A wharf structure multi-source data processing method and system based on digital twinning

Abstract

The application provides a wharf structure multi-source data processing method and system based on digital twinning, applied to the technical field of digital twinning, through acquiring multi-source data of the wharf structure, and time stamp alignment of the multi-source data; through constructing a three-dimensional digital model, and establishing a spatial mapping relationship between data points in the multi-source sensing data sequence and structure units in the three-dimensional digital model; through calculating the internal stress state of the steel bundle, the fine perception of the internal stress state of the structure is realized; through evaluating the fatigue damage accumulation degree of the steel bundle according to the internal stress state, the internal stress state and the fatigue damage accumulation degree are rendered to the three-dimensional digital model for display, and a warning signal is generated when the fatigue damage accumulation degree exceeds a preset threshold, therefore, the application has the advantages of improving the accuracy and early warning capability of wharf structure health monitoring.

Description

Technical Field

[0001] This application relates to the field of digital twin technology, and in particular to a method and system for processing multi-source data of a dock structure based on digital twins. Background Technology

[0002] Real-time monitoring systems are crucial for the maintenance and safety management of large infrastructure such as docks. These systems collect data through various sensors, including satellite positioning, fiber optics, acoustic waves, and underwater imaging, to construct a digital twin that changes synchronously with the physical structure, providing a comprehensive reflection of its health. However, existing technologies have a significant problem: after a program update, the temperature compensation settings of the time server may fail, causing nonlinear drift in the time signal. This drift fluctuates complexly with temperature changes, resulting in a cumulative, difficult-to-correct, deviation in the local clocks of each sensor when edge computing nodes receive calibration signals.

[0003] This bias affects the temporal correlation of data. For example, the system may presuppose that a physical event generates multiple signals within a short period, but timestamp discrepancies cause signal misalignment, exceeding the associated time window and rendering the data correlation logic ineffective. During preprocessing, the cloud platform's automated program, unable to find mutually corroborating signals within the preset time window, may misjudge and filter out valid low-amplitude event data as noise. This could lead system administrators to raise the threshold for data change, further exacerbating the blind spot in perceiving subtle structural changes.

[0004] Because low-amplitude event data is filtered out, the subtle fatigue accumulation of the wharf structure under normal operating conditions cannot trigger digital twin status updates. For example, the minute stresses and acoustic emission signals generated by ship berthing are misinterpreted as noise, causing the digital twin to fail to map the structural fatigue accumulation process in real time, resulting in a deviation from the actual physical structural state. This deviation leads to the omission of latent fatigue damage in the steel reinforcement bundles, whose local anchorage areas gradually deteriorate under long-term micro-vibrations and stress cycles. Furthermore, the monitoring system, due to the failure of time information alignment logic and automated data filtering, cannot provide accurate structural health assessments.

[0005] Ultimately, when the wharf is subjected to moderate winds and waves, the steel reinforcement bundles already suffering from fatigue damage may partially yield or loosen their anchorage. However, due to time and location alignment issues and lagging update mechanisms in its data processing framework, the digital twin system cannot promptly identify the decline in the load-bearing capacity of the steel reinforcement bundles, delaying repair decisions and exposing the wharf to structural safety risks. Therefore, existing technologies urgently need improvement to ensure the accuracy and reliability of the monitoring system. Summary of the Invention

[0006] In view of the shortcomings of the prior art, this application provides a method and system for processing multi-source data of wharf structures based on digital twins. It has the advantages of solving the problems of inaccurate timestamp alignment of multi-source data and the inability of digital twin models to reflect the accumulation of subtle fatigue in the structure in real time, and improving the accuracy and early warning capability of wharf structure health monitoring.

[0007] Firstly, a method for processing multi-source data on a port structure based on digital twins, the method comprising the following steps: S1: Acquire macroscopic spatial displacement data, internal strain distribution data of concrete, and acoustic emission data of microscopic damage of the wharf structure; S2: Time-stamp align the macroscopic spatial displacement data, the internal strain distribution data of concrete, and the acoustic emission data of microscopic damage to generate a time-aligned multi-source sensing data sequence; S3: Construct a three-dimensional digital model containing the geometric parameters and material mechanical properties of the concrete structure and internal steel reinforcement bundles, and establish a spatial mapping relationship between the data points in the multi-source sensing data sequence and the structural units in the three-dimensional digital model; S4: Based on the strain distribution data inside the concrete, the stress field of the concrete region is inverted, and combined with the stress field and the macroscopic spatial displacement data, the internal stress state of the steel reinforcement bundle is calculated. S5: Evaluate the degree of fatigue damage accumulation of the steel bar bundle based on the internal stress state, render the internal stress state and the degree of fatigue damage accumulation to the three-dimensional digital model for display, and generate an early warning signal when the degree of fatigue damage accumulation exceeds a preset threshold.

[0008] Furthermore, step S2 includes: S21: Receive external timing signals at the edge computing node and fine-tune the local clock frequency based on local temperature data using an adaptive filtering algorithm; S22: Use the fine-tuned local clock to timestamp-align the macroscopic spatial displacement data, the concrete internal strain distribution data, and the microscopic damage acoustic emission data to generate a time-aligned multi-source induction data sequence.

[0009] Furthermore, step S21 includes: S211: Establish a clock error state equation, wherein the state equation takes clock phase deviation and frequency drift as state variables, and the local temperature data as a control input variable affecting frequency drift; S212: Based on the state estimate of the previous moment and the current local temperature data, predict the clock error state at the current moment; S213: When the external timing signal is received, calculate the observation residual between the external timing signal and the predicted value, use the observation residual to correct the clock error state at the current time, and obtain the optimal clock frequency drift estimate. S214: Dynamically compensate the counting frequency of the local clock based on the optimal clock frequency drift estimate to generate the fine-tuned local clock.

