Integral bridge joint connection system of steel-concrete composite beam based on equivalent stiffness matching

Abstract

This invention belongs to the technical field of structural health monitoring, and relates to a node connection system for a steel-concrete composite beam bridge based on equivalent stiffness matching. The system includes: a data acquisition module for generating a four-dimensional dynamic strain tensor; a manifold mapping module for constructing a dynamic strain manifold; a risk identification module for outputting precursory stress concentration areas and their topological locations; a command generation module for generating spatially addressed active driving commands; an execution management module for switching from an energy harvesting state to an active stiffness modulation state; a stress intervention module for generating local active reverse strain; and a closed-loop verification module for completing the entire closed-loop control process. This invention solves the problem that many monitoring systems remain at the level of data acquisition and threshold alarms, making it difficult to identify implicit precursors of stress concentration evolution and their topological locations from massive amounts of sensor data.

Description

Technical Field

[0001] This invention belongs to the technical field of structural health monitoring and relates to a joint connection system for integral steel-concrete composite beam bridges based on equivalent stiffness matching. Background Technology

[0002] Steel-concrete composite beam bridges, due to their ability to fully utilize the tensile strength of steel and the compressive strength of concrete, possess advantages such as lightweight structure, high stiffness, convenient construction, and high material utilization, and have been widely used in highway, railway, and urban bridge construction. However, steel and concrete differ in their physical and mechanical properties, especially in the joint areas. Due to the coupling of multiple factors such as abrupt stiffness changes, complex structures, repeated loads, and environmental erosion, stress concentration is prone to occur. Under long-term action, these stress concentration areas can lead to structural defects such as concrete cracking, steel fatigue damage, and shear connector failure, thus restricting the durability, safety, and service life of composite beam bridges.

[0003] To address the aforementioned issues, existing technologies have proposed passive crack resistance and structural optimization measures. These methods focus on pre-releasing or reducing adverse stresses during the design and construction phases by optimizing structural construction and connection methods. Common approaches include applying prestress to the bridge deck in the negative bending moment zone, employing the fulcrum lifting method, or using segmented construction techniques for the bridge deck. For example, Chinese invention patent CN114016392A discloses a crack control measure and design method for composite beam bridges based on finite displacement. This method involves setting finite displacement dual-control shear studs with elliptical holes and an outer layer of polymer elastomer in the negative bending moment zone at the pier top. This allows for limited sliding between the bridge deck and the steel beam under design loads, and further releases tensile stress through the deformation of the elastomer during overload or strong earthquakes.

[0004] However, existing technologies lack the ability to perceive the internal stress state and damage evolution of nodes in real time, making it impossible to achieve risk warning or proactive intervention in the early stages of damage. Their crack resistance depends entirely on pre-set displacement tolerance, and their reliability and adaptability under long-term, random loads face challenges. Most high-performance actuators require continuous external power supply, making them difficult to deploy in the field environment of bridges, and the extensive cabling increases construction and maintenance costs. Monitoring systems are mostly limited to data acquisition and threshold alarms, making it difficult to identify hidden stress concentration precursors and their topological locations from massive amounts of sensor data. Monitoring, decision-making, and execution modules are often independent of each other, lacking a closed-loop verification mechanism, making it difficult to ensure the timeliness and effectiveness of intervention actions, and even more difficult to automatically evaluate the effect and reset to a low-power monitoring state after intervention. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a steel-concrete composite beam integral bridge node connection system based on equivalent stiffness matching.

[0006] A steel-concrete composite beam integral bridge node connection system based on equivalent stiffness matching includes: The data acquisition module controls the heterogeneous distributed sensor array embedded in the steel-concrete node to collect the original strain wavelength of each measuring point, and combines it with the pre-stored topological location metadata to parse and encapsulate it to generate a four-dimensional dynamic strain tensor. The manifold mapping module receives a four-dimensional dynamic strain tensor and calls the built-in topology mapping engine to map the tensor to a Riemannian manifold to construct a dynamic strain manifold. The risk identification module receives the dynamic strain manifold and calculates the manifold curvature entropy and its rate of change over time for each local region. It compares the manifold curvature entropy with the preset singularity peak threshold to identify and output the precursory stress concentration area and its topological location. The instruction generation module receives the topological location of the precursor stress concentration area and generates a spatial addressing active drive instruction by searching a preset drive parameter lookup table. The execution management module sends a spatial addressing active drive command to the target micro-motion unit that matches the topological mirror image of the precursor stress concentration area, triggering the target micro-motion unit to switch from the energy harvesting state to the active stiffness modulation state. The stress intervention module, in the modulation state, adjusts the driving voltage signal according to the received driving parameter set and applies it to the piezoelectric material to generate local active reverse strain. The closed-loop verification module obtains the updated rate of change of manifold curvature entropy in the target region and compares it with the preset singularity peak threshold to confirm the effectiveness of the intervention. After completing the drive, the stress intervention module automatically switches back to the energy harvesting state to complete the closed-loop control process.

[0007] A further aspect of the present invention includes a data acquisition module, which is used to perform the following operations: Real-time temperature compensation is performed on the acquired raw strain wavelength data to isolate the wavelength change caused purely by mechanical strain, and the change is converted into a real-time micro-strain value through calibration coefficients. The real-time strain value calculated at each measuring point at the current sampling time is bound to its corresponding topological location metadata to form a structured data record with complete spatiotemporal labels; Based on the time dimension and the spatial three-dimensional grid defined by the topological location metadata, the structured data records of all measuring points are organized and stacked to generate a four-dimensional dynamic strain tensor, in which the four dimensions of the tensor are the time dimension and three spatial coordinate dimensions.

[0008] A further aspect of the present invention includes a manifold mapping module, used to perform the following operations: The topology mapping engine uses an interpolation algorithm based on radial basis functions to perform surface fitting on discrete data point sets mapped to Riemannian manifolds; Each data point is treated as a field source, and the weighting coefficients of its influence decaying with distance are calculated using radial basis functions to generate a continuously differentiable four-dimensional hypersurface that runs through all data points. The four-dimensional hypersurface is a dynamic strain manifold that deforms in real time as time progresses and as the data in the four-dimensional dynamic strain tensor is updated.

[0009] A further aspect of the present invention includes a risk identification module, which performs the following steps: For dynamic strain manifolds, the Gaussian curvature of each local region is calculated in sections to quantify the local bending degree of stress distribution; Based on the Gaussian curvature calculation results of continuous time series, we can understand the manifold curvature entropy of each local region and its rate of change over time. The rate of change of manifold curvature entropy is compared with a preset singularity peak threshold. If the rate of change exceeds the singularity peak threshold, the local area is locked as a precursor stress concentration area.

[0010] A further aspect of the present invention includes an instruction generation module, used to perform the following operations: Extract the topological coordinates of the precursory stress concentration region; Based on topological coordinates, the set of driving parameters containing trigger time, phase and amplitude information is determined by querying a preset driving parameter lookup table; The driving parameter set and topological coordinates are combined into a spatially addressed active driving command.

[0011] A further aspect of the present invention includes an execution management module, used to perform the following operations: The dual-mode micro-motion array laid on the steel-concrete interface controls the mechanical vibration energy caused by external loads to be converted into electrical energy through the positive piezoelectric effect and stored in the local microcapacitor of each unit. Send a spatially addressed active drive command to the target micro-motion unit corresponding to the physical location of the precursor stress concentration area; After receiving the instruction, the target micro-motion unit switches from the energy harvesting state to the active stiffness modulation state and wakes up its internal microcontroller.

