A method and device for early warning of damage to drainage pipes in transportation hubs

By collecting multi-source pipeline health data and combining it with simulation models, the location of damage to the drainage network can be automatically predicted, solving the problem of low efficiency in manual inspection, realizing a scientific early warning mechanism and accurate positioning, reducing operation and maintenance costs, and improving the effectiveness of risk prevention and control.

CN123116041APending Publication Date: 2026-11-17CENT SOUTH UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611080114.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-21
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

The operation and maintenance of existing drainage pipe networks in transportation hubs relies on manual inspections and lacks scientific health status grading standards and early warning mechanisms, resulting in high operation and maintenance costs and poor risk prevention and control.

Method used

By collecting multi-source pipeline health data, and combining high-fidelity fluid-structure interaction simulation models and intelligent damage location models, the system automatically outputs predicted pipeline damage locations and establishes scientific health status grading standards and early warning mechanisms.

Benefits of technology

It enables real-time and accurate damage location of drainage pipe networks, reduces operation and maintenance costs, and improves risk prevention and control capabilities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123116041A_ABST
    Figure CN123116041A_ABST
Patent Text Reader

Abstract

This application provides a method and device for early warning of damage to drainage pipes in transportation hubs. The method includes: collecting multi-source pipeline health data along the drainage network of the transportation hub; inputting the multi-source pipeline health data into a preset high-fidelity fluid-structure interaction simulation model, and obtaining the dynamic response simulation results of the pipeline structure by performing dynamic simulation on the pipeline structure of the drainage network; fusing the multi-source pipeline health data and the dynamic response simulation results to generate a comprehensive input feature tensor, and inputting the comprehensive input feature tensor into a damage intelligent localization model to output the predicted pipeline damage location; obtaining the maximum principal stress value corresponding to the predicted pipeline damage location from the dynamic response simulation results, and determining the health status rating of the drainage network based on the maximum principal stress value; matching the health status rating with a preset graded early warning rule base, and outputting graded early warning signals, which effectively reduces the operation and maintenance cost of the pipeline network and improves the risk prevention and control capability of the transportation hub pipeline network.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of intelligent operation and maintenance technology for transportation hub infrastructure, and in particular to a method and device for early warning of damage to drainage pipes in transportation hubs. Background Technology

[0002] The drainage network of transportation hubs is a core infrastructure ensuring the normal operation, fire-fighting water supply, and domestic water supply of various transportation hubs such as high-speed rail stations, conventional rail stations, and bus stations. This network is often buried underground in complex environments and is highly susceptible to problems such as pipe aging, joint corrosion, scale buildup, and localized damage due to soil corrosion, geological subsidence, water erosion, and traffic load vibrations. These issues can lead to pipeline leaks. Statistics show that the leakage rate of existing transportation hub networks often exceeds 20%, reaching as high as 50% in some areas, with daily leakage reaching nearly 100 tons. This not only causes serious water waste but can also trigger secondary problems such as ground subsidence, equipment short circuits, and insufficient fire-fighting water supply, seriously threatening the operational safety of transportation hubs and the safety of personnel and property.

[0003] Currently, the operation and maintenance of drainage pipe networks in transportation hubs mainly relies on traditional manual inspection methods. Manual inspections are inefficient and lack timeliness, making it impossible to locate damage and leaks in real time and accurately. There is a lack of scientific health status grading standards and early warning mechanisms. Pipeline operation and maintenance are mostly reactive, passively responding to leak accidents. This results in high operation and maintenance costs, poor risk control, and may also affect the normal operation of transportation hubs. Summary of the Invention

[0004] This application provides a method and device for early warning of damage to drainage pipes in transportation hubs, which addresses the problems of high operation and maintenance costs and poor risk prevention and control effects caused by the lack of scientific health status classification standards and early warning mechanisms in related technologies.