[0010] Furthermore, step S214 includes: S2141: Real-time monitoring of the instantaneous power consumption of high-power components inside the edge computing node, and acquisition of the local hot spot temperature near the high-power components; S2142: Calculate the internal disturbance compensation amount based on the instantaneous power consumption and the local hot spot temperature; S2143: The optimal clock frequency drift estimate is used as the external offset compensation amount, and the external offset compensation amount is superimposed with the internal disturbance compensation amount to calculate the final frequency adjustment command. S2144: The counting frequency of the local clock is adjusted using the frequency adjustment command to generate the fine-tuned local clock.

[0011] Furthermore, in step S3, the step of establishing a spatial mapping relationship between the data points in the multi-source sensing data sequence and the structural units in the three-dimensional digital model includes: S31: The three-dimensional digital model is meshed using a spatial indexing algorithm to generate a spatial mesh structure containing several voxel units; S32: Obtain the physical coordinates of each sensor in the wharf structure corresponding to the multi-source heterogeneous sensing data; S33: Based on the physical coordinates, determine the voxel unit to which each sensor data point belongs using the nearest neighbor search algorithm; S34: Using a spatial interpolation algorithm, the discrete strain distribution data inside the concrete is mapped to the nodes of each voxel element in the three-dimensional digital model to form continuous strain field data; the spatial mapping relationship includes the strain field data.

[0012] Furthermore, step S4 includes: S41: Using the finite element inverse solution algorithm, the strain distribution data inside the concrete is input as a boundary condition into the three-dimensional digital model to calculate the three-dimensional stress tensor field of the concrete region. S42: Based on the geometric topological relationships defined by the three-dimensional digital model, identify the spatial trajectory and bonding interface of the steel reinforcement bundle within the concrete area; S43: Extract the strain component along the reinforcement direction of the concrete at the bonding interface, and map the strain component along the reinforcement direction to the axial strain of the steel bar bundle; S44: Combine the macroscopic spatial displacement data to perform geometric nonlinear correction on the axial strain, and use the constitutive equation in the material mechanics properties to convert the corrected axial strain into the axial stress of the steel bar bundle, thereby obtaining the internal stress state.

[0013] Furthermore, in step S5, the step of assessing the cumulative degree of fatigue damage of the steel reinforcement bundle based on the internal stress state includes: S51: Extract the incremental load spectrum of the axial stress of the steel bar bundle over time within the current monitoring time window from the internal stress state; S52: Perform cycle counting on the incremental load spectrum, identify several stress cycles, and determine the stress amplitude and average stress of each stress cycle; S53: Based on the SN curve in the mechanical properties of the material, find the fatigue life corresponding to each stress cycle and calculate the damage value of a single cycle; S54: The single-cycle damage values ​​of all stress cycles identified in the incremental load spectrum are accumulated to obtain the current incremental damage value, and the current incremental damage value is added to the stored historical cumulative damage value at the previous moment to calculate the degree of fatigue damage accumulation.

[0014] Furthermore, the method also includes the step of: S03: Identify whether there is a localized micro-corrosion environment around the internal steel reinforcement bundles and quantify the activity of micro-corrosion; Step S03 includes: S031: Statistically analyze the event density and total event energy of the acoustic emission data of the microscopic damage within a preset area; S032: Calculate the local tensile strain drift of the internal strain distribution data of the concrete within the preset area; S033: The micro-corrosion activity is calculated by combining the event density, the total energy of the event, and the local tensile strain drift.

[0015] Furthermore, step S54 includes: S541: Determine whether the inherent fatigue sensitivity level of the steel bar bundle is a sensitive level, and determine whether the micro-corrosion activity exceeds a preset activity threshold. S542: If the above conditions are met simultaneously, the nonlinear acceleration coefficient is calculated based on the micro-corrosion activity and the inherent fatigue sensitivity level. S543: The current incremental damage value is amplified using the nonlinear acceleration coefficient to obtain a corrected current incremental damage value; S544: Add the corrected current incremental damage value to the stored historical cumulative damage value from the previous moment to calculate the degree of fatigue damage accumulation.

[0016] Secondly, a multi-source data processing system for a wharf structure based on digital twins, applied in any of the methods described above, the system comprising: The data acquisition module is used to acquire macroscopic spatial displacement data, internal strain distribution data of concrete, and acoustic emission data of microscopic damage of the wharf structure. The calculation module is used to align the macroscopic spatial displacement data, the internal strain distribution data of concrete, and the acoustic emission data of microscopic damage with timestamps to generate a time-aligned multi-source sensing data sequence. The construction module is used to construct a three-dimensional digital model containing the geometric parameters and material mechanical properties of the concrete structure and internal steel reinforcement bundles, and to establish a spatial mapping relationship between the data points in the multi-source sensing data sequence and the structural units in the three-dimensional digital model. The estimation module is used to invert the stress field of the concrete region based on the internal strain distribution data of the concrete, and to estimate the internal stress state of the steel reinforcement bundle by combining the stress field and the macroscopic spatial displacement data. The display module is used to assess the degree of fatigue damage accumulation of the steel bar bundle based on the internal stress state, render the internal stress state and the degree of fatigue damage accumulation to the three-dimensional digital model for display, and generate an early warning signal when the degree of fatigue damage accumulation exceeds a preset threshold.

[0017] Beneficial Effects: This application proposes a multi-source data processing method and system for wharf structures based on digital twins. By acquiring macroscopic spatial displacement data, internal concrete strain distribution data, and microscopic damage acoustic emission data of the wharf structure, and aligning these data with timestamps, a time-aligned multi-source sensing data sequence is generated, solving the problem of inaccurate timestamp alignment in existing technologies. Furthermore, by constructing a three-dimensional digital model containing the geometric parameters and material mechanical properties of the concrete structure and internal steel reinforcement bundles, and establishing a spatial mapping relationship between data points in the multi-source sensing data sequence and structural units in the three-dimensional digital model, a precise physical basis for the digital twin model is provided. Finally, by inverting the concrete zone based on the internal concrete strain distribution data... By analyzing the stress field of the domain and combining it with macroscopic spatial displacement data, the internal stress state of the steel reinforcement bundles is calculated, enabling a refined perception of the internal stress conditions of the structure. The degree of fatigue damage accumulation in the steel reinforcement bundles is assessed based on the internal stress state, and the internal stress state and fatigue damage accumulation are rendered onto a 3D digital model for display. An early warning signal is generated when the fatigue damage accumulation exceeds a preset threshold. This solves the problem that digital twin models cannot reflect subtle fatigue accumulation in the structure in real time, and provides a timely and effective early warning mechanism. It addresses the issues of inaccurate timestamp alignment of multi-source data and the inability of digital twin models to reflect subtle fatigue accumulation in the structure in real time in existing technologies, thus improving the accuracy and early warning capabilities of wharf structural health monitoring. Attached Figure Description

[0018] Figure 1 This is a flowchart of a multi-source data processing method for a wharf structure based on digital twins proposed in this application.

