Steel pipe-ecc concrete bridge pier health state monitoring method and system

Abstract

The application relates to the technical field of bridge health monitoring, and discloses a steel pipe-ECC concrete pier health state monitoring system, which comprises the following modules: a multi-source sensing module, which is responsible for monitoring interface strain, cracks and micron-level slip amounts; a data acquisition and transmission module, which is responsible for acquiring sensor data and uploading the sensor data to the cloud; and a health assessment model, which runs on the cloud, analyzes damage by fusing physical theories and measured data, and outputs pier health states. The steel pipe-ECC concrete pier health state monitoring method and system arrange strain sensors, distributed crack monitoring nets and special micro slip displacement meters in key areas of the steel pipe-ECC interface, build a multi-physical quantity coupled sensing array, can synchronously and real-timely capture micron-level interface slip and internal concrete micro crack initiation, improve early monitoring sensitivity and spatial resolution of the interface damage evolution process of the composite material, and solve the technical blind spot that the traditional monitoring technology cannot reach the interface damage origin area.

Description

Technical Field

[0001] This invention relates to the field of bridge health monitoring technology, specifically to a method and system for monitoring the health status of steel pipe-ECC concrete bridge piers. Background Technology

[0002] Steel-tube-ECC concrete composite piers are increasingly widely used in modern bridge engineering due to the excellent strain hardening characteristics and multi-crack resistance of ECC material and the good restraint effect of steel tube. The stress performance of this composite structure is highly dependent on the interfacial bond integrity between the steel tube and the ECC filling material. However, under long-term environmental erosion and cyclic loading, the interface is prone to latent early damage such as debonding and slippage. Such micro-damage is difficult to effectively detect through macroscopic observation or traditional vibration monitoring methods, but it is the source of structural performance degradation.

[0003] Current health monitoring technologies for this type of composite structure largely adopt the sensing and assessment approaches used for conventional concrete structures, which have significant technical limitations: sensors are mostly deployed in areas with significant macroscopic responses, failing to achieve continuous and targeted sensing coverage of the interface—the core damage origin region—and easily missing early micro-damage signals; the physical models and algorithms used for assessment are based on the mechanical assumptions of ordinary concrete, which are incompatible with the nonlinear characteristics of ECC strain hardening and multi-crack development, resulting in high false alarm rates and poor interpretability of damage identification results; monitoring data is singular and discrete, lacking a multi-source information fusion mechanism that can synchronously correlate interface slip, crack propagation, and material strain, and the data acquisition strategy is rigid, making it difficult to balance low-power operation with high-frequency capture of key events; existing assessment methods are mostly based on instantaneous threshold judgments, lacking dynamic tracking and forward-looking early warning capabilities for damage evolution trends. Therefore, there is an urgent need to develop a high-precision, low-power, and evolvable health monitoring technology that adapts to the interface damage mechanism of composite materials and integrates material constitutive properties and intelligent algorithms. Summary of the Invention

[0004] To address the problems mentioned in the background section, the present invention provides the following technical solution: a health status monitoring system for steel pipe-ECC concrete bridge piers, comprising:

[0005] A multi-source sensing module, designed for the heterogeneous interface characteristics of steel pipe and ECC concrete, deploys an array of multi-physical quantity sensors in the interface region and key stress-bearing parts of the bridge pier; the multi-source sensing module includes:

[0006] Strain sensors are used to sense the strain response in interface regions.

[0007] Crack gauges are used to sense the crack propagation behavior of ECC concrete.

[0008] The miniature sliding displacement meter adopts a dual-anchor direct measurement structure, including a first anchoring end rigidly fixed to the inner wall of a steel pipe, a second anchoring end pre-embedded and wrapped in ECC concrete, and a displacement sensing component connected between the two. It is used to directly obtain the micron-level relative slip in the interface normal and tangential directions, eliminating the signal hysteresis and cumulative error caused by indirect conversion through material strain difference integration or elastic elements.

[0009] The data acquisition and transmission module is communicatively connected to the multi-source sensing module and is used to synchronously acquire and preprocess multi-channel sensing signals including strain, crack width and interface slip, and upload the data to the cloud.