[0012] A further aspect of the present invention includes a stress intervention module, used to perform the following operations: The microcontroller of the target micro-motion unit mobilizes the electrical energy stored in its local microcapacitor to start the built-in waveform generation hardware module; Based on the phase and amplitude parameters contained in the space-addressed active drive command, DC power is inverted and modulated to generate a dedicated drive voltage signal. A driving voltage signal is applied to the piezoelectric composite material of the target micro-motion unit, causing it to generate local active reverse strain that is opposite to the current stress deformation trend through the inverse piezoelectric effect. This changes the local tangential stiffness of the region in real time, forcing the concentrated stress flow to diverge towards the surrounding low-stress region.

[0013] A further aspect of the present invention provides a method for generating local active reverse strain, comprising the following steps: By utilizing the inverse piezoelectric effect, the driving voltage signal is converted into mechanical deformation; The phase of the local active reverse strain is controlled by the phase parameters in the driving parameter set, so that it cancels out the detected stress wave; By changing the local tangential stiffness of the region, the stress accumulated in the region is dispersed to the surrounding areas.

[0014] A further aspect of the present invention includes a closed-loop verification module, used to perform the following steps: In the next time step after applying local active reverse strain, the data acquisition module is repeatedly executed to the risk identification module to recalculate the rate of change of manifold curvature entropy in the target region. The rate of change of manifold curvature entropy was confirmed to have dropped below the singularity peak threshold to verify the effectiveness of stress dissipation; After the target micro-motion unit completes the driving action, if no new command arrives, it will automatically control its internal circuit switch to switch from the active driving path back to the energy harvesting path. The microcontroller of the target micro-motion unit enters a low-power sleep mode, and its local microcapacitor resumes energy collection to store energy for the next potential intervention action, thereby completing the entire closed-loop control process.

[0015] A further aspect of this invention confirms that the rate of change of manifold curvature entropy has decreased to the singularity peak threshold, comprising the following steps: In the next sampling time step after intervention, the monitoring and analysis cycle is automatically restarted by utilizing the embedded edge computing core and the heterogeneous distributed sensor array in collaboration. For the target area corresponding to the precursor stress concentration zone, the rate of change of manifold curvature entropy is recalculated based on the latest time window data; If the recalculated rate of change of manifold curvature entropy is less than the singularity peak threshold, then the stress relief effect generated by the stress intervention module is deemed effective, and the precursors to abnormal geometric mutations have disappeared.

[0016] In summary, the present invention has the following beneficial technical effects: 1. Construct a four-dimensional dynamic strain tensor to characterize the stress field inside the node, and use a topology mapping engine to generate a dynamic strain manifold. Transform discrete multi-source sensor data into a continuous and visualized geometric model. Calculate the rate of change of the local curvature entropy of the manifold to capture the instantaneous evolution of the stress field geometry from order to disorder. From monitoring the stress amplitude itself to identifying the trend of stress field evolution, the system can predict the risk area of ​​stress concentration before physical damage occurs, improving the lead time for structural damage warning.

[0017] 2. Extract the topological coordinates of the precursor stress concentration area and generate a spatial addressing active driving command to activate the target micro-motion element. The target micro-motion element uses the inverse piezoelectric effect to generate a local active reverse strain opposite to the current stress deformation trend, actively changing the local tangential stiffness of the region and forcing the stress flow to redistribute to the surrounding area. This realizes the transformation from passively bearing the load to actively intervening in the stress distribution. By real-time drainage of local stress, the formation of stress peaks is suppressed, reducing the probability of fatigue accumulation in the material due to excessive local strain.

[0018] 3. In its default state, the dual-mode micro-motion array converts the mechanical vibration energy caused by external loads into electrical energy through the positive piezoelectric effect and stores it locally. When performing active intervention, the micro-motion unit mobilizes the locally stored electrical energy to complete the drive. After the intervention action is completed, the system re-evaluates the stress state of the target area through a closed-loop verification step to confirm the intervention effect, thus reducing the dependence on external energy supply. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. The drawings are used to provide a further understanding of the present invention.

[0020] Figure 1 This discloses a schematic diagram of the framework in the embodiments of this application.

[0021] Figure 2 This discloses a flowchart of an embodiment of this application. Detailed Implementation

[0022] The following is in conjunction with the appendix Figures 1-2 A preferred description of the present invention is provided below.

[0023] See attached document Figures 1-2 This invention proposes a steel-concrete composite beam integral bridge node connection system based on equivalent stiffness matching, comprising: The data acquisition module controls the heterogeneous distributed sensor array embedded inside the steel-concrete node to collect the original strain wavelengths of each measuring point, and combines the pre-stored topological location metadata for parsing and encapsulation to generate a four-dimensional dynamic strain tensor.

[0024] A further aspect of the present invention includes a data acquisition module, which is used to perform the following operations: Real-time temperature compensation is performed on the acquired raw strain wavelength data to isolate the wavelength change caused purely by mechanical strain, and the change is converted into a real-time micro-strain value through calibration coefficients. The real-time strain value calculated at each measuring point at the current sampling time is bound to its corresponding topological location metadata to form a structured data record with complete spatiotemporal tags. The structured data records of all measuring points are organized and stacked according to the time dimension and the spatial three-dimensional grid defined by the topological location metadata to generate a four-dimensional dynamic strain tensor, in which the four dimensions of the tensor are the time dimension and three spatial coordinate dimensions.

[0025] Specifically, the embedded system or industrial control computer deployed on-site at the steel-concrete composite beam joint serves as the execution entity of this module. It first controls the heterogeneous distributed sensing array embedded within the concrete of the joint to begin operation. This array consists of two types of fiber Bragg grating sensors with different sensing densities and functions. One type is a micro-sensitive fiber cluster used for precisely capturing areas with large stress gradient changes, such as densely arranged micro-sensitive fiber clusters below the steel beam flange and around shear studs. The other type is a sparse reference fiber used to provide background reference and system calibration benchmark. A preset sampling frequency, such as 50Hz, is used. This 50Hz setting is based on the dominant frequency range of bridge vibrations caused by typical vehicle loads. According to the Nyquist sampling theorem, this frequency can effectively capture dynamic strain signals without aliasing, and is a standard setting for this type of dynamic monitoring in this field. A demodulator continuously acquires the raw data of Bragg wavelength drift caused by strain and temperature changes in each fiber Bragg grating sensor in the array.

[0026] From the locally stored configuration file, the topology location metadata for each fiber Bragg grating sensor, which was pre-stored and associated during construction and deployment, is parsed. This topology location metadata includes the three-dimensional spatial coordinates of the sensor's measuring point in the global coordinate system of the node structure, its structural partition identifier, and a unit vector representing the force direction along its principal sensing axis. For example, it includes at least three-dimensional coordinates expressed as floating-point numbers in millimeters. Its accuracy is typically within ±1mm, derived from the final layout location determined during the construction phase using surveying instruments such as total stations; a structural partition identifier of type string or integer, such as "Zone_A_Shear_Connector_Cluster"; and a three-dimensional force direction unit vector. This is used to indicate which direction the sensor is sensitive to the strain component.