[0019] Figure 2 This is a structural diagram of a multi-source data processing system for a wharf structure based on digital twins, as proposed in this application.

[0020] Figure 3 This is a schematic diagram of a multi-source data processing system for a wharf structure based on digital twins, as proposed in this application.

[0021] Labeling Explanation: 201, Data Acquisition Module; 202, Calculation Module; 203, Construction Module; 204, Estimation Module; 205, Display Module. Detailed Implementation

[0022] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and marked in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0023] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0024] This application proposes a multi-source data processing method for wharf structures based on digital twins. This method aims to address the problem that, during the long-term operation of large wharf structures, the difficulty in accurately aligning and fusing multi-source heterogeneous monitoring data across time and space prevents digital twin models from accurately reflecting the internal structure, particularly the true health status of critical load-bearing components such as reinforcing steel bundles, especially the accumulation of latent fatigue damage. This can potentially delay critical maintenance decisions and lead to safety hazards. This method constructs a digital mirror image that is synchronized in real-time with the physical wharf, achieving state resonance and enabling in-depth insights and proactive early warning of structural health conditions.

[0025] Please refer to Figure 1The method specifically includes the following steps: Step S1, acquiring macroscopic spatial displacement data, concrete internal strain distribution data, and microscopic damage acoustic emission data of the wharf structure. Step S2, aligning the acquired macroscopic spatial displacement data, concrete internal strain distribution data, and microscopic damage acoustic emission data with timestamps to generate a time-aligned multi-source sensing data sequence. Step S3, constructing a three-dimensional digital model containing the geometric parameters and material mechanical properties of the concrete structure and internal steel reinforcement bundles, and establishing a spatial mapping relationship between the data points in the multi-source sensing data sequence and the structural units in the three-dimensional digital model. Step S4, inverting the stress field of the concrete region based on the concrete internal strain distribution data, and combining the stress field and macroscopic spatial displacement data to calculate the internal stress state of the steel reinforcement bundles. Step S5, assessing the degree of fatigue damage accumulation of the steel reinforcement bundles based on the internal stress state, rendering the internal stress state and the degree of fatigue damage accumulation onto the three-dimensional digital model for display, and generating an early warning signal when the degree of fatigue damage accumulation exceeds a preset threshold.

[0026] In the initial stage of the methodology, it is necessary to comprehensively collect data reflecting the health status of the wharf structure at different scales, from macro to micro. This step is the foundation for all subsequent analyses, and the comprehensiveness and accuracy of the data directly determine the reliability of the final assessment results.

[0027] Macroscopic spatial displacement data is primarily used to characterize the deformation, settlement, or tilting of the wharf structure as a whole in three-dimensional space. This data reflects the overall response of the structure under long-term self-weight, ship berthing impact, wave and tidal currents, and other external loads. In a specific implementation scenario, multiple high-precision Global Navigation Satellite System (GNSS) receivers, such as those compatible with China's BeiDou Navigation Satellite System and Global Positioning System (GPS), can be deployed at key locations on the wharf, such as the top surface of the caisson, the wharf's leading edge, and the connection points of approach bridges. These receivers continuously collect the three-dimensional coordinates of each monitoring point at a set sampling frequency, such as once per hour or increased to once per minute during critical events like ship berthing. By performing differential calculations with stable reference points, millimeter-level displacement data sequences are obtained. In another implementation scenario, a ground-based robotic total station can be used. Several reflecting prisms are pre-positioned on the wharf structure as monitoring points. The total station automatically performs periodic inspections according to a preset program, calculating the changes in the three-dimensional coordinates of each monitoring point by measuring angles and distances, thereby obtaining macroscopic displacement data. For larger-scale regional settlement monitoring, interferometric synthetic aperture radar technology can be used to obtain the wide-area ground settlement field of the wharf area by analyzing the phase differences of satellite radar images at different times, providing background reference for macroscopic displacement analysis.

[0028] Data on the internal strain distribution of concrete is used to reveal the local stress state within the structure. As the primary building material for wharves, the internal strain distribution of concrete directly affects the structure's load-bearing capacity and durability. Various sensing technologies can be employed to acquire this data. A preferred implementation is to pre-embed distributed fiber optic sensors within the reinforcing cages of key load-bearing components, such as pile foundations, beams, and panels, during the wharf structure construction phase. These sensors utilize principles such as Brillouin time-domain reflectometry or Rayleigh scattering to achieve continuous strain measurement with centimeter-level spatial resolution over a range of several kilometers along the fiber optic path. When the concrete deforms under stress, the fiber optic cable is stretched or compressed accordingly. By analyzing the changes in the optical signal, the strain value at each point along the fiber optic cable can be deduced, forming a detailed internal strain distribution cloud map. Another implementation is to use discretely positioned fiber Bragg grating strain sensors. Multiple gratings with different center wavelengths are connected in series on a single fiber, and these grating points are precisely fixed within the concrete at locations requiring focused monitoring. The reflected wavelength of each grating drifts linearly with the strain magnitude at its location, allowing for simultaneous acquisition of strain data from multiple key points using a demodulator. In addition, traditional vibrating wire strain gauges can also be used as a supplement or alternative, calculating strain by measuring changes in the vibration frequency of the internal steel wire.