[0005] The first aspect of this application provides a method for early warning of damage to drainage pipes in transportation hubs, the method comprising: Collect multi-source pipeline health data along the drainage network in transportation hubs; The multi-source pipeline health data is input into a preset high-fidelity fluid-structure interaction simulation model. By performing dynamic simulation on the pipeline structure of the drainage network, the dynamic response simulation results of the pipeline structure are obtained. The multi-source pipeline health data and the dynamic response simulation results are fused to generate a comprehensive input feature tensor. The comprehensive input feature tensor is then input into a preset intelligent damage localization model to output the predicted pipeline damage location of the drainage network. The maximum principal stress value corresponding to the predicted pipeline damage location is obtained from the dynamic response simulation results, and the health status rating of the drainage network is determined based on the maximum principal stress value. The health status rating is matched with a preset tiered early warning rule base to generate and output tiered early warning signals.

[0006] Optionally, in the first implementation of the first aspect of this application, the step of collecting multi-source pipeline health data along the drainage network in the transportation hub includes: Based on the geometric layout, material properties, and historical operation and maintenance records of the drainage pipe network of the transportation hub, the key inspection areas and vulnerable parts of the drainage pipe network are determined. The key inspection areas and vulnerable parts of the pipeline are scanned by phased array ultrasonic testing equipment, the original radio frequency echo signal is received, the original radio frequency echo signal is imaged and processed, and the equivalent size and spatial coordinates of the defect are extracted from the imaging results to obtain quantitative characterization parameters of local damage to the drainage pipeline. By deploying distributed accelerometers, high-precision flow meters, and pressure transmitters along the drainage pipeline, the vibration acceleration time series data, cross-sectional flow velocity time series data, and dynamic water pressure time series data of the drainage pipeline under operating conditions are collected simultaneously. Time alignment and filtering denoising are performed on the vibration acceleration time series data, the cross-sectional flow velocity time series data, and the hydrodynamic pressure time series data to obtain a multi-source dynamic monitoring data stream. The quantitative characterization parameters are associated and stored with the multi-source dynamic monitoring data stream to obtain the multi-source pipeline health data of the drainage pipeline.

[0007] Optionally, in the second implementation of the first aspect of this application, the step of inputting the multi-source pipeline health data into a preset high-fidelity fluid-structure interaction simulation model, and obtaining the dynamic response simulation results of the pipeline structure by performing dynamic simulation on the pipeline structure of the drainage network, includes: Based on the dynamic water pressure time series data and cross-sectional flow velocity time series data, the transient pressure fluctuation value and flow velocity gradient value of the fluid in the pipeline under unsteady flow conditions are determined. By mapping the relationship between dynamic water pressure and pipe wall stress, the transient pressure fluctuation value and flow velocity gradient value are transformed into a dynamic distributed force system acting on the inner wall of the pipe. Based on the vibration transmission path and vibration amplitude and frequency characteristics of the track structure during the operation of the transportation hub, the corresponding traffic load excitation is transformed into periodic additional dynamic loads acting on the pipe supports and surrounding soil. The dynamic distributed force system and the periodic additional dynamic load are vector superimposed to generate a composite dynamic load spectrum on the pipeline structure. Based on the composite dynamic load spectrum, the stress tensor and nodal vibration displacement data of each location in the drainage network under the composite dynamic load are determined, and the full-field dynamic response simulation results of the pipeline structure under the combined action of fluid-structure interaction and traffic load are generated.

[0008] Optionally, in the third implementation of the first aspect of this application, the step of fusing the multi-source pipeline health data and the dynamic response simulation results to generate a comprehensive input feature tensor includes: Extract nodal vibration displacement data corresponding to the spatial location of the vibration acceleration time series data from the full-field dynamic response simulation results; The vibration acceleration time series data and the node vibration displacement data are time-stamp aligned and spatial coordinate matched to establish a correspondence between the two sets of data in the time and spatial dimensions. The vibration acceleration time series data and the nodal vibration displacement data after time stamp alignment and spatial coordinate matching are spliced ​​at the feature level to generate a comprehensive input feature tensor.