[0027] The system performs real-time temperature compensation on the acquired raw wavelength drift data, extracts the wavelength change caused purely by mechanical strain, and converts it into real-time micro-strain values ​​through calibration coefficients. The system binds and encapsulates the real-time strain value calculated at each measurement point at the current sampling time with its corresponding topological position metadata data to form a structured data record with complete spatiotemporal labels.

[0028] As the time series progresses, the system organizes these structured records from all measurement points according to the time dimension and a three-dimensional spatial grid defined by metadata, ultimately generating and outputting a four-dimensional dynamic strain tensor. This four-dimensional dynamic strain tensor has a four-index data structure used to systematically store the strain field information that evolves within each node over time. Its mathematical expression is typically... Or in a higher-dimensional form to include directional information, with four dimensions being the time dimension and three spatial coordinate dimensions. Each element in the four-dimensional dynamic strain tensor contains the strain value at a specific time, at a specific spatial location, along a specific force direction, and its associated topological label.

[0029] For example, suppose a sensor from a cluster of micro-sensitive optical fibers is deployed near the interface between the lower flange of a steel beam and the concrete. The system parses the sensor's topological location metadata as three-dimensional coordinates from its configuration file. The unit is mm, the structural partition identifier is "Flange_Interface_Hotspot_01", and the force direction vector is... This indicates strain sensitive to the Z-axis direction.

[0030] At a certain sampling time The system reads the original wavelength shift of the sensor from the demodulator as follows: After using data from the collaborative temperature-sensing fiber optic cable to perform temperature compensation calculations, the equivalent wavelength shift caused by pure strain is obtained. for According to the strain sensitivity coefficient pre-calibrated by the sensor , Indicates each microstrain The corresponding Bragg wavelength offset in picometers , is a key parameter for measuring the strain sensitivity of a sensor.

[0031] The real-time strain value at this moment is obtained through calculation. The calculated strain value timestamp Binding the topology location metadata packet generates the following structured record: {time: coord: (1500.0, 200.0, 50.0), zone: "Flange_Interface_Hotspot_01", direction: (0, 0, 1), strain: 0.833}, records the time index of the final generated four-dimensional dynamic strain tensor. Spatial index corresponding coordinates The data elements at each location are processed in parallel across all sensor nodes to construct a spatially topologically encoded strain field snapshot covering the entire node region at each sampling time, which is then stacked in time series to form a four-dimensional dynamic strain tensor.

[0032] The manifold mapping module receives a four-dimensional dynamic strain tensor and calls the built-in topology mapping engine to map the tensor to a Riemannian manifold to construct the dynamic strain manifold.

[0033] A further aspect of the present invention includes a manifold mapping module, used to perform the following operations: The topology mapping engine uses an interpolation algorithm based on radial basis functions to fit the discrete data point set mapped to the Riemannian manifold to a surface. Each data point is regarded as a field source, and the weight coefficient of its influence decaying with distance is calculated using radial basis functions to generate a continuously differentiable four-dimensional hypersurface that runs through all data points. The four-dimensional hypersurface is the dynamic strain manifold, which deforms in real time as time progresses and as the data in the four-dimensional dynamic strain tensor is updated.

[0034] Specifically, an embedded edge computing core refers to an embedded processor with specific computing capabilities integrated in the node monitoring field, such as a SoC based on the ARM Cortex-A series core. It is responsible for running lightweight manifold construction algorithms. As the execution body of this module, the embedded edge computing core receives the four-dimensional dynamic strain tensor generated and transmitted by the data acquisition module through its data input interface. The edge computing core is pre-installed with and runs a dedicated topology mapping engine that implements specific geometric mapping and surface fitting algorithms.

[0035] The topology mapping engine first parses the topological location metadata bound to each data point in the four-dimensional dynamic strain tensor, especially the three-dimensional spatial coordinates. Force direction vector The topology mapping engine uses the three-dimensional coordinates of each data point. As its base position in three-dimensional Euclidean space, and its corresponding real-time strain value As an additional feature dimension representing the state of the points, the topology mapping engine uses this mapping relationship to map each discrete data point... The original spatial coordinate-strain value data space is embedded into a higher-dimensional four-dimensional Riemannian manifold space. The higher-dimensional Riemannian manifold is a four-dimensional mathematical space used to intuitively represent the spatial position-strain state relationship. Locally, it possesses Euclidean space properties, but globally it exhibits a complex curved structure. The local coordinate representation of the manifold space can be denoted as... ,in Strain value is related to physical space coordinates. As a fourth dimension, it gives manifolds an intuitive physical meaning.

[0036] To construct a smooth and continuous geometric surface, i.e., a manifold, from a discrete set of mapped points, the topology mapping engine employs an interpolation algorithm that uses radial basis functions to reconstruct the continuous surface from discrete data points. A surface fitting is performed on a set of spatial points. The interpolation algorithm treats each data point as a field source, and its influence decays with distance. The weighting coefficients are determined by solving a system of linear equations, thereby generating a continuous and differentiable four-dimensional hypersurface that runs through all data points. This hypersurface is the dynamic strain manifold, which satisfies the following formula:

[0037] in, Indicates at time The dynamic strain manifold, which is the final output of the manifold mapping module, exists as a mathematical surface over time. Continuously changing four-dimensional geometric objects serve as the data foundation for subsequent topological geometric analysis; symbols This represents the mapping relationship from spatial coordinates to strain amplitude; For point At any moment The corresponding strain amplitude coordinates; It is a linear or low-order polynomial trend term with respect to space and time, used to capture the global changing context of strain; It is the total number of discrete data points participating in the interpolation; It is the index number of the data point; It is the first The spatial coordinates of each data point; The first is obtained by interpolation. The weighting coefficients for each data point; It is a radial basis function; The Euclidean norm is used to calculate the Euclidean distance between two points. Among them, radial basis functions Gaussian function can be used , It is the bandwidth parameter that controls the smoothness of the function. The deformation occurs in real time as the data in the four-dimensional dynamic strain tensor is updated over time, and its geometric shape intuitively represents the distribution and evolution of the stress field inside the reinforced concrete joint.

[0038] For example, the specific record containing sensor data output by the edge computing core's data acquisition module, along with the records from all other sensors at the same moment, together constitute a time frame. A slice of the four-dimensional dynamic strain tensor.

[0039] The topology mapping engine extracts metadata and coordinates from the example records. mm and strain value The topology mapping engine points are mapped to a four-dimensional manifold space, with coordinates denoted as... Assuming there are two other adjacent sensor points within the same structural partition "Flange_Interface_Hotspot_01", The time-mapped coordinates are (1505.0, 200.0, 50.0, 1.200) and (1500.0, 205.0, 50.0, 0.950), respectively.

[0040] Interpolation is performed using Gaussian radial basis functions, with bandwidth parameter... The smoothness of the fitted surface is controlled, and its value must match the average spacing of the sensing points, typically ranging from 5 mm to 15 mm. The bandwidth parameter... The value of is matched to the spatial resolution of the sensor network, and its empirical formula is: ,in, The average Euclidean distance between sensing points used for manifold mapping, i.e., the spacing between all adjacent sensor pairs. The average spacing of the micro-sensitive fiber clusters deployed at the junction of the lower flange of the steel beam in this embodiment is [value missing]. Substituting into the formula, we get Rounded down to This ensures that the fitted manifold reflects details without excessive oscillation, and that the radial basis function-based interpolation smooths sensing noise while preserving key spatial gradient features of the strain field, resulting in a fitted dynamic strain manifold. To ensure physical realism, assuming the sensor spacing is approximately 10 mm, therefore, the following settings are made: The topology mapping engine calculates the distances between example points and other points, such as the distance to the first neighboring point. mm, corresponding basis function values The topology mapping engine solves for the weight coefficients by simultaneously solving the equations for all data points, including the example point and its neighbors. A continuous, smooth surface passing through all mapping points is constructed using these weights and basis functions. For any point on the surface, for example, coordinates At this point, the corresponding strain amplitude can be calculated using interpolation formulas. This fully defines the time. The dynamic strain manifold.