[0010] A health assessment model, deployed on a cloud server, is used to receive the data and run a physics-data hybrid-driven damage identification model. The model uses the ECC material nonlinear constitutive equation with damage variables as physical constraints and is embedded in the training process of an attention-enhanced temporal neural network. By fusing measured data and theoretical constitutive relations, it dynamically reconstructs the interface damage development path and outputs the graded health status assessment results of the bridge pier.

[0011] Preferably, the displacement sensing component is a linear variable differential transformer, and the two ends of its measuring rod are rigidly connected to the first anchoring end and the second anchoring end, respectively, so as to realize the direct linear measurement of the relative displacement of the interface. Compared with the indirect measurement method using a combination of spring and force sensors or the measurement structure based on strain gauges pasted on elastic elements, this system achieves 1μm resolution within a ±2mm range and has a relative measurement error ≤0.5% by eliminating the nonlinear influence of the intermediate elastic element.

[0012] Preferably, the crack gauge is a distributed fiber Bragg grating array, which is embedded in the ECC concrete surface in a cross-grid pattern with the spacing between adjacent grating points ≤10cm, forming a crack monitoring network with high spatial resolution; by monitoring local strain gradient anomalies, it can achieve millimeter-level positioning of crack initiation location and real-time tracking of crack width evolution, with a crack monitoring sensitivity of 0.02mm.

[0013] Preferably, the data acquisition and transmission module has an adaptive sampling frequency adjustment function triggered by multiple cascading events, which is implemented by a microcontroller unit:

[0014] Under normal circumstances, the first sampling frequency run;

[0015] When the rate of change of acceleration is monitored Exceeding the first preset threshold At that time, the sampling frequency of the strain and displacement channels will be automatically switched to a frequency higher than [the specified frequency]. Second sampling frequency When the rate of change of acceleration exceeds the second preset threshold simultaneously ;

[0016] And the cumulative change in slip exceeds the threshold. At that time, the third sampling frequency is triggered. The ultra-high frequency sampling mode is used, and event data packets are marked synchronously;

[0017] After the event ends, the system automatically returns to the first sampling frequency. ;

[0018] The sampling frequency switching logic is decided by the edge intelligent unit based on the real-time data stream status. Compared with the traditional fixed threshold triggering mechanism, this system reduces the false triggering rate through multi-cascaded conditions and edge computing, while extending the continuous working time in passive outdoor scenarios.

[0019] Preferably, the physical-data hybrid driven model running in the health assessment model is constructed as follows:

[0020]

[0021] As a physical constraint, among which For stress, In response, The equivalent elastic modulus function varies with strain. For continuous damage variables ( , It represents no loss. (This indicates complete failure).

[0022] This constitutive equation is introduced into the training loss function of the attention-enhanced long short-term memory network in the form of a physical consistency loss term, which is expressed as:

[0023]

[0024] in As a weighting factor, and These represent the stress and damage variable values ​​predicted by the network, respectively.

[0025] This hybrid model adaptively focuses on key features in the input sequence through a temporal attention mechanism, improving generalization ability under limited training samples, while ensuring that the output results conform to the basic mechanical laws of ECC materials. Compared with the pure data-driven model, it reduces the false alarm rate by more than 40%.

[0026] Preferably, the health assessment model also integrates a digital twin-assisted parameter optimization module: before deployment, the physical-data hybrid model is pre-trained using finite element simulation data; after deployment, the model parameters are fine-tuned online using Bayesian optimization based on real-time monitoring data streams, so as to... The trigger threshold and damage determination boundary are dynamically updated using the score as the objective function, enabling the monitoring system to adaptively evolve in complex service environments.

[0027] Preferably, the health status level output by the health assessment model is divided into at least four levels: normal, minor damage, moderate deterioration, and severe deterioration, and the level determination is dynamically corrected based on the following three indicators:

[0028] Current damage index ;

[0029] Historical data trend slope , used to characterize the rate of damage development;

[0030] Acceleration of trend change , used to characterize damage acceleration features;

[0031] When satisfied and and When any combination of conditions is met, a corresponding level of early warning is triggered, enabling multi-dimensional and proactive diagnosis from local interface hidden damage to overall performance degradation.