[0009] Optionally, in the fourth implementation of the first aspect of this application, before the step of inputting the comprehensive input feature tensor into a preset intelligent damage localization model and outputting the predicted pipeline damage location of the drainage network, the method further includes: Based on the combination of different damage locations, damage degrees, and flow conditions, the vibration acceleration time series data and the node vibration displacement data are sliced, and the corresponding damage location coordinates are labeled for each slice to generate a training sample set. The training sample set is input into the neural network model. The encoder of the neural network model processes the input sequence in the training sample set to obtain a spatiotemporal feature tensor. The decoder of the neural network model performs position regression decoding on the spatiotemporal feature tensor to generate the predicted damage location coordinates. A localization error loss function is generated based on the difference between the marked damage location coordinates in the training sample set and the predicted damage location coordinates. The weight parameters of the neural network model are iteratively updated according to the positioning error loss function until the positioning error loss function converges to a preset threshold, thereby obtaining a trained and optimized intelligent damage positioning model.

[0010] Optionally, in the fifth implementation of the first aspect of this application, the step of obtaining the maximum principal stress value corresponding to the predicted pipeline damage location from the dynamic response simulation results, and determining the health status rating of the drainage network based on the maximum principal stress value, includes: Extract the maximum principal stress value at the location corresponding to the predicted pipeline damage location from the full-field dynamic response simulation results, as well as the stress gradient distribution data within the preset area of ​​the predicted pipeline damage location; Based on the yield strength limit and tensile strength limit of the pipe material, the allowable stress benchmark value of the material is determined, and the maximum principal stress value is compared with the allowable stress benchmark value of the material to generate a stress level index; Defect size data corresponding to the predicted pipeline damage location is extracted from the multi-source pipeline health data, and distortion features are extracted from the vibration spectrum corresponding to the predicted pipeline damage location in the vibration acceleration time series data to generate spectral distortion feature values. The stress level index, the defect size level constituted by the defect size data, and the spectral distortion characteristic value are used as joint judgment indicators to determine the health status rating of the drainage pipe network.

[0011] Optionally, in the sixth implementation of the first aspect of this application, the step of matching the health status rating with a preset graded early warning rule base to generate and output a graded early warning signal includes: Based on the health status rating, a preset hierarchical early warning rule library is matched, and the corresponding threshold rules and condition triggering logic are retrieved from the hierarchical early warning rule library; By using the threshold rules and condition triggering logic, the stress level index, the defect size level, and the spectral distortion feature value are compared with the corresponding early warning thresholds at each level in the hierarchical early warning rule base to determine whether the health status rating meets the early warning triggering conditions. When the health status rating is determined to meet the early warning triggering conditions, the corresponding early warning information template is extracted from the graded early warning rule base according to the triggered early warning level. The predicted pipeline damage location, the health status rating, and the stress level index are then filled into the early warning information template to generate a graded early warning signal for early warning.

[0012] A second aspect of this application provides a damage warning device for drainage pipes in transportation hubs. The damage warning device is used to implement a method for warning of damage to drainage pipes in transportation hubs. The damage warning device includes: The acquisition module is used to collect multi-source pipeline health data along the drainage network in transportation hubs; The simulation module is used to input the multi-source pipeline health data into a preset high-fidelity fluid-structure interaction simulation model, and obtain the dynamic response simulation results of the pipeline structure by performing dynamic simulation on the pipeline structure of the drainage network. The prediction module is used to fuse the multi-source pipeline health data and the dynamic response simulation results to generate a comprehensive input feature tensor, and input the comprehensive input feature tensor into a preset intelligent damage localization model to output the predicted pipeline damage location of the drainage network. The determination module is used to obtain the maximum principal stress value corresponding to the predicted pipeline damage location from the dynamic response simulation results, and determine the health status rating of the drainage network based on the maximum principal stress value; The early warning module is used to match the health status rating with a preset hierarchical early warning rule base, generate and output hierarchical early warning signals.