[0041] The risk identification module receives the dynamic strain manifold and calculates the manifold curvature entropy and its rate of change over time for each local region. It compares this entropy with the preset singularity peak threshold to identify and output the precursory stress concentration area and its topological location.

[0042] A further aspect of the present invention includes a risk identification module, which performs the following steps: For dynamic strain manifolds, the Gaussian curvature of each local region is calculated in sections to quantify the local bending degree of stress distribution; based on the Gaussian curvature calculation results of continuous time series, the manifold curvature entropy of each local region and its rate of change over time are solved; the rate of change of manifold curvature entropy is compared with a preset singularity peak threshold, and if it exceeds the singularity peak threshold, the local region is locked as a precursor stress concentration area.

[0043] Specifically, the embedded edge computing core acts as the execution entity for this step. Its internal manifold analysis module reads the dynamic strain manifold generated by the manifold mapping module. Within the 3D spatial projection region corresponding to the dynamic strain manifold, this module defines a regular 3D voxel mesh, discretizing and partitioning the manifold space. For each partition, i.e., a small local region... Local area Small volume units, typically cubes with side lengths of 5 mm to 10 mm, are obtained by spatial discretization of the dynamic strain manifold using a three-dimensional voxel mesh. This size matches the spatial resolution of the sensor network. First, for the two-dimensional topological plane of the steel-concrete interface, the four-dimensional dynamic strain manifold is dimensionality-reduced by slicing, and the two-dimensional dynamic strain scalar field corresponding to the stressed section is extracted. The unit is .

[0044] To construct dimensionlessly Euclidean geometric surfaces, the manifold mapping module uses a preset spatial mapping scale factor. The microstrain values ​​are mapped to the height of the geometric surface, in units of... Construct the equivalent geometric surface equation The continuous surface satisfies Gaussian curvature. The calculation formula is as follows:

[0045] in, These are the first fundamental form coefficients of the equivalent geometric surface, defined as follows: , , ; These are the coefficients of the second fundamental form, defined as follows: In the formula, , Representing strain fields respectively right The first-order partial derivative, , , This represents the corresponding second-order partial derivative.

[0046] The risk identification module identifies each partition. According to a preset time window, such as 10 consecutive sampling periods, with a single sampling period of 20 ms corresponding to a sampling frequency of 50 Hz, and a total time window of 0.2 s, the sequence of its Gaussian curvature changing over time is collected. Based on this sequence, the module calculates the statistical uncertainty of the Gaussian curvature distribution within this time window, i.e., the manifold curvature entropy. manifold curvature entropy An index defined based on the concept of information entropy is used to quantify the disorder or uncertainty of the distribution of Gaussian curvature values ​​in a local area over a short period of time; the larger the value, the more disordered the curvature changes.

[0047] Subsequently, the risk identification module further calculates the change in curvature entropy between adjacent time windows and divides it by the time interval to obtain the rate of change of curvature entropy over time. rate of change To capture moments of accelerated change in curvature distribution disorder, the risk identification module calculates each partition... rate of change of manifold curvature entropy With the preset singularity peak threshold Real-time comparison, singularity spike threshold This is a preset critical value used to determine whether the rate of change is abnormal; the unit is seconds (s). -1 Based on statistical analysis of a large amount of historical health data and simulated injury case data, the typical range was determined to be within 0.01 s. -1 up to 0.05 s -1 Between. Assuming the application scenario is a medium-sized highway bridge, based on data analysis under its typical vehicle load spectrum, set... .

[0048] If a certain partition Value exceeds singularity spike threshold The module will immediately partition the partition. The potential risk area is marked and its spatial topological coordinates, i.e. the index or center coordinates of the voxel mesh of that partition, are encapsulated and output as a precursor stress concentration area. A precursor stress concentration area is a set of spatial coordinates of one or more local areas whose rate of change of curvature entropy exceeds a threshold.

[0049] For local areas Suppose that it is continuous At each sampling time point, the range of values ​​in the sequence is divided into equal intervals. A range, such as The statistical sequence values ​​fall within each interval Number of items inside Calculate its probability The manifold curvature entropy within the time window Entropy The larger the value, the more chaotic the distribution of the Gaussian curvature in the region is over a short period of time, and the more disordered the evolution of the stress field is.

[0050] Let the current time window end at... The corresponding entropy value is The endpoint of the previous time window is The entropy value is The time interval between the two windows is For example, if 0.2 s, then the rate of change of manifold curvature entropy It satisfies the following formula:

[0051] in, Indicates partition The manifold curvature entropy over a time window; It is the Gaussian curvature value within the time window Classified to After the nth discrete interval, it falls on the nth The probability of each interval; This represents the rate of change of curvature entropy; and This represents the end point of two adjacent time windows; It is the length of the time window, and the rate of change is used to capture the moment when the curvature distribution disorder changes rapidly. It is a key indicator for identifying precursory stress concentration areas.

[0052] For example, the manifold analysis module reads the time corresponding to the manifold mapping module example. The module contains the dynamic strain manifold sequence of its preceding time steps, including example coordinates. A voxel mesh with a side length of 5 mm is defined in the region near mm.

[0053] Assuming partitions Including example points and their neighbors, the manifold analysis module uses the local surface formed by these points to calculate the region's... Mean Gaussian curvature at time t The manifold analysis module collects data from partitions. arrive A Gaussian curvature sequence at 10 time points, [0.001, 0.002, 0.004, 0.005, 0.007, 0.008, 0.010, 0.011, 0.011, 0.012] mm. -2 Divide the numerical range of this sequence into 5 equal intervals, and calculate the curvature entropy of the current time window by statistically analyzing the probability distribution of each interval. .

[0054] Assuming the previous window The calculated entropy value is Time window length Then the rate of change of curvature entropy This rate of change is compared with a preset singularity peak threshold. In comparison, due to partition The rate of change of curvature entropy significantly exceeds the singularity peak threshold. Therefore, the manifold analysis module will lock the partition and set the center coordinates of its voxel mesh, for example... mm output, marked as a precursor stress concentration area.

[0055] The instruction generation module receives the topological location of the precursor stress concentration area and generates a spatial addressing active drive instruction by searching a preset drive parameter lookup table.

[0056] A further aspect of the present invention includes an instruction generation module, used to perform the following operations: Extract the topological coordinates of the precursor stress concentration zone; based on the topological coordinates, determine the set of driving parameters containing trigger time, phase and amplitude information by querying a preset driving parameter lookup table; combine the driving parameter set and the topological coordinates into a spatially addressed active driving command.