[0029] Microscopic damage acoustic emission data is used to capture transient elastic waves released during microscopic damage events such as the initiation and propagation of microcracks within concrete materials, slippage at the steel-concrete interface, and the formation of corrosion products. These signals provide direct evidence of early, minute damage to the structure and possess extremely high sensitivity. Acquiring this type of data typically requires installing a series of high-frequency piezoelectric acoustic emission sensors on or near the surface of the wharf structure. These sensors are arranged in an array, forming a two-dimensional or three-dimensional sensor network, for example, on key components such as pile foundations and beams. When a microscopic damage event occurs within the structure, the generated acoustic emission waves propagate outwards and are received by the sensor array. By analyzing the arrival time difference of the same event signals received by different sensors, triangulation algorithms can be used to accurately determine the location, time, and intensity of the damage event.

[0030] Through the above multi-level and multi-dimensional data collection, a complete data foundation has been formed, providing the necessary input for the subsequent construction of a digital twin that can truly reflect the physical port status.

[0031] After acquiring the three types of heterogeneous data mentioned above, a key challenge lies in the fact that these data are typically generated by different acquisition systems at different sampling frequencies, each with its own independent local clock. If these data cannot be precisely aligned in the time dimension, it is impossible to correctly correlate the responses of different sensors under the same physical event. For example, a minor impact during ship berthing may simultaneously trigger millimeter-level macroscopic displacement, localized stress concentrations of dozens of micro-strains, and a series of low-energy acoustic emission events. If the timestamps of these data have millisecond or even second-level deviations, subsequent fusion analysis will misclassify these strongly correlated signals as independent, unrelated noise, thus losing valuable early warning information about structural damage. Therefore, it is necessary to align the timestamps of macroscopic spatial displacement data, concrete internal strain distribution data, and microscopic damage acoustic emission data to generate a time-aligned multi-source sensing data sequence.

[0032] Furthermore, in order to address the problem of low data timestamp alignment accuracy caused by nonlinear time drift of the time synchronization server mentioned in the background technology, the step of data timestamp alignment can specifically include receiving external time synchronization signals at the edge computing node and fine-tuning the local clock frequency based on local temperature data using an adaptive filtering algorithm; then, using the fine-tuned local clock to timestamp align macroscopic spatial displacement data, concrete internal strain distribution data, and microscopic damage acoustic emission data to generate a time-aligned multi-source induction data sequence.

[0033] The edge computing nodes here are data acquisition and preprocessing units deployed on-site at the dock, responsible for directly connecting to various sensors. External timing signals can originate from global navigation satellite systems, such as the high-precision second pulse signal provided by the BeiDou system, or from a network time protocol server. However, the frequency of the crystal oscillator inside the edge computing node, i.e., the frequency source of the local clock, drifts with changes in ambient temperature, and this drift is typically non-linear. Hard synchronization relying solely on periodic external timing signals cannot completely eliminate the accumulated error between two synchronization points. Therefore, this application introduces an adaptive filtering algorithm based on local temperature data. The edge computing node integrates a temperature sensor to monitor its operating environment temperature in real time. The adaptive filtering algorithm uses temperature data as input to dynamically predict and compensate for clock frequency drift caused by temperature changes, thereby fine-tuning the local clock frequency.

[0034] Specifically, the process of fine-tuning the local clock frequency using an adaptive filtering algorithm can be broken down into the following steps. First, a clock error state equation is established, which uses clock phase deviation and frequency drift as state variables, and local temperature data as a control input variable affecting frequency drift. This state equation mathematically describes the physical process of clock error evolution over time. For example, the clock error model can be represented as a two-dimensional linear system, with the state vector containing phase error and frequency error. The state transition equation describes how the current state evolves from the state at the previous moment, where the change in frequency error is related to the difference between the current temperature and the temperature at the previous moment.

[0035] Secondly, based on the state estimate from the previous moment and the current local temperature data, the clock error state at the current moment is predicted. This step involves extrapolating the state using the established state equations without external time synchronization, continuously tracking the dynamic changes in the clock error.

[0036] Next, upon receiving an external time signal, the observation residual between the external time signal and the predicted value is calculated. This observation residual is used to correct the clock error state at the current moment, yielding the optimal clock frequency drift estimate. The external time signal provides an absolutely accurate time reference. The difference between this and the predicted value of the local clock is the observation residual. This residual reflects the accumulated error of the prediction model. Using Kalman filtering or its variants, this observation residual can be incorporated into the state estimation to correct the predicted state, resulting in a statistically optimal clock error estimate for the current moment, including optimal estimates of phase deviation and frequency drift.

[0037] Finally, the counting frequency of the local clock is dynamically compensated based on the optimal clock frequency drift estimate to generate a fine-tuned local clock. Specifically, this can be achieved by adjusting the frequency control word of the digital oscillator or by increasing or decreasing the count value of the clock counter via software, thereby offsetting the estimated frequency drift and bringing the local clock frequency as close as possible to the ideal frequency.

[0038] To further improve the accuracy of clock frequency compensation, considering that the heat generated by the edge computing node itself can also interfere with the clock crystal oscillator, compensation for internal thermal disturbances can be introduced in the step of dynamically compensating the counting frequency of the local clock based on the optimal clock frequency drift estimate. Specifically, this step includes real-time monitoring of the instantaneous power consumption of high-power components inside the edge computing node and obtaining the local hot spot temperature near the high-power components. High-power components mainly refer to central processing units, graphics processing units, or application-specific integrated circuits, etc., whose power consumption fluctuates drastically under different computing loads, causing rapid temperature changes in their vicinity and forming local hot spots. This internal heat source has a more direct and severe impact on the adjacent crystal oscillator than the ambient temperature.

[0039] Then, the internal disturbance compensation amount is calculated based on the monitored instantaneous power consumption and local hot spot temperature. A functional relationship model between power consumption, local temperature, and crystal oscillator frequency drift can be established in advance through experimental calibration or thermal simulation. During operation, based on real-time monitored data, a frequency compensation amount to offset the internal thermal disturbance is calculated using this model.