[0013] A third aspect of this application provides an electronic device, including a memory and a processor, wherein the processor is configured to execute a computer program stored in the memory, and when the processor executes the computer program, it implements the steps of the traffic hub drainage pipe damage early warning method provided in the first aspect of this application.

[0014] The fourth aspect of this application provides a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the steps of the traffic hub drainage pipe damage early warning method provided in the first aspect of this application.

[0015] Compared with the prior art, this application has the following beneficial effects: By acquiring multi-source pipeline health data in real time through data collection devices deployed along the pipeline, and combining fluid-structure interaction simulation models and intelligent damage location models, the system automatically outputs predicted pipeline damage locations. It also establishes scientific health status grading standards and early warning mechanisms, changing the existing post-event maintenance and passive response to leakage accidents operation and maintenance mode of pipeline networks, effectively reducing pipeline network operation and maintenance costs, and improving the risk prevention and control capabilities of transportation hub pipeline networks. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart illustrating the early warning method for damage to drainage pipes in transportation hubs provided in this application embodiment; Figure 2 This is a schematic diagram of the neural network architecture provided in the embodiments of this application; Figure 3A schematic diagram of the program modules of the traffic hub drainage pipe damage early warning device provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0018] To make the inventive objectives, features, and advantages of this application more apparent and understandable, the technical solutions in 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 them. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] To address the issues of high maintenance costs and poor risk control caused by the lack of scientific health status grading standards and early warning mechanisms in manual inspections of related technologies, this application provides a method for early warning of drainage pipe damage in transportation hubs. Figure 1 This is a flowchart illustrating the early warning method for damage to drainage pipes in transportation hubs provided in this embodiment. The method includes the following steps: Step 110: Collect health data of multi-source pipelines along the drainage network in the transportation hub.

[0020] Specifically, distributed accelerometers, high-precision flow meters, and pressure transmitters deployed along the pipeline are used to simultaneously collect data on vibration acceleration, cross-sectional flow velocity, and hydrodynamic pressure during pipeline operation. Simultaneously, phased-array ultrasonic testing technology is employed to obtain quantitative characterization parameters of local pipeline damage. Phased-array ultrasonic testing transmits ultrasonic waves and receives the raw radio frequency echo signals. After processing with a full-focusing imaging method, the equivalent size and spatial coordinates of defects are extracted, thus obtaining static baseline data of pipeline damage. The distributed sensor network continuously collects dynamic response signals during pipeline operation at a preset sampling frequency. After time alignment and filtering for noise reduction, a multi-source dynamic monitoring data stream is formed. This combines static damage detection and dynamic operation monitoring techniques to create a multi-source sensing system covering the entire pipeline condition.

[0021] Step 120: Input the multi-source pipeline health data into the preset high-fidelity fluid-structure interaction simulation model, and obtain the dynamic response simulation results of the pipeline structure by performing dynamic simulation on the pipeline structure of the drainage network.

[0022] Specifically, the transient pressure fluctuations and velocity gradients generated by the fluid in the pipeline under unsteady flow conditions are transformed into a dynamic distributed force system acting on the inner wall of the pipeline through a mapping relationship. Simultaneously, based on the vibration transmission path and amplitude-frequency characteristics of the track structure during the operation of the transportation hub, the traffic load excitation is transformed into periodic additional dynamic loads acting on the pipeline supports and surrounding soil. These two loads are then vector-superimposed to form a composite dynamic load spectrum. Based on this, a finite element discrete model is constructed by combining the pipeline's basic geometric information and material physical parameters. The stress tensor and nodal vibration displacement response of each pipeline element under the composite dynamic load are solved using structural dynamics time history analysis. This allows for the acquisition of full-field dynamic response simulation results reflecting the entire stress and deformation state of the pipeline at a relatively low computational cost.