[0057] Specifically, the instruction generation module first receives and parses the data of the precursor stress concentration area identified and output by the risk identification module. The data at least includes the topological coordinate information of the locked area. The instruction generation module extracts the spatial coordinates corresponding to the precursor stress concentration area from the data. The coordinates are usually represented in the form of the three-dimensional coordinates of the center point of the voxel mesh where the precursor stress concentration area is located, denoted as . The instruction generation module uses the topological coordinates of the three-dimensional location information extracted from the data of the precursory stress concentration zone to uniquely locate the risk area in physical space. It then determines the driving parameter set through a driving parameter lookup table. This driving parameter lookup table is a pre-set database mapping the coordinates of the storage space to the driving parameters. Its content is based on extensive offline finite element simulations and physical calibration experiments. The construction method of the driving parameter lookup table is as follows: Finite element simulation is used to establish a refined finite element model of the target steel-concrete joint and simulate its mechanical response under various typical, extreme and damaged conditions. Data generation simulates the sensing and manifold mapping process in the simulation, and virtually calculates the risk characteristics such as the coordinates and curvature entropy change rate of each local area under different working conditions. Parameter optimization involves trying different combinations of driving parameters (trigger time) in the simulation for each identified virtual risk area. Phase Φ, Amplitude By calculating the rate of decrease in curvature entropy change in the driven region, matching driving parameters can be found. Create a table to store a large number of risk feature-driving parameter mapping pairs in a database, namely the driving parameter lookup table. During actual operation, the system can obtain the corresponding driving parameters by performing the nearest neighbor matching query in the table based on the risk features identified in real time.

[0058] To ensure the targeted nature and effectiveness of the drive intervention, the instruction generation module accesses a preset lookup table stored in local non-volatile memory. This lookup table uses spatial coordinates or coordinate ranges as index keys and stores the associated drive parameters. The instruction generation module then uses the extracted coordinates... Perform a lookup table query to find the matching or nearest entry and read the driver parameter values ​​stored therein.

[0059] The driver parameter set contains three key elements: trigger time. The start time of instruction execution is defined, and considering system processing and executor response latency, its typical value is set to 0 to 50 ms. The instruction generation module directly reads or interpolates the required driving parameters from the lookup table based on the identified risk characteristics. The final set of driving parameters contains three core parameters: trigger time, etc. Typically, this is a delay time relative to the current moment, measured in milliseconds (ms), representing the phase parameter of the driving signal. Units are ° or rad, phase parameter The initial phase of the driving voltage signal is defined to form a matching cancellation relationship with the current stress wave, with a value ranging from 0° to 360°; the amplitude parameter of the driving signal is also defined. The unit is V, amplitude parameter The peak value of the driving voltage is defined, which is positively correlated with the predicted severity of stress concentration, typically ranging from 10 V to 100 V, with the specific value depending on the performance calibration of the actuator's piezoelectric material.

[0060] The instruction generation module will generate topological coordinates With drive parameter set The system encapsulates and encodes commands in a predefined binary or JSON format to generate a complete control instruction package with a clear spatial orientation, namely a spatial addressing active-driven instruction. As the final structured data packet, the spatial addressing active-driven instruction ensures that subsequent actions can be accurately located at the specified physical space position and performed according to the predetermined timing and intensity.

[0061] For example, the instruction generation module receives the precursory stress concentration zone data output by the risk identification module, which includes the center coordinates of the locked partition. The instruction generation module connects to the module by querying the path of the driver parameter lookup table. It searches the local lookup table using coordinates to find matching entries. Assuming the lookup result is the trigger time... Phase parameters Amplitude parameters , = V determines the peak value of the drive voltage, which is related to the magnitude of the active strain generated by the piezoelectric actuator.

[0062] The 180° phase, used as a driving signal that generates a signal opposite to the detected strain trend, forms an optimized cancellation relationship based on the anti-phase interference of strain waves. The system generates an active strain wave with a phase difference of 180°, opposite to the harmful strain wave in the stress concentration area. When the two waves superimpose, the peaks weaken each other, ultimately reducing the local stress peak and achieving stress relief. At 0° phase, the two waves have the same vibration trend, and the superposition increases the strain amplitude, exacerbating stress concentration. At 180° phase, the two waves have completely opposite vibration trends; one wave stretches while the other compresses, and the superposition cancels each other out. The 180° phase setting is not a fixed value, but rather a dynamic calculation based on the real-time detected strain wave phase, ensuring phase reversal at the moment of triggering. Trigger time The 10 ms setting defines the delay from instruction generation to the application of the drive voltage. This delay primarily compensates for the inherent processing and response time within the system, ensuring that the drive action and the identified risk signal are synchronized in the time domain. Specifically, it includes edge computing and instruction issuance delays, the execution time of manifold mapping, risk identification, and instruction generation algorithms on the embedded edge computing core (typically approximately 3-5 ms), instruction communication and parsing delays (the time it takes for spatially addressed instructions to be transmitted to the target micro-motion unit via a fieldbus, such as CAN bus, and parsed by its microcontroller, typically approximately 1-2 ms), and hardware switching and stabilization delays (the time required for the target micro-motion unit's internal circuitry to switch from energy harvesting state to active drive state, and for the drive voltage generation circuit to reach a stable operating state, typically approximately 2-4 ms).

[0063] The instruction generation module packages the coordinates and parameter set to generate a structured instruction object. For example, in JSON format, the instruction can be represented as {"address": {"x": 1502.5", "y": 202.5", "z": 50.0}, "params": {"delay_ms": 10", "phase_deg": 180", "amplitude_v": 50}}. This complete JSON string, or equivalent binary data stream, constitutes the space addressing active driving instruction sent to subsequent execution units.

[0064] The execution management module sends a spatial addressing active drive command to the target micro-motion unit that matches the topological mirror image of the precursor stress concentration area, triggering the target micro-motion unit to switch from the energy harvesting state to the active stiffness modulation state.

[0065] A further aspect of the present invention includes an execution management module, used to perform the following operations: The dual-mode micro-motion array laid on the steel-concrete interface converts the mechanical vibration energy caused by external loads into electrical energy through the positive piezoelectric effect and stores it in the local microcapacitor of each unit; it sends a spatial addressing active drive command to the target micro-motion unit corresponding to the pre-stress concentration area in physical location; after receiving the command, the target micro-motion unit switches from the energy harvesting state to the active stiffness modulation state and wakes up its internal microcontroller.

[0066] Specifically, the central controller of the dual-mode micro-motion array laid on the steel-concrete interface serves as the execution body of this module. The dual-mode micro-motion array refers to an actuator network composed of multiple independent units with two working modes, such as energy harvesting and active drive. Its deployment position is strictly mirrored with the sensor array topology in the data acquisition module, and it manages the working modes of all target micro-motion units.

[0067] By default, the central controller keeps all target micro-motion units in the energy harvesting state, which is the default operating mode of the target micro-motion units. In this mode, the target micro-motion units convert environmental mechanical vibration energy into electrical energy through the positive piezoelectric effect. The efficiency range of the energy management circuit is determined based on the performance of the target micro-motion unit components. The piezoelectric material used in the target micro-motion units, such as PZT-5H, has an electromechanical conversion efficiency of about 40%-70% in typical bridge vibration environments, such as accelerations of 0.1 to 0.5 g and frequencies of 2 to 10 Hz. The subsequent full-bridge rectifier circuit will suffer losses of about 10%-20% due to the diode forward voltage drop. The static power consumption and conversion efficiency of the low dropout linear regulator (LDO) are about 85%-95%.