[0032] A method for monitoring the health status of steel pipe-ECC concrete bridge piers includes the following steps:

[0033] S1: A multi-source sensor array deployed at the interface between the steel pipe and ECC concrete and at key stress points of the pier is used to acquire real-time time-series monitoring data including strain, crack width and interface slip. The interface slip is directly measured by a miniature slip displacement meter with a double-anchored structure, eliminating the cumulative error introduced by indirect measurement.

[0034] S2: The time-series monitoring data is synchronously acquired, preprocessed, and transmitted through a data acquisition and transmission module with multi-cascade event-triggered adaptive sampling frequency adjustment function;

[0035] S3: Input the processed data into the health assessment model, which is a physics-data hybrid driven model that incorporates an attention mechanism enhanced by ECC material constitutive constraints. The damage index is calculated. And combined with historical data trend slope With acceleration It dynamically assesses and outputs the health status level of the bridge piers.

[0036] Preferably, the adaptive sampling frequency adjustment step in step S2 further includes: real-time monitoring of the rate of change of acceleration via a microcontroller unit. Cumulative change with interface slip ;

[0037] when When this occurs, the first-level event flag is triggered, and the sampling frequency is changed from... Upgraded to ;

[0038] when and When this occurs, the second-level event flag is triggered, and the sampling frequency is changed from... Upgraded to ; in And the triggering condition threshold After the incident ended and no further anomalies were confirmed, the system automatically returned to its normal low-power sampling frequency. .

[0039] The health assessment model described in step S3 is further dynamically updated through a parameter optimization process assisted by digital twins: S31: In the pre-deployment stage, the attention-enhanced LSTM network is trained offline using finite element simulation data of bridge piers;

[0040] S32: Online monitoring phase, with Using the score as the objective function, a Bayesian optimization algorithm is employed to determine the trigger threshold. Damage determination boundary and the physical consistency loss trade-off coefficient Perform periodic fine-tuning;

[0041] S33: Synchronize the optimized parameters to the edge intelligent unit and health assessment model to achieve adaptive optimization and continuous evolution of the monitoring system throughout the entire life cycle of the structure.

[0042] Compared with the prior art, the present invention provides a method and system for monitoring the health status of steel pipe-ECC concrete bridge piers, which has the following beneficial effects:

[0043] 1. The method and system for monitoring the health status of steel pipe-ECC concrete bridge piers constructs a multi-physical quantity coupled sensing array by strategically deploying strain sensors, a distributed crack monitoring network, and specially designed micro-slip displacement gauges in key areas of the steel pipe-ECC interface. This array can synchronously and in real-time capture micron-level slip at the interface and the initiation of microcracks inside the concrete, improving the early monitoring sensitivity and spatial resolution of the damage evolution process of the composite material interface and solving the technical blind spot of traditional monitoring technologies that cannot reach the origin area of ​​interface damage.

[0044] 2. The method and system for monitoring the health status of steel pipe-ECC concrete bridge piers utilizes a micro-slip displacement meter with a double-anchored mechanical structure where both ends are anchored to the steel pipe and ECC concrete respectively. It directly measures relative displacement using an LVDT, eliminating signal hysteresis and cumulative errors caused by intermediate medium transmission in traditional surface-adhesive sensors or indirect conversion sensors based on elastic elements. It achieves 1μm-level resolution within a ±2mm range and can reliably detect initial slip at the 0.02mm level, providing an unprecedentedly accurate data source for interface debonding early warning.

[0045] 3. This method and system for monitoring the health status of steel pipe-ECC concrete bridge piers uses the nonlinear constitutive equation of ECC material with damage variables as a physical constraint. This equation is embedded as a physical consistency loss term into the training process of an attention-enhanced temporal neural network, constructing a physics-data hybrid-driven damage identification model. This fusion mechanism utilizes physical laws to guide data learning, effectively limiting the model's solution space and enabling it to maintain excellent generalization performance even with limited training samples. Compared to a purely data-driven model, it reduces the false alarm rate by more than 40%, while ensuring the physical interpretability of the output results.

[0046] 4. The steel pipe-ECC concrete bridge pier health status monitoring method and system has an adaptive sampling frequency adjustment function based on multi-cascade event triggering in the data acquisition module. The edge intelligent unit dynamically decides the sampling frequency based on multiple conditions such as the rate of change of acceleration and the cumulative change of slip (normally 1Hz, which can be increased to 100Hz after event triggering). While ensuring the complete capture of key transient events, it reduces power consumption by more than 85% compared with the traditional fixed high-frequency sampling mode, which greatly extends the system's sustainable working time in passive outdoor scenarios.