[0023] Step 130: Perform feature fusion on the multi-source pipeline health data and dynamic response simulation results to generate a comprehensive input feature tensor, and input the comprehensive input feature tensor into the preset intelligent damage localization model to output the predicted pipeline damage location of the drainage network.

[0024] Specifically, nodal vibration displacement data corresponding to the spatial locations of measured vibration acceleration time-series data are extracted from the full-field dynamic response simulation results. The measured vibration signals and simulated displacement data are time-stamp aligned and spatial coordinate matched to establish a one-to-one correspondence between the two sets of data from different sources and with different physical meanings in the spatiotemporal dimension. Subsequently, the two types of aligned data are spliced ​​at the feature level to form a comprehensive input feature tensor that simultaneously contains measured perception information and simulated physical information. This enables the localization model to simultaneously utilize the ability of measured data to capture the real damage response and the ability of simulated data to constrain the physical field of the entire field, providing more complete input features for the data-driven localization model, thereby obtaining more accurate predictions of pipeline damage location information.

[0025] Step 140: Obtain the maximum principal stress value corresponding to the predicted pipeline damage location from the dynamic response simulation results, and determine the health status rating of the drainage network based on the maximum principal stress value.

[0026] Specifically, the spatial coordinates output by the intelligent damage location model are spatially matched with the stress tensor in the full-field dynamic response simulation results. The maximum principal stress value at the corresponding coordinate location is extracted from the simulation results. Simultaneously, the allowable stress benchmark value of the material is determined based on the yield strength limit and tensile strength limit of the pipeline material. The stress level index is obtained by calculating the ratio of the maximum principal stress value to the allowable stress value. This index serves as the core indicator for quantitatively characterizing the safety margin of the pipeline structure. By establishing a spatial correlation between the damage location results and the simulated stress field, the determination of the structural health status is transformed from an experience-based qualitative judgment to a quantitative judgment based on stress ratios.

[0027] Step 150: Match the health status rating with the preset graded early warning rule library to generate and output graded early warning signals.

[0028] Specifically, a hierarchical early warning rule base linked to the operational level of transportation hubs is pre-built. The rule base is configured with corresponding threshold rules and condition triggering logic for different health status levels. The stress level index, defect size level, and spectral distortion characteristic value are used as joint judgment indicators and compared with the early warning thresholds of each level in the rule base. When the current status is determined to meet the early warning triggering conditions, the corresponding early warning information template is extracted according to the triggered early warning level. The predicted pipeline damage location, health status level, and stress level index are filled into the template to form a structured hierarchical early warning signal.

[0029] In one optional implementation, the step of collecting multi-source pipeline health data along the drainage network in a transportation hub includes: determining the key inspection areas and vulnerable parts of the drainage network based on its geometric layout, material properties, and historical maintenance records; scanning the key inspection areas and vulnerable parts of the pipeline using a phased array ultrasonic testing device, receiving the original radio frequency echo signal, performing imaging processing on the original radio frequency echo signal, and extracting the equivalent size and spatial coordinates of the defects from the imaging results to obtain quantitative characterization parameters of local damage to the drainage pipeline; simultaneously collecting vibration acceleration time-series data, cross-sectional flow velocity time-series data, and dynamic water pressure time-series data of the drainage pipeline under operating conditions using distributed acceleration sensors, high-precision flow meters, and pressure transmitters deployed along the drainage pipeline; performing time alignment and filtering denoising processing on the vibration acceleration time-series data, cross-sectional flow velocity time-series data, and dynamic water pressure time-series data to obtain a multi-source dynamic monitoring data stream; and associating and storing the quantitative characterization parameters with the multi-source dynamic monitoring data stream to obtain multi-source pipeline health data of the drainage pipeline.