[0068] Taking all the above factors into account, the overall conversion efficiency of the entire energy harvesting pathway from mechanical vibration energy to stable electrical energy stored in microcapacitors is empirically between 30% and 60%. Under these conditions, the positive piezoelectric material layer inside each target micro-motion unit deforms due to the small mechanical vibration of the steel-concrete interface caused by external loads on the bridge, such as vehicle traffic, generating alternating induced charges. The micro-energy management circuit built into the target micro-motion unit is usually composed of a full-bridge rectifier and a low-dropout linear regulator, which converts the alternating charges into stable direct current and stores them in the unit's local surface-mount tantalum capacitor or supercapacitor in a trickle-charging manner. The local microcapacitor is an independent energy storage element inside each target micro-motion unit, and its voltage range is usually designed between 0V and 5V. The capacity selection needs to balance the stored energy and the unit volume. The typical capacity is 100 µF to 1000 µF, and a typical value of 470 µF can store about 2.1 mJ of energy at 3 V.

[0069] When the central controller receives the spatial addressing active drive command generated and transmitted by the command generation module, it parses the topological coordinates contained in the "address" field of the command packet. The controller searches for one or more target micro-motion units that correspond to the topological coordinates in physical space based on a pre-stored target micro-motion unit physical location mapping table. The controller then sends the set of drive parameters contained in the "params" field of the instruction packet to the target micro-motion unit via a local communication bus, such as based on RS-485 or CAN protocol. And mode switching commands.

[0070] Upon receiving an instruction, the interface circuit of the target micro-motion unit wakes its internal microcontroller from sleep mode. The microcontroller, a low-power processor integrated within each target micro-motion unit, is responsible for instruction parsing, timing control, and state switching. The microcontroller first parses the instruction and then... The internal timer is set with parameters, and then the solid-state relay or MOSFET switch in the control unit switches the circuit from the energy harvesting path to the active drive path. After the circuit switch is completed, the target micro-motion unit officially enters the active stiffness modulation state. The active stiffness modulation state is the trigger working mode of the target micro-motion unit. In this mode, the unit is prepared to use the stored electrical energy to generate active mechanical strain through the inverse piezoelectric effect. Its microcontroller remains active, waiting to execute the next drive action, and preparing the electrical energy stored in the local microcapacitor to be used for drive.

[0071] Among them, topological mirror matching refers to the physical installation position of each target micro-motion unit in the actuator array corresponding one-to-one with the center position of a specific high-density sensor cluster in the sensor array in three-dimensional space. The positional deviation is usually controlled within ±2mm, which is the basis for realizing spatial addressing intervention.

[0072] For example, the central controller receives a spatially addressed active drive instruction generated by the instruction generation module, the address of which is... mm, parameter is V.

[0073] The controller queries the mapping table and finds the target micro-motion unit with the unique address identifier DMU_ZoneA_07. The controller then sends a command frame containing parameters to the unit DMU_ZoneA_07 via the CAN bus. At this time, the unit DMU_ZoneA_07 is in the default energy harvesting state, and its local 470 µF tantalum capacitor has been charged by harvesting vibration energy. After receiving the command, the CAN transceiver of the V unit triggers an interrupt, waking up the microcontroller from its sleep state. The microcontroller parses the command and, upon learning that it needs to act in 10 ms, starts a 10 ms hardware timer and simultaneously sends a high-level signal. This controls the MOSFET switch to switch the circuit from the energy harvesting terminal connected to the rectifier to the driving terminal connected to the power amplifier. After the timer countdown ends, the microcontroller confirms that the unit has stabilized in active stiffness modulation mode, and the energy in its local capacitor is approximately... This is used to subsequently drive and wait for the specific drive commands of the stress intervention module to be executed.

[0074] The stress intervention module, whose control input is connected to the execution management module, adjusts the driving voltage signal according to the received driving parameter set in the modulation state and applies it to the piezoelectric material to generate local active reverse strain.

[0075] A further aspect of the present invention includes a stress intervention module, used to perform the following operations: The microcontroller of the target micro-motion unit mobilizes the electrical energy stored in its local microcapacitor to start the built-in waveform generation hardware module; according to the phase and amplitude parameters contained in the spatial addressing active drive command, it inverts and modulates the DC power to generate a dedicated drive voltage signal; the drive voltage signal is applied to the piezoelectric composite material of the target micro-motion unit, so that it generates a local active reverse strain opposite to the current stress deformation trend through the inverse piezoelectric effect, thereby changing the local tangential stiffness of the region in real time and forcing the concentrated stress flow to diverge to the surrounding low stress area.

[0076] A further aspect of the present invention provides a method for generating local active reverse strain, comprising the following steps: By utilizing the inverse piezoelectric effect, the driving voltage signal is converted into mechanical deformation; the phase of the local active reverse strain is controlled by the phase parameters in the driving parameter set, so that it cancels out the detected stress wave; by changing the local tangential stiffness of the region, the stress flow accumulated in the region is dispersed to the surrounding region.

[0077] Specifically, after the target micro-motion unit enters the active stiffness modulation state, its internally integrated low-power microcontroller acts as the execution body of this module. The controller first reads the set of driving parameters from the received instructions. Waiting time Upon arrival, the digital-to-analog converter (DAC) or high-frequency pulse-width modulation (PWM) module inside the target micro-motion unit is immediately activated. Subsequently, the microcontroller starts the built-in DC-DC high-frequency boost converter circuit to boost the released transient drive voltage peak to the 50V drive voltage matched by command parameter A. The high-lumen pulse or anti-phase alternating signal generated by the drive module is applied to the multilayer stacked piezoelectric ceramic integrated within the unit.

[0078] The stress intervention module uses the DC power stored in the local microcapacitor as the power rail, according to the phase parameters in the instruction. and amplitude parameters An AC drive waveform is synthesized. The local microcapacitor is an energy storage element integrated on the circuit board of each target micro-motion unit. It has been charged by collecting environmental vibration energy in the preceding module, providing explosive energy support for instantaneous high-power drive without the need for external power supply cables.

[0079] The waveform signal, after passing through a miniature power amplifier circuit, is applied to the two end electrodes of the core component of the target micro-motion unit, namely the piezoelectric ceramic / polymer composite material layer. Based on the inverse piezoelectric effect, which refers to the physical phenomenon that a piezoelectric material undergoes mechanical deformation due to an external electric field when not subjected to external mechanical force, it is the core mechanism for achieving electromechanical conversion. The material layer undergoes mechanical deformation under the influence of the electric field, due to the phase of the driving signal... It is set to have a specific reverse relationship with the detected structural strain wave, such as 180° phase reversal. The deformation direction generated by the piezoelectric material is exactly opposite to the current mechanical deformation trend of the local area of ​​the steel-concrete node. This artificially introduced small reverse deformation is the local active reverse strain. The local active reverse strain is the physical manifestation of the intervention action. It is a micron-level mechanical displacement field. Its purpose is to generate counterforce at the stress concentration point, rather than change the macroscopic structural configuration.

[0080] From a mechanical perspective, the introduction of this reverse strain is equivalent to instantaneously adjusting the local tangential stiffness of the material at a local micro-element, breaking the original stress concentration equilibrium state. This forces the stress flow that originally converged towards the singularity to be redistributed and dispersed to the surrounding areas with relatively low stiffness or low stress levels under the action of the stiffness gradient, thereby achieving active guidance of stress concentration.