[0040] Next, the optimal clock frequency drift estimate obtained through adaptive filtering is used as the external offset compensation amount. This external offset compensation amount is then superimposed with the calculated internal disturbance compensation amount to obtain the final frequency adjustment command. The advantage of this approach is that it simultaneously considers the impact of both the slow changes in external ambient temperature and the rapid fluctuations in internal component power consumption on the clock frequency, resulting in more comprehensive and accurate compensation.

[0041] Finally, this final frequency adjustment command is used to adjust the counting frequency of the local clock, generating a highly stable and accurate fine-tuned local clock. All data collected by this edge computing node, whether displacement, strain, or acoustic emission signals, will be timestamped using this finely tuned local clock, thus ensuring high consistency of multi-source data in the time dimension from the source, laying a solid foundation for subsequent data fusion and analysis.

[0042] After aligning the timestamps of the data, it is necessary to associate these abstract time-series data with the physical entities of the wharf structure. This requires constructing a three-dimensional digital model and establishing a spatial mapping relationship between the data points in the multi-source sensing data sequence and the structural units in the three-dimensional digital model. This three-dimensional digital model is the skeleton of the digital twin, containing not only the precise geometric shape of the wharf structure but also its internal structure, such as concrete grade, reinforcement arrangement, and especially the spatial orientation, diameter, and material properties of the reinforcement bundles. The model can be constructed based on the wharf's original design drawings, such as computer-aided design drawings or building information models, or it can be reverse-engineered using technologies such as 3D laser scanning to obtain more realistic geometric information.

[0043] To establish a spatial mapping relationship between data points in a multi-source sensing data sequence and structural units in a 3D digital model, the following steps can be used to achieve efficient and accurate mapping. First, a spatial indexing algorithm is used to mesh the 3D digital model, generating a spatial mesh structure containing numerous voxel units. This is equivalent to discretizing the continuous geometric model into a set composed of a large number of tiny cubes or other shaped units. For example, an octree algorithm can be used to recursively partition the model space, using smaller voxel units in structurally complex or high-resolution areas, and larger voxel units in structurally simple areas, thereby improving computational efficiency while maintaining accuracy.

[0044] Secondly, it is necessary to obtain the physical coordinates of each sensor in the wharf structure corresponding to the multi-source heterogeneous sensing data. During the sensor deployment phase, it is necessary to accurately measure and record the three-dimensional coordinates of each sensor, whether it is a global navigation satellite system antenna, fiber optic grating, or acoustic emission probe, in the unified coordinate system of the wharf.

[0045] Then, based on the acquired physical coordinates, the nearest neighbor search algorithm is used to determine the voxel unit to which each sensor data point belongs. For each sensor, based on its physical coordinates, the voxel unit containing or closest to that coordinate point is quickly searched within the predefined spatial grid structure. A spatial indexing structure, such as an octree, can greatly accelerate this search process. In this way, the data collected by each sensor is bound to a specific voxel unit in the model.

[0046] Finally, for field data like strain, simply mapping discrete sensor data points to individual voxels is insufficient; a continuous field distribution is required. Therefore, spatial interpolation algorithms are needed to map the discrete strain distribution data within the concrete to the nodes of each voxel element in the 3D digital model, forming continuous strain field data. For example, Kriging interpolation or inverse distance weighted interpolation can be used to calculate the strain value at any location in the model, especially at voxel elements without sensors, based on the strain values ​​at known strain sensor locations. In this way, the discrete strain measurement data is expanded into a continuous 3D strain field covering the entire concrete region. This continuous strain field data is a crucial component of the spatial mapping relationship, providing complete input for subsequent mechanical analysis.

[0047] Through the above steps, all time-aligned sensing data is precisely "anchored" to the corresponding positions in the three-dimensional digital model, achieving data unification in both time and space dimensions. The digital twin begins to possess the ability to perceive the state of the physical world.

[0048] Next, we move into the core analysis phase of the digital twin, which involves inferring key physical quantities within the structure that cannot be directly measured, based on the sensor data already mapped onto the model. In this application, the core objective is to estimate the internal stress state of the reinforcing steel bundles. Since directly installing sensors on the reinforcing steel bundles is very difficult and may affect their performance, an indirect estimation method is required. This method inverts the stress field in the concrete region based on the strain distribution data within the concrete, and combines the stress field with macroscopic spatial displacement data to estimate the internal stress state of the reinforcing steel bundles.

[0049] When estimating the internal stress state of reinforcing steel strands, since the stress in the strands cannot be directly measured, this application proposes a reverse estimation method based on measured strain data, specifically including the following steps. First, using a finite element inverse algorithm, the continuous internal strain distribution data of the concrete generated in the previous step is input as boundary conditions into a three-dimensional digital model to calculate the three-dimensional stress tensor field of the concrete region. Traditional finite element analysis is a forward problem, i.e., applying loads and solving for displacement and strain. The reverse solution here uses the known internal strain field as a constraint condition, and solves in reverse for the stress distribution that induces this strain field. This is equivalent to solving an inverse mechanical problem, thereby obtaining a detailed three-dimensional stress distribution within the entire concrete member.

[0050] The second step involves identifying the spatial trajectory and bond interface of the rebar bundles within the concrete area based on the geometric topological relationships defined in the 3D digital model. During the construction of the 3D digital model, the geometric paths of the rebar bundles were precisely defined. By querying the geometric information of the model, the spatial curve equation of each rebar bundle and its interface with the surrounding concrete can be accurately extracted.