[0047] 5. This steel-tube-ECC concrete bridge pier health status monitoring method and system not only outputs a continuous damage index based on current data in its health assessment model, but also innovatively integrates historical trend slope and trend change acceleration indicators of damage development. It constructs a dynamically corrected, multi-dimensional hierarchical criterion system and combines digital twin-assisted Bayesian optimization to achieve online adaptive evolution of model parameters. The four-level health status output by this system and its corresponding operation and maintenance recommendations have stronger temporal continuity, forward-looking perspective, and engineering practicality, effectively supporting bridge management units in formulating scientific preventive maintenance and emergency intervention plans. Attached Figure Description

[0048] Figure 1 A schematic diagram of the overall structure of the steel pipe-ECC concrete bridge pier health monitoring system;

[0049] Figure 2 A block diagram of the physical-data hybrid model within the health assessment model;

[0050] Figure 3 This is a time-series evolution trend chart of the interface damage index ID(t).

[0051] In the diagram: 100, Multi-source sensing module; 101, Strain sensor; 102, Crack gauge; 103, Miniature sliding displacement gauge; 200, Data acquisition and transmission module; 300, Health assessment model. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] Please see Figures 1-3 The present invention provides a technical solution:

[0054] Example 1: A Health Status Monitoring System for Steel Pipe-ECC Concrete Bridge Piers

[0055] This embodiment provides a health monitoring system for steel pipe-ECC concrete bridge piers based on multi-source sensing and physical-data hybrid modeling, aiming to solve the problem that existing technologies are unable to identify latent damage at the interface of composite materials in an early and accurate manner.

[0056] like Figure 1 As shown in the system architecture diagram, the system mainly includes a multi-source sensing module 100, a data acquisition and transmission module 200, and a health assessment model 300.

[0057] 1. Multi-source sensing module 100

[0058] The multi-source sensing module 100 is the system's perception layer, specifically designed to address the damage mechanism at the steel-concrete composite (ECC) bridge pier interface. Compared to existing steel-concrete composite bridge monitoring technologies, the core innovation of this module lies in its ability to achieve targeted sensing of multiple physical quantities at the heterogeneous steel-concrete composite (ECC) interface.

[0059] Strain sensor 101: A high-precision resistive foil strain gauge (model: BFH120-3AA, grid length 3mm, resistance 120Ω, sensitivity coefficient 2.0±1%) is attached to the inner wall of the steel pipe, especially in the stress concentration area or the initial section area of ​​ECC concrete pouring. The gauges are arranged at 20cm intervals along the pier height, and the spacing is increased to 10cm within 1.5 times the pier diameter from the bottom of the pier, to monitor the local strain response of the steel pipe wall under load.

[0060] Crack gauge 102: In this embodiment, a distributed fiber Bragg grating sensor array (FBG, center wavelength 1525-1565nm, grating length 8mm, reflectivity ≥90%) is preferably used. This array is prefabricated in the factory onto a flexible polyimide substrate, forming a sensing strip 20cm wide and with a length matching the pier height. During construction, it is directly embedded 3mm below the ECC concrete surface in a 45° intersecting grid pattern along the vertical direction of the pier, with an 8cm spacing between adjacent grating points. This arrangement forms a high-density (approximately 120 measuring points / pier), distributed crack monitoring network, capable of accurately capturing the initiation location and propagation direction of microcracks ≥0.02mm in width by identifying minute strain gradient anomalies (spatial resolution up to 3mm).

[0061] The miniature sliding displacement gauge 103 employs a dual-anchor direct measurement structure to overcome the signal hysteresis caused by deformation of the protective or adhesive layer in traditional surface-mounted sensors. The specific structure includes:

[0062] The first anchoring end is made of 304 stainless steel and is rigidly fixed to the stainless steel embedded plate pre-welded to the inner wall of the steel pipe by M6 bolts, ensuring that its displacement is completely synchronized with the steel pipe wall.