[0081] Among them, the driving voltage signal is an analog AC signal, whose frequency is usually consistent with the dominant frequency of the local vibration of the structure, and whose waveform is usually a modified sine wave or pulse wave.

[0082] Local tangential stiffness refers to the ratio of stress increment to strain increment in a material under a specific stress state. By introducing active strain, this ratio can be dynamically modified to achieve equivalent stiffness matching.

[0083] For example, the microcontroller of the target micro-motion unit "DMU_ZoneA_07" is in After the countdown ends, the data stored in the local capacitor will be used. V electrical energy, after being boosted by the voltage boosting circuit, has a peak output of [value missing]. The inverted sinusoidal signal, assuming the piezoelectric stack contains piezoelectric thin films, their monolayer piezoelectric strain constant The typical value is 500 pm / V, in transient... The axial mechanical displacement deformation generated by the piezoelectric stack under the driving voltage. ,Although While the displacement is small in the macroscopic structure, in the early stages of microstructural damage initiation, such as at the tip of a microcrack or at an interface microdefect, this micrometer-scale local active displacement instantaneously alters the microscopic stress intensity factor at the defect tip. This is equivalent to instantaneously introducing a reverse yield force at the local micro-element, blunting the originally rapidly rising local microscopic stress peak at the singularity, and forcing the stress concentration at the microcrack tip to diffuse towards the surrounding undamaged matrix, thereby suppressing the initial initiation of fatigue cracking at the microscopic level.

[0084] The closed-loop verification module obtains the updated rate of change of manifold curvature entropy in the target region and compares it with the preset singularity peak threshold to confirm the effectiveness of the intervention. After completing the drive, the stress intervention module automatically switches back to the energy harvesting state to complete the closed-loop control process.

[0085] A further aspect of the present invention includes a closed-loop verification module, used to perform the following steps: In the next time step after applying local active reverse strain, the data acquisition module is repeatedly executed to the risk identification module to recalculate the rate of change of manifold curvature entropy in the target region; it is confirmed that the rate of change of manifold curvature entropy has dropped below the singularity peak threshold to verify the effectiveness of stress relief; after the target micro-motion unit completes the driving action, if no new instructions arrive, it automatically controls its internal circuit switch to switch from the active driving path back to the energy harvesting path; the microcontroller of the target micro-motion unit enters a low-power sleep mode, and its local microcapacitor restarts energy harvesting to store energy for the next potential intervention action, thereby completing the entire closed-loop control process.

[0086] A further aspect of this invention confirms that the rate of change of manifold curvature entropy has decreased to the singularity peak threshold, comprising the following steps: In the next sampling time step after intervention, the monitoring and analysis cycle is automatically restarted by utilizing the embedded edge computing core and the heterogeneous distributed sensing array in collaboration. For the target area corresponding to the precursory stress concentration zone, the rate of change of manifold curvature entropy is recalculated based on the latest time window data. If the recalculated rate of change of manifold curvature entropy is less than the singularity peak threshold... If the stress relief effect generated by the stress intervention module is effective, then the precursor to abnormal geometric mutation has disappeared.

[0087] Specifically, after the target micro-motion unit in the stress intervention module completes the local active reverse strain application action, after a preset system clock cycle, for example, the next 20 ms sampling time step... The embedded edge computing core works in conjunction with the sensor array to automatically restart the miniaturized monitoring-analysis cycle. The system clock cycle or the next sampling time step is the basic time unit for the system to perform periodic data acquisition and processing, which corresponds to the sampling frequency of 50 Hz defined in the data acquisition module, i.e., 20 ms.

[0088] The sensor array acquires data in the same configuration as the data acquisition module. The system generates a new round of original strain wavelength data for the time node region and regenerates the four-dimensional dynamic strain tensor for that time. The edge computing kernel, like the manifold mapping module, maps the new four-dimensional dynamic strain tensor onto the virtual geometric space, constructing... The dynamic strain manifold updated at each moment.

[0089] The manifold analysis module within the computational kernel focuses on the local spatial region corresponding to the previously identified precursory stress concentration area. Using the same algorithm as the risk identification module, it recalculates the target region within the specified area. Rate of change of manifold curvature entropy within the latest time window at time t_t After the calculation is completed, the manifold analysis module will use the newly calculated results. Singularity spike threshold defined in the value and risk identification module and remaining constant throughout the control process Compare them, if the comparison result is If the manifold analysis module determines that the stress relief effect produced by the intervention action of the stress intervention module is effective in the target area, the previously detected abnormal geometric change precursors have disappeared.

[0090] After the drive voltage signal is output, the microcontroller of the target micro-motion unit continuously monitors its local status. When it confirms that the drive action is completed and the duration reaches the preset length, such as 20 ms, if no new instructions arrive, it automatically controls the internal circuit switch to switch the working mode from the active stiffness modulation state back to the default energy harvesting state.

[0091] The internal clock and state machine of the target micro-motion unit are reset, and its microcontroller re-enters the low-power sleep mode. Its local microcapacitor restarts to collect environmental vibration energy through the positive piezoelectric effect, thereby completing energy storage and state preparation for the next potential intervention action. This forms a complete and cyclical technology chain consisting of all modules from anomaly perception, decision-making and instruction generation, execution preparation, active intervention to effect verification and system reset, realizing a self-consistent operation and complete closed-loop control process from prediction to intervention to verification.

[0092] The criterion for validating the effectiveness is that the rate of change of curvature entropy in the target region after intervention falls back to the singularity peak threshold. The following characterizes the evolution of the local strain field from accelerated disorder to relative stability.

[0093] The execution of the data acquisition module to the risk identification module aims to automatically trigger the entire process from data collection to the calculation of the rate of change of manifold curvature entropy within the first sampling period after intervention.

[0094] Automatic switching back to the energy harvesting state is a built-in logic in the design of the target micro-motion unit, guaranteed by the timer and state machine hardware logic within the unit's microcontroller. This eliminates the need for the central controller to send additional reset commands, reducing communication overhead and improving reliability.

[0095] For example, in the stress intervention module example, the target micro-motion unit "DMU_ZoneA_07" is in Approximately 10 ms after the initial sampling time, a driving strain with an amplitude of 3.5V and a phase of 180° was applied, and the system waited until the next sampling time. ,exist The sensor array reacquires data at all times and generates new tensors.

[0096] Edge computing kernels construct manifolds for new time moments, specifically for coordinate systems. For the target region centered at mm, calculate its latest rate of change of curvature entropy, assuming based on to The new time window data was used to calculate... Compare this value with the invariant singularity spike threshold. In comparison, due to The intervention was deemed effective, and the risk of stress concentration has been mitigated.

[0097] Meanwhile, after the target micro-motion unit "DMU_ZoneA_07" finishes its driving action for 20ms, the microcontroller automatically switches the circuit back to the energy harvesting path, clears the unit status flag, and enters sleep mode. The voltage of its local microcapacitor may drop from 2.8V to 2.5V due to driving consumption. Then, it begins to slowly charge and recover under the energy harvesting state due to environmental vibration. This completes the closed-loop operation of monitoring-prediction-intervention-verification for a specific risk area and restores to the low-power monitoring state, ready to deal with the next possible risk.

[0098] Each of the modules can be implemented in whole or in part through software, hardware, or a combination thereof. It supports hardware embedded in or independent of the processor in the computer device, and also supports software stored in the memory of the computer device, so that the processor can call and execute the operations corresponding to each of the above modules.