[0051] The third step involves extracting the strain components along the reinforcement direction of the concrete at the bond interface. During this process, the calculated three-dimensional stress tensor field of the concrete region is used to verify the state of the bond interface. Specifically, it determines whether the stress value of the concrete surrounding the reinforcement bundle exceeds its tensile strength or has entered a state of plastic damage. If the stress field indicates local cracking or damage in the concrete, a bond slip correction factor is introduced to reduce the extracted strain components along the reinforcement direction; otherwise, it is considered a good bond. The (corrected) strain components along the reinforcement direction are then mapped to the axial strain of the reinforcement bundle. Based on the assumption that the reinforcement and concrete work together through a good bond, at the bond interface, the axial strain of the reinforcement and the strain components of the surrounding concrete along the reinforcement direction are coordinated. Therefore, the component of the concrete strain field along the tangential direction of the reinforcement bundle can be extracted from the bond interface identified in the previous step and used as the axial strain of the reinforcement bundle itself.

[0052] The fourth step involves applying geometric nonlinear correction to the axial strain using macroscopic spatial displacement data. The corrected axial strain is then converted into axial stress in the reinforcing bar bundle using constitutive equations from material mechanics, yielding the internal stress state. When the wharf structure undergoes significant overall displacement or rotation, considering only local strain may neglect the effects of geometric nonlinearity. Therefore, it is necessary to use the collected macroscopic spatial displacement data to describe the deformation state of the entire structure and correct the axial strain calculation of the reinforcing bar bundle based on this. The corrected axial strain, combined with the constitutive relationship of the reinforcing bar bundle material (i.e., the stress-strain curve), allows for accurate calculation of the axial stress in the reinforcing bar bundle. This time-varying axial stress is the final internal stress state of the reinforcing bar bundle that we need to obtain.

[0053] After obtaining data on the internal stress state of the steel reinforcement bundles, particularly the changes in axial stress over time, a quantitative assessment of the degree of fatigue damage accumulation is needed to evaluate their long-term health. Fatigue damage is the phenomenon where materials accumulate damage and eventually fracture under cyclic loading, even when the stress is far below their yield strength. For steel reinforcement bundles in wharf structures, berthing of ships and wave impacts create cyclic loads, leading to the continuous accumulation of fatigue damage.

[0054] The steps for assessing the cumulative fatigue damage of steel reinforcement bundles based on their internal stress state include: First, extracting the incremental load spectrum of axial stress over time within the current monitoring time window (e.g., the past 10 minutes or 1 hour) from the calculated internal stress state. This is actually a short-time series of data recording the stress change history of the steel reinforcement bundle in the latest monitoring cycle, rather than historical data for the entire life cycle.

[0055] Secondly, the incremental load spectrum is cyclically counted to identify several stress cycles, and the stress amplitude and mean stress of each stress cycle are determined. For irregular load spectra, a specific algorithm is needed to decompose them into a series of equivalent, regular stress cycles. The rainflow counting method, widely used in the industry, is an effective cycle counting method that can accurately identify all closed stress-strain hysteresis loops in the load spectrum, corresponding to a complete fatigue cycle.

[0056] Then, based on the predefined SN curve for reinforcing steel bundles in the material mechanics properties, the fatigue life corresponding to each stress cycle is found, and the damage value per cycle is calculated. The SN curve, or stress-life curve, describes the number of cycles a material can withstand at a specific stress amplitude. For each identified stress cycle, the allowable number of cycles N can be found on the SN curve based on its stress amplitude and mean stress. The damage value D caused to the material by a single cycle can be defined as 1 / N.

[0057] Finally, the single-cycle damage values ​​of all stress cycles identified in the current incremental load spectrum are summed to calculate the damage increment for the current cycle. Subsequently, this damage increment is added to the historical cumulative damage value stored in the digital twin database for the previous moment to update and obtain the current fatigue damage accumulation level. Based on linear cumulative damage criteria (e.g., the Palmgren-Mainner criterion), this incremental update mechanism ensures the real-time nature of damage assessment and the continuity of historical data. Theoretically, fatigue failure will occur when the updated fatigue damage accumulation level reaches 1.

[0058] Given that the wharf structure is exposed to a humid and salt spray environment for extended periods, corrosion is a significant factor accelerating fatigue damage to the reinforcing steel strands. Corrosion creates stress concentration points on the steel surface, accelerating the initiation of fatigue cracks. Therefore, prior to fatigue assessment, this application proposes a method for identifying and quantifying localized micro-corrosion environments. This method includes identifying the presence of localized micro-corrosion environments around the internal reinforcing steel strands and quantifying the activity of these micro-corrosion.

[0059] Specifically, this step includes: statistically analyzing the event density and total energy of acoustic emission data related to microscopic damage within a predetermined region. The predetermined region refers to a concrete area within a certain range surrounding a specific steel reinforcement bundle in a three-dimensional digital model. Corrosion processes, especially active corrosion, are accompanied by microscopic events such as hydrogen embrittlement and cracking of the corrosion product layer, all of which generate acoustic emission signals. Therefore, a higher density and energy of acoustic emission events within a region indicates more intense microscopic damage activity in that region, potentially related to corrosion activity.

[0060] Simultaneously, the local tensile strain drift within the preset area is calculated based on the internal strain distribution data of the concrete. After steel reinforcement corrodes, the volume of corrosion products expands, exerting a compressive effect on the surrounding concrete, leading to localized tensile strain. If the tensile strain in a certain area shows a continuous increasing trend over time, indicating tensile strain drift, this is likely caused by the expansion of internal steel reinforcement due to corrosion.

[0061] Finally, the micro-corrosion activity is calculated by combining the acoustic emission event density, total event energy, and local tensile strain drift. These three indicators can be weighted and fused, for example, using a pre-calibrated empirical formula: Micro-corrosion activity = w1 * event density + w2 * total event energy + w3 * local tensile strain drift, where w1, w2, and w3 are weighting coefficients. By combining monitoring data from different physical phenomena in this way, the activity level of the micro-corrosion environment around the rebar bundle can be quantified more reliably.