[0063] The second anchoring end is designed as a conical head with a spiral pattern (50mm in length and 12mm in maximum diameter), with a knurled surface to enhance the gripping force. It is pre-positioned before the ECC concrete is poured and is finally firmly held in the concrete matrix.

[0064] Displacement sensing component: A linear variable differential transformer (LVDT, model: GA-HC-2, range ±2mm, linearity ≤0.25%FS, resolution 1μm) is used. The two ends of its measuring rod are rigidly connected to the first anchoring end and the second anchoring end through universal joints to realize direct linear measurement of the relative displacement of the interface in the normal and tangential directions.

[0065] Compared to indirect measurement methods using a combination of spring and force sensors or measurement structures based on strain gauges bonded to elastic elements, this design eliminates the nonlinear effects of the intermediate elastic element, achieving a resolution of 1μm within a ±2mm range with a relative measurement error of ≤0.5%. It can reliably detect initial slip at the 0.02mm level, providing an unprecedentedly accurate data source for interface debonding early warning.

[0066] 2. Data Acquisition and Transmission Module 200

[0067] This module is responsible for conditioning, acquiring, and remotely transmitting the analog signals output by the multi-source sensing module 100, and has an adaptive sampling frequency adjustment function triggered by multiple cascaded events.

[0068] The module integrates a 24-bit high-precision multi-channel synchronous acquisition card (sampling rate up to 1kHz / channel) and an ARM Cortex-M4 microcontroller unit (168MHz main frequency), normally operating at the first sampling frequency. =1Hz periodic sampling of all sensor channels, daily power consumption ≤5Wh, suitable for passive outdoor scenarios.

[0069] The microcontroller continuously monitors signals from triaxial accelerometers (installed at the top of the piers and at half the pier height), while simultaneously tracking the cumulative change in slippage output by the miniature slip displacement meter 103. The sampling frequency adjustment logic is decided by the edge intelligence unit based on the real-time data stream status:

[0070] Level 1 trigger: When the rate of change of the composite acceleration is detected. Exceeding the first preset threshold for three consecutive sampling periods When a potential mechanical event is identified, the sampling frequency of the strain sensor 101 and the miniature sliding displacement gauge 103 channels is automatically adjusted. Increase to the second sampling frequency and activate the event flag;

[0071] Second-level trigger: When the rate of change of acceleration exceeds the second preset threshold simultaneously. And the cumulative change in slip Exceed When a "major mechanical event" is identified (such as severe vehicle overloading, collision, or ground vibration), the third sampling frequency is triggered. The system employs an ultra-high frequency sampling mode and initiates a high-density data recording window that lasts for at least 30 seconds, with data packets automatically labeled with event type and priority.

[0072] Event End and Resumption: When no triggering conditions are met for 30 consecutive seconds, and When the rate of change returns to the baseline level, the system automatically reverts to the low-power normal sampling frequency. .

[0073] Compared to traditional fixed threshold triggering mechanisms, this system reduces the false triggering rate by more than 85% and extends the continuous working time in the field by using multi-cascaded conditions and edge computing decisions, while ensuring the complete capture of critical transient events.

[0074] 3. Health Assessment Model 300

[0075] Deployed on a cloud server, it integrates a physical-data hybrid driven model enhanced with attention mechanism and a parameter optimization module assisted by digital twin.

[0076] 3.1 Model Architecture

[0077] The model uses a three-layer Long Short-Term Memory (LSTM) network (128 hidden units per layer) as its main framework and introduces a temporal attention mechanism. The input layer receives a multi-dimensional temporal vector extracted through feature engineering. ,in For the mean strain, The width of the crack. This represents the amount of interface sliding. This is an acceleration characteristic.

[0078] Physical constraint fusion: Based on the strain hardening and multi-crack characteristics of ECC materials, a system with continuous damage variables is established. , It represents no loss. The modified constitutive equation (representing complete failure) is used as a physical prior:

[0079]

[0080] in, The equivalent elastic modulus function, considering the nonlinear characteristics of the material, is calibrated using an ECC uniaxial tensile test as follows:

[0081]

[0082] In the formula The initial elastic modulus is taken as 35 GPa. The softening coefficient is 120.

