm level, which correspond to three response states: daily monitoring, risk attention, and emergency response, respectively. Warning information is disseminated through various channels, including prominent display on the platform interface, automatic push notifications via SMS and email.

Images