[0062] After quantifying the micro-corrosion activity, to obtain fatigue damage assessment results that more closely reflect actual working conditions, the traditional fatigue damage accumulation calculation method needs to be modified. The specific modification process is as follows: First, determine whether the inherent fatigue sensitivity level of the rebar bundle is sensitive, and determine whether the calculated micro-corrosion activity exceeds a preset activity threshold. Different types or batches of rebar bundles may have different sensitivities to corrosive environments; this can be pre-input as a material property. Subsequent modifications are only necessary when the rebar bundle itself is sensitive to corrosion fatigue, and the corrosion activity of its environment has indeed reached a significant level.

[0063] If the above conditions are met simultaneously, a nonlinear acceleration factor is calculated based on the micro-corrosion activity and the inherent fatigue sensitivity level. This acceleration factor reflects the degree to which corrosion shortens fatigue life. Its value can be determined based on a corrosion fatigue model established from experimental data. Generally, the higher the corrosion activity and the more sensitive the material, the larger the acceleration factor.

[0064] Then, this nonlinear acceleration factor is used to amplify the damage increment of the current cycle calculated earlier, resulting in a corrected damage increment. For example, the corrected damage increment = nonlinear acceleration factor * original damage increment.

[0065] Finally, the corrected damage increment is added to the stored historical cumulative damage value from the previous moment to calculate the final fatigue damage accumulation degree that takes into account the corrosion-fatigue interaction. This processing method only uses the current corrosion environment parameters to accelerate the current damage increment, avoiding erroneous corrections to damage from historical healthy periods, thus more accurately reflecting the health status of the rebar strands under real service conditions.

[0066] After completing all calculations and assessments, the final step of the method is to visualize and provide early warnings for the results. The calculated internal stress state of the rebar bundles and the assessed cumulative fatigue damage are rendered into a 3D digital model for display. For example, different colors can be used in the 3D model to represent the stress magnitude or damage level at different locations within the rebar bundles, forming an intuitive cloud map. Managers can view detailed information for any location through interactive operations such as rotating, zooming, and sectioning the model. When the cumulative fatigue damage at any location exceeds a preset safety threshold, such as 0.7, an early warning signal is automatically generated. This warning signal can manifest as a pop-up alert window on the monitoring interface, a text message or email sent to the manager's mobile phone, or even trigger an audible and visual alarm, prompting a detailed inspection or maintenance of the area.

[0067] Through the complete methodology described above, this application achieves a closed loop from multi-source data acquisition, spatiotemporal alignment, model mapping, internal state estimation to final health assessment and early warning, providing strong technical support for the safe operation of wharf structures.

[0068] To implement the above method, this application also provides a multi-source data processing system for wharf structures based on digital twins. Please refer to... Figure 2 , Figure 3The system comprises a data acquisition module 201, a calculation module 202, a construction module 203, an estimation module 204, and a display module 205. The data acquisition module 201 is responsible for performing the data acquisition steps, integrating various sensor interfaces to acquire macroscopic spatial displacement data of the wharf structure, internal strain distribution data of concrete, and acoustic emission data of microscopic damage. The calculation module 202 is responsible for performing the timestamp alignment step, incorporating the aforementioned local clock fine-tuning algorithm based on adaptive filtering, used for precise time synchronization of multi-source data, generating time-aligned multi-source sensing data sequences. The construction module 203 is responsible for establishing a three-dimensional digital model and data mapping, including three-dimensional modeling tools and spatial mapping algorithms, used to construct a three-dimensional digital model containing structural geometry and material properties, and to establish spatial mapping relationships between sensing data points and model units. The estimation module 204 is the core of the system's analysis, integrating a finite element inverse solution algorithm and mechanical estimation logic, used to invert the stress field based on strain data and estimate the internal stress state of the reinforcing steel bundles. Display module 205 is responsible for result evaluation, visualization, and early warning. It incorporates a fatigue damage accumulation assessment algorithm, including a modified model that considers corrosion effects, and renders the calculation results onto a 3D model. Simultaneously, it generates early warning signals based on preset thresholds. These modules work together to form a complete wharf structure health monitoring and early warning platform. In this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, without necessarily requiring or implying any actual relationship or order between these entities or operations.

[0069] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for processing multi-source data on a wharf structure based on digital twins, characterized in that, The method includes the following steps: S1: Acquire macroscopic spatial displacement data, internal strain distribution data of concrete, and acoustic emission data of microscopic damage of the wharf structure; S2: Time-stamp align the macroscopic spatial displacement data, the internal strain distribution data of concrete, and the acoustic emission data of microscopic damage to generate a time-aligned multi-source sensing data sequence; S3: Construct a three-dimensional digital model containing the geometric parameters and material mechanical properties of the concrete structure and internal steel reinforcement bundles, and establish a spatial mapping relationship between the data points in the multi-source sensing data sequence and the structural units in the three-dimensional digital model; S4: Based on the strain distribution data inside the concrete, the stress field of the concrete region is inverted, and combined with the stress field and the macroscopic spatial displacement data, the internal stress state of the steel reinforcement bundle is calculated. S5: Evaluate the degree of fatigue damage accumulation of the steel bar bundle based on the internal stress state, render the internal stress state and the degree of fatigue damage accumulation to the three-dimensional digital model for display, and generate an early warning signal when the degree of fatigue damage accumulation exceeds a preset threshold.

2. The method for processing multi-source data of a wharf structure based on digital twins according to claim 1, characterized in that, Step S2 includes: S21: Receive external timing signals at the edge computing node and fine-tune the local clock frequency based on local temperature data using an adaptive filtering algorithm; S22: Use the fine-tuned local clock to timestamp-align the macroscopic spatial displacement data, the concrete internal strain distribution data, and the microscopic damage acoustic emission data to generate a time-aligned multi-source induction data sequence.

3. The method for processing multi-source data of a wharf structure based on digital twins according to claim 2, characterized in that, Step S21 includes: S211: Establish a clock error state equation, wherein the state equation takes clock phase deviation and frequency drift as state variables, and the local temperature data as a control input variable affecting frequency drift; S212: Based on the state estimate of the previous moment and the current local temperature data, predict the clock error state at the current moment; S213: When the external timing signal is received, calculate the observation residual between the external timing signal and the predicted value, use the observation residual to correct the clock error state at the current time, and obtain the optimal clock frequency drift estimate. S214: Dynamically compensate the counting frequency of the local clock based on the optimal clock frequency drift estimate to generate the fine-tuned local clock.