[0083] During LSTM training, this constitutive equation is introduced into the loss function as a physical consistency loss term:

[0084]

[0085] in,

[0086]

[0087]

[0088] To balance this, a coefficient of 0.3 is used in this embodiment. This ensures that while the network learns the patterns in the data, its output must conform as closely as possible to the fundamental mechanical laws of ECC materials, thereby improving generalization ability and physical interpretability when training data (usually from limited laboratory experiments) is insufficient.

[0089] 3.2 Parameter Optimization for Digital Twin Assistance

[0090] Before deployment, a digital twin model of the bridge pier was established using ANSYS finite element simulation software to simulate the interface damage evolution process under different load conditions, generating 500 sets of simulation data for pre-training of the hybrid model.

[0091] After deployment, based on real-time monitoring data streams, Using score (with a focus on recall) as the objective function, a Bayesian optimization algorithm is employed to fine-tune key parameters online, including the trigger threshold. Damage determination boundary and the physical consistency loss trade-off coefficient The optimization cycle is 7 days, enabling the monitoring system to adaptively evolve throughout the entire lifespan of the structure.

[0092] 3.3 Health Status Grading Assessment

[0093] The model outputs a continuous damage index. The system is not simply based on instantaneous... Instead of relying on value-based grading, a dynamic assessment is conducted by combining historical trend analysis.

[0094] Level 4 health status: Classified as Level I (normal, ... Grade II (minor injury, 0.2 ≤ 0.2) <0.4), Level III (moderate degradation, 0.4≤ <0.7), Level IV (severe degradation), ≥0.7).

[0095] Multi-dimensional dynamic criteria: such as Figure 3 As shown in the time-series evolution trend chart of the damage index, it is periodically (every 24 hours) adjusted... Regression analysis of the historical sequence yielded the following results:

[0096] Trend slope This reflects the rate of damage development;

[0097] Acceleration of trend change This reflects the characteristics of accelerated damage.

[0098] The damage level is determined based on the current damage index. Trend slope and acceleration The three indicators are dynamically adjusted. For example, even if currently... The value did not reach 0.4, but if the trend slope... A sharp increase (exceeding 0.06 / day) and acceleration Even if the result is positive, the system may still issue an early warning and classify it as Level III moderate degradation. Each level is associated with specific operation and maintenance recommendations, ranging from "intensified monitoring" to "activating emergency plans," providing direct decision support for preventative maintenance.

[0099] 3.4 Brief Description of Working Principle

[0100] During the service life of the bridge piers, when micron-level slippage or microcracks initiation occur at the interface due to fatigue, overload, or other reasons, the dual-anchored micro-slip displacement gauge 103 and the distributed FBG crack gauge 102 are the first to generate a sensitive response. The data acquisition module 200 simultaneously acquires signals, capturing the entire event process at a high sampling rate of 100Hz when necessary. All data is uploaded to the health assessment model 300. The physics-data hybrid model in the engine analyzes the data and calculates the damage index by simultaneously satisfying both data regularity and physical regularity. By combining its evolution trend, the overall health level of the bridge piers can be dynamically and accurately assessed, enabling early warning and step-by-step diagnosis from local interface hidden damage to overall performance degradation.

[0101] 3.5 Experimental Verification Results

[0102] In comparative tests conducted in the laboratory on scaled-down bridge pier specimens (1:5 scale), the specimens using this system showed an average of 35% earlier identification time for interface debonding damage compared to specimens monitored solely by traditional strain gauges. In tests simulating sudden impact loads, the hybrid evaluation model of this system reduced the false alarm rate for damage by 42% compared to the pure data-driven model, improving the reliability of the monitoring results and their engineering practical value.

[0103] Example 2: System Deployment and Operation Method

[0104] This embodiment provides a method for monitoring the health status of steel pipe-ECC concrete bridge piers based on the system described in Embodiment 1, including the following steps:

[0105] S1: Sensor Deployment and Initial Calibration

[0106] Strain sensor 101 is attached to the inner wall of the steel pipe at the designed position;

[0107] The flexible sensing strip with the prefabricated FBG array is embedded in a cross-grid pattern 3mm below the surface of the ECC concrete.