[0099] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A steel-concrete composite beam integral bridge node connection system based on equivalent stiffness matching, characterized in that, include: The data acquisition module controls the heterogeneous distributed sensor array embedded in the steel-concrete node to collect the original strain wavelength of each measuring point, and combines it with the pre-stored topological location metadata to parse and encapsulate it to generate a four-dimensional dynamic strain tensor. The manifold mapping module receives a four-dimensional dynamic strain tensor and calls the built-in topology mapping engine to map the tensor to a Riemannian manifold to construct a dynamic strain manifold. The risk identification module receives the dynamic strain manifold and calculates the manifold curvature entropy and its rate of change over time for each local region. It compares the manifold curvature entropy with the preset singularity peak threshold to identify and output the precursory stress concentration area and its topological location. The instruction generation module receives the topological location of the precursor stress concentration area and generates a spatial addressing active drive instruction by searching a preset drive parameter lookup table. The execution management module sends a spatial addressing active drive command to the target micro-motion unit that matches the topological mirror image of the precursor stress concentration area, triggering the target micro-motion unit to switch from the energy harvesting state to the active stiffness modulation state. The stress intervention module, in the modulation state, adjusts the driving voltage signal according to the received driving parameter set and applies it to the piezoelectric material to generate local active reverse strain. The closed-loop verification module obtains the updated rate of change of manifold curvature entropy in the target region and compares it with the preset singularity peak threshold to confirm the effectiveness of the intervention. After completing the drive, the stress intervention module automatically switches back to the energy harvesting state to complete the closed-loop control process.

2. The steel-concrete composite beam integral bridge node connection system based on equivalent stiffness matching according to claim 1, characterized in that, The data acquisition module is used to perform the following operations: Real-time temperature compensation is performed on the acquired raw strain wavelength data to isolate the wavelength change caused purely by mechanical strain, and the change is converted into a real-time micro-strain value through calibration coefficients. The real-time strain value calculated at each measuring point at the current sampling time is bound to its corresponding topological location metadata to form a structured data record with complete spatiotemporal labels; Based on the time dimension and the spatial three-dimensional grid defined by the topological location metadata, the structured data records of all measuring points are organized and stacked to generate a four-dimensional dynamic strain tensor, in which the four dimensions of the tensor are the time dimension and three spatial coordinate dimensions.

3. The steel-concrete composite beam integral bridge node connection system based on equivalent stiffness matching according to claim 1, characterized in that, The manifold mapping module is used to perform the following operations: The topology mapping engine uses an interpolation algorithm based on radial basis functions to perform surface fitting on discrete data point sets mapped to Riemannian manifolds; Each data point is treated as a field source, and the weighting coefficient of its influence decaying with distance is calculated using radial basis functions to generate a continuous and differentiable four-dimensional hypersurface that runs through all data points. The four-dimensional hypersurface is a dynamic strain manifold that deforms in real time as time progresses and as the data in the four-dimensional dynamic strain tensor is updated.

4. The steel-concrete composite beam integral bridge node connection system based on equivalent stiffness matching according to claim 1, characterized in that, The risk identification module is used to perform the following steps: For dynamic strain manifolds, the Gaussian curvature of each local region is calculated in sections to quantify the local bending degree of stress distribution; Based on the Gaussian curvature calculation results of continuous time series, the manifold curvature entropy of each local region and its rate of change with time are solved; The rate of change of manifold curvature entropy is compared with a preset singularity peak threshold. If the rate of change exceeds the singularity peak threshold, the local area is locked as a precursor stress concentration area.

5. The steel-concrete composite beam integral bridge node connection system based on equivalent stiffness matching according to claim 1, characterized in that, The instruction generation module is used to perform the following steps: Extract the topological coordinates of the precursory stress concentration region; Based on topological coordinates, the set of driving parameters containing trigger time, phase and amplitude information is determined by querying a preset driving parameter lookup table; The driving parameter set and topological coordinates are combined into a spatially addressed active driving command.

6. The integral bridge node connection system based on equivalent stiffness matching of steel-concrete composite beams according to claim 1, characterized in that, The execution management module is used to perform the following operations: The dual-mode micro-motion array laid on the steel-concrete interface controls the mechanical vibration energy caused by external loads to be converted into electrical energy through the positive piezoelectric effect and stored in the local microcapacitor of each unit. Send a spatially addressed active drive command to the target micro-motion unit corresponding to the physical location of the precursor stress concentration area; After receiving the instruction, the target micro-motion unit switches from the energy harvesting state to the active stiffness modulation state and wakes up its internal microcontroller.

7. The steel-concrete composite beam integral bridge node connection system based on equivalent stiffness matching according to claim 1, characterized in that, The stress intervention module is used to perform the following steps: The microcontroller of the target micro-motion unit mobilizes the electrical energy stored in its local microcapacitor to start the built-in waveform generation hardware module; Based on the phase and amplitude parameters contained in the space-addressed active drive command, DC power is inverted and modulated to generate a dedicated drive voltage signal. A driving voltage signal is applied to the piezoelectric composite material of the target micro-motion unit, causing it to generate local active reverse strain that is opposite to the current stress deformation trend through the inverse piezoelectric effect. This changes the local tangential stiffness of the region in real time, forcing the concentrated stress flow to diverge towards the surrounding low-stress region.

8. The integral bridge node connection system based on equivalent stiffness matching of steel-concrete composite beams according to claim 7, characterized in that, A method for generating local active reverse strain includes the following steps: By utilizing the inverse piezoelectric effect, the driving voltage signal is converted into mechanical deformation; The phase of the local active reverse strain is controlled by the phase parameters in the driving parameter set, so that it cancels out the detected stress wave; By changing the local tangential stiffness of the region, the stress accumulated in the region is dispersed to the surrounding areas.

9. The integral bridge node connection system based on equivalent stiffness matching of steel-concrete composite beams according to claim 1, characterized in that, The closed-loop verification module is used to perform the following steps: In the next time step after applying local active reverse strain, the data acquisition module is repeatedly executed to the risk identification module to recalculate the rate of change of manifold curvature entropy in the target region. The rate of change of manifold curvature entropy was confirmed to have dropped below the singularity peak threshold to verify the effectiveness of stress dissipation; After the target micro-motion unit completes the driving action, if no new command arrives, it will automatically control its internal circuit switch to switch from the active driving path back to the energy harvesting path. The microcontroller of the target micro-motion unit enters a low-power sleep mode, and its local microcapacitor resumes energy collection to store energy for the next potential intervention action, thus completing the entire closed-loop control process.

10. The integral bridge node connection system based on equivalent stiffness matching of steel-concrete composite beams according to claim 9, characterized in that, Confirm that the rate of change of manifold curvature entropy has dropped to the singularity peak threshold, including the following steps: In the next sampling time step after intervention, the monitoring and analysis cycle is automatically restarted by utilizing the embedded edge computing core and the heterogeneous distributed sensor array in collaboration. For the target area corresponding to the precursor stress concentration zone, the rate of change of manifold curvature entropy is recalculated based on the latest time window data; If the recalculated rate of change of manifold curvature entropy is less than the singularity peak threshold, then the stress relief effect generated by the stress intervention module is deemed effective, and the precursors to abnormal geometric mutations have disappeared.

Images