4. The method for processing multi-source data of a wharf structure based on digital twins according to claim 3, characterized in that, Step S214 includes: S2141: Real-time monitoring of the instantaneous power consumption of high-power components inside the edge computing node, and acquisition of the local hot spot temperature near the high-power components; S2142: Calculate the internal disturbance compensation amount based on the instantaneous power consumption and the local hot spot temperature; S2143: The optimal clock frequency drift estimate is used as the external offset compensation amount, and the external offset compensation amount is superimposed with the internal disturbance compensation amount to calculate the final frequency adjustment command. S2144: The counting frequency of the local clock is adjusted using the frequency adjustment command to generate the fine-tuned local clock.

5. The method for processing multi-source data of a wharf structure based on digital twins according to claim 1, characterized in that, In step S3, the step of establishing a spatial mapping relationship between the data points in the multi-source sensing data sequence and the structural units in the three-dimensional digital model includes: S31: The three-dimensional digital model is meshed using a spatial indexing algorithm to generate a spatial mesh structure containing several voxel units; S32: Obtain the physical coordinates of each sensor in the wharf structure corresponding to the multi-source heterogeneous sensing data; S33: Based on the physical coordinates, determine the voxel unit to which each sensor data point belongs using the nearest neighbor search algorithm; S34: Using a spatial interpolation algorithm, the discrete strain distribution data inside the concrete is mapped to the nodes of each voxel element in the three-dimensional digital model to form continuous strain field data; the spatial mapping relationship includes the strain field data.

6. The method for processing multi-source data of a wharf structure based on digital twins according to claim 1, characterized in that, Step S4 includes: S41: Using the finite element inverse solution algorithm, the strain distribution data inside the concrete is input as a boundary condition into the three-dimensional digital model to calculate the three-dimensional stress tensor field of the concrete region. S42: Based on the geometric topological relationships defined by the three-dimensional digital model, identify the spatial trajectory and bonding interface of the steel reinforcement bundle within the concrete area; S43: Extract the strain component along the reinforcement direction of the concrete at the bonding interface, and map the strain component along the reinforcement direction to the axial strain of the steel bar bundle; S44: Combine the macroscopic spatial displacement data to perform geometric nonlinear correction on the axial strain, and use the constitutive equation in the material mechanics properties to convert the corrected axial strain into the axial stress of the steel bar bundle, thereby obtaining the internal stress state.

7. The method for processing multi-source data of a wharf structure based on digital twins according to claim 6, characterized in that, In step S5, the step of assessing the degree of fatigue damage accumulation of the steel reinforcement bundle based on the internal stress state includes: S51: Extract the incremental load spectrum of the axial stress of the steel bar bundle over time within the current monitoring time window from the internal stress state; S52: Perform cycle counting on the incremental load spectrum, identify several stress cycles, and determine the stress amplitude and average stress of each stress cycle; S53: Based on the SN curve in the mechanical properties of the material, find the fatigue life corresponding to each stress cycle and calculate the damage value of a single cycle; S54: The single-cycle damage values ​​of all stress cycles identified in the incremental load spectrum are accumulated to obtain the current incremental damage value, and the current incremental damage value is added to the stored historical cumulative damage value at the previous moment to calculate the degree of fatigue damage accumulation.

8. The method for processing multi-source data of a wharf structure based on digital twins according to claim 7, characterized in that, The method further includes the following steps: S03: Identify whether there is a localized micro-corrosion environment around the internal steel reinforcement bundles and quantify the activity of micro-corrosion; Step S03 includes: S031: Statistically analyze the event density and total event energy of the acoustic emission data of the microscopic damage within a preset area; S032: Calculate the local tensile strain drift of the internal strain distribution data of the concrete within the preset area; S033: The micro-corrosion activity is calculated by combining the event density, the total energy of the event, and the local tensile strain drift.

9. The method for processing multi-source data of a wharf structure based on digital twins according to claim 8, characterized in that, Step S54 includes: S541: Determine whether the inherent fatigue sensitivity level of the steel bar bundle is a sensitive level, and determine whether the micro-corrosion activity exceeds a preset activity threshold. S542: If the above conditions are met simultaneously, the nonlinear acceleration coefficient is calculated based on the micro-corrosion activity and the inherent fatigue sensitivity level. S543: The current incremental damage value is amplified using the nonlinear acceleration coefficient to obtain a corrected current incremental damage value; S544: Add the corrected current incremental damage value to the stored historical cumulative damage value from the previous moment to calculate the degree of fatigue damage accumulation.

10. A multi-source data processing system for a wharf structure based on digital twins, characterized in that, The system, applied in the method of any one of claims 1-9, comprises: The data acquisition module is used to acquire macroscopic spatial displacement data, internal strain distribution data of concrete, and acoustic emission data of microscopic damage of the wharf structure. The calculation module is used to align the macroscopic spatial displacement data, the internal strain distribution data of concrete, and the acoustic emission data of microscopic damage with timestamps to generate a time-aligned multi-source sensing data sequence. The construction module is used to construct a three-dimensional digital model containing the geometric parameters and material mechanical properties of the concrete structure and internal steel reinforcement bundles, and to establish a spatial mapping relationship between the data points in the multi-source sensing data sequence and the structural units in the three-dimensional digital model. The estimation module is used to invert the stress field of the concrete region based on the internal strain distribution data of the concrete, and to estimate the internal stress state of the steel reinforcement bundle by combining the stress field and the macroscopic spatial displacement data. The display module is used to assess the degree of fatigue damage accumulation of the steel bar bundle based on the internal stress state, render the internal stress state and the degree of fatigue damage accumulation to the three-dimensional digital model for display, and generate an early warning signal when the degree of fatigue damage accumulation exceeds a preset threshold.

Images