[0108] Install the miniature sliding displacement meter 103: First, fix the first anchoring end to the embedded plate on the inner wall of the steel pipe. After the second anchoring end is pre-positioned, pour ECC concrete. After the concrete reaches the design strength, install the LVDT displacement sensing component and perform zero-point calibration.

[0109] Initial state data is collected to establish a baseline database.

[0110] S2: Data Acquisition and Adaptive Sampling

[0111] The data acquisition and transmission module 200 runs continuously. The microcontroller monitors the rate of change of acceleration and the cumulative change of slip in real time. It dynamically adjusts the sampling frequency according to the multi-cascade triggering logic and simultaneously collects multi-source data such as strain, crack width, and slip. After preprocessing, the data is uploaded to the cloud via 4G / 5G network.

[0112] S3: Health Assessment and Status Output

[0113] The health assessment model 300 receives a data stream, runs an attention-enhanced physical-data hybrid model, and calculates the damage index. And analyze historical trends, and synthesize current trends. Value, trend slope and acceleration Dynamically determine the health level and output an assessment report and maintenance recommendations.

[0114] S4: Online Model Optimization

[0115] The model optimization process is triggered every 7 days, using the test data from the past week, to... Using the score as the objective function, a Bayesian optimization algorithm is employed to fine-tune the trigger threshold, damage determination boundary, and physical constraint trade-off coefficients online, thereby achieving adaptive evolution of the system.

[0116] Example 3: Engineering Application Case

[0117] This system has been applied to a steel-tube-ECC concrete pier (18m high, 1.2m in diameter) in an actual bridge project. Since its deployment in March 2024, the system has successfully captured interface response data during three overloaded vehicle passage events and one minor earthquake (magnitude 3.2). During a routine inspection in August 2024, the system provided a 14-day advance warning of minor damage to the interface at the pier base. =0.28, =0.04 / day), subsequent manual inspection confirmed the presence of localized micro-detachment of approximately 0.15mm, verifying the system's early warning capability.

[0118] 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 health status monitoring system for steel pipe-ECC concrete bridge piers, characterized in that, include: The multi-source sensing module (100) is designed for the heterogeneous interface characteristics of steel pipe and ECC concrete, and deploys a multi-physical quantity sensor array in the interface area and key stress-bearing parts of the pier. The multi-source sensing module (100) includes: A strain sensor (101) is used to sense the strain response of the interface region; Crack gauge (102) for sensing crack propagation behavior in ECC concrete; The miniature sliding displacement meter (103) adopts a dual-anchor direct measurement structure, including a first anchor end rigidly fixed to the inner wall of the steel pipe, a second anchor end pre-embedded and wrapped in ECC concrete, and a displacement sensing component connected between the two, which is used to directly obtain the micron-level relative slip in the interface normal and tangential directions, eliminating the signal hysteresis and cumulative error caused by the indirect conversion through material strain difference integration or elastic element. The displacement sensing component is a linear variable differential transformer, and its measuring rod is rigidly connected to the first anchoring end and the second anchoring end respectively to realize direct linear measurement of the relative displacement of the interface. The data acquisition and transmission module (200) is communicatively connected to the multi-source sensing module (100) and is used to synchronously acquire and preprocess multi-channel sensing signals including strain, crack width, and interface slip, and upload the data to the cloud; the data acquisition and transmission module (200) has an adaptive sampling frequency adjustment function triggered by multi-cascaded events, which is implemented by a microcontroller unit: Under normal circumstances, the first sampling frequency run; When the rate of change of acceleration is monitored Exceeding the first preset threshold At that time, the sampling frequency of the strain and displacement channels will be automatically switched to a frequency higher than [the specified frequency]. Second sampling frequency When the rate of change of acceleration exceeds the second preset threshold simultaneously ; And the cumulative change in slip exceeds the threshold. When the event is triggered, the ultra-high frequency sampling mode is activated, and the event data packet is synchronously marked. The sampling frequency of the ultra-high frequency sampling mode is the third sampling frequency. ,in, and ; After the event ends, the system automatically returns to the first sampling frequency. ; Among them, the sampling frequency switching logic is decided by the edge intelligent unit based on the real-time data stream status; The health assessment model (300) is deployed on a cloud server to receive the data and run a physics-data hybrid-driven damage identification model. The model uses the nonlinear constitutive equation of ECC material with damage variables as physical constraints and is embedded in the training process of the attention mechanism-enhanced temporal neural network. It dynamically reconstructs the interface damage development path by fusing measured data and theoretical constitutive relations, and outputs the graded health status assessment results of the bridge pier. The physical-data hybrid driven model running in the health assessment model (300) is specifically constructed as follows: ； As a physical constraint, among which For stress, In response, ; The equivalent elastic modulus function varies with strain. For continuous damage variables, the value ranges from 0 to 1. The time indicates that the material is undamaged. This indicates that the material has completely failed. This constitutive equation is introduced into the training loss function of the attention-enhanced long short-term memory network in the form of a physical consistency loss term, which is expressed as: ； in As a weighting factor, and These represent the stress and damage variable values ​​predicted by the network, respectively.

2. The system according to claim 1, characterized in that, The crack gauge (102) is a distributed fiber Bragg grating array, which is embedded in the ECC concrete surface in a cross grid pattern, with the spacing between adjacent grating points ≤10cm, forming a crack monitoring network with high spatial resolution.

3. The system according to claim 1, characterized in that, The health assessment model (300) also integrates a digital twin-assisted parameter optimization module: before deployment, the physical-data hybrid model is pre-trained using finite element simulation data; after deployment, the model parameters are fine-tuned online using Bayesian optimization based on real-time monitoring data streams. The trigger threshold and damage determination boundary are dynamically updated using the score as the objective function, enabling the monitoring system to adaptively evolve in complex service environments.

4. The system according to claim 1, characterized in that, The health assessment model (300) outputs a health status level that is at least divided into four levels: normal, minor damage, moderate deterioration, and severe deterioration. The level determination is dynamically corrected based on the following three indicators: Current damage index ; Historical data trend slope , used to characterize the rate of damage development; Acceleration of trend change , used to characterize damage acceleration features; When satisfied and and When any combination of conditions is met, a corresponding level of early warning is triggered, enabling multi-dimensional and proactive diagnosis from local interface hidden damage to overall performance degradation.

5. A method for monitoring the health status of steel-tube-ECC concrete bridge piers based on the system described in any one of claims 1 to 4, characterized in that, Includes the following steps: S1: A multi-source sensor array deployed in the interface area between the steel pipe and ECC concrete and the key stress-bearing parts of the pier is used to acquire real-time time-series monitoring data including strain, crack width and interface slip; among which, the interface slip is directly measured by a miniature slip displacement meter (103) with a double anchoring structure, eliminating the cumulative error introduced by indirect measurement. S2: The time-series monitoring data is synchronously acquired, preprocessed and transmitted through the data acquisition and transmission module (200) with multi-cascade event-triggered adaptive sampling frequency adjustment function; S3: Input the processed data into the health assessment model (300), which is a physics-data hybrid driven model that integrates the attention mechanism enhanced by ECC material physical constitutive constraints, and calculates the damage index. And combined with historical data trend slope With acceleration It dynamically assesses and outputs the health status level of the bridge piers.

6. The method according to claim 5, characterized in that, The adaptive sampling frequency adjustment step in step S2 further includes: real-time monitoring of the rate of change of acceleration via a microcontroller unit. Cumulative change with interface slip ; when When this occurs, the first-level event flag is triggered, and the sampling frequency is changed from... Upgraded to ; when and When this occurs, the second-level event flag is triggered, and the sampling frequency is changed from... Upgraded to ; in And the triggering condition threshold ; After the incident ended and it was confirmed that there were no ongoing anomalies, the system automatically returned to its normal low-power sampling frequency. .

7. The method according to claim 5, characterized in that, The health assessment model (300) described in step S3 is further dynamically updated through a digital twin-assisted parameter optimization process: S31: In the pre-deployment phase, the attention-enhanced LSTM network is trained offline using finite element simulation data of bridge piers; S32: Online monitoring phase, with Using the score as the objective function, a Bayesian optimization algorithm is employed to determine the trigger threshold. Damage determination boundary and the physical consistency loss trade-off coefficient Perform periodic fine-tuning; S33: Synchronize the optimized parameters to the edge intelligent unit and health assessment model (300) to realize the adaptive optimization and continuous evolution of the monitoring system throughout the entire life cycle of the structure.

Images