[0030] In this embodiment, in the operating environment of the drainage pipe network of the transportation hub, a structure-risk mapping relationship is constructed based on the geometric routing, material properties, and historical operation and maintenance records of the drainage pipe network. The geometric routing refers to the direction, bending angle, and connection topology of the pipe in the underground space. The material properties refer to the differences in mechanical response and corrosion sensitivity of materials such as steel, ductile iron, or polymer materials. The historical operation and maintenance records are used to reflect the locations of past leaks, repairs, and corrosion. Thus, the key detection areas and vulnerable parts of the drainage pipe network are comprehensively determined. The key detection areas are the key observation sections in areas of significant stress concentration or fluid scouring. Vulnerable parts are areas that are more prone to microcracks or wall thinning due to long-term load coupling.

[0031] A phased array ultrasonic testing device is used to scan the target area. The phased array ultrasonic testing device uses a polycrystalline array transducer to electronically control and focus the ultrasonic beam, achieving high-resolution detection of the internal structure of the pipe wall. The device receives the original radio frequency echo signal, which is the unenvelope high-frequency electrical signal formed after the ultrasonic wave is reflected at the material interface. The original radio frequency echo signal is then processed into an image. This image processing is based on a full-focusing algorithm to superimpose and reconstruct a high-resolution cross-sectional image from the echoes of different transmission and reception channels. The equivalent size and spatial coordinates of the defect are extracted from the imaging results. The equivalent size is used to characterize the equivalent geometric scale of the defect, and the spatial coordinates are used to describe the three-dimensional positioning information of the defect in the pipe wall thickness and circumferential position, thus obtaining quantitative characterization parameters of local damage in the drainage pipe.

[0032] Distributed accelerometers, high-precision flow meters, and pressure transmitters deployed along the drainage pipeline are used to simultaneously collect time-series data on vibration acceleration, cross-sectional flow velocity, and dynamic water pressure during operation. The distributed accelerometers capture changes in pipe wall vibration response, the high-precision flow meters acquire cross-sectional flow rate characteristics, and the pressure transmitters measure fluid dynamic pressure fluctuations. The vibration acceleration, cross-sectional flow velocity, and dynamic water pressure time-series data undergo time alignment and filtering / denoising. Time alignment corrects sampling clock deviations between different acquisition channels, ensuring that multi-source signals correspond to the same physical state under a unified time reference. Filtering / denoising suppresses high-frequency noise interference introduced by environmental electromagnetic interference and structural background vibration, resulting in a multi-source dynamic monitoring data stream. This multi-source dynamic monitoring data stream represents a fused multi-physics continuous observation sequence under a unified time reference.

[0033] By associating and storing quantitative characterization parameters with multi-source dynamic monitoring data streams, multi-source pipeline health data of drainage pipelines can be obtained. The association storage establishes a mapping relationship between defect spatial coordinates and corresponding time series signals, thereby achieving synchronous binding of structural damage status and operational response data.

[0034] In one optional implementation, the steps of inputting multi-source pipeline health data into a preset high-fidelity fluid-structure interaction simulation model and obtaining the dynamic response simulation results of the pipeline structure by dynamically simulating the pipeline structure of the drainage network include: determining the transient pressure fluctuation value and velocity gradient value of the fluid in the pipeline under unsteady flow state based on the time series data of dynamic water pressure and the time series data of cross-sectional velocity; converting the transient pressure fluctuation value and velocity gradient value into a dynamic distributed force system acting on the inner wall of the pipeline through the mapping relationship between dynamic water pressure and pipe wall stress; converting the corresponding traffic load excitation into periodic additional dynamic loads acting on the pipeline supports and surrounding soil based on the vibration transmission path and vibration amplitude-frequency characteristics of the track structure during the operation of the transportation hub; vector superimposing the dynamic distributed force system and the periodic additional dynamic load to generate a composite dynamic load spectrum of the pipeline structure; determining the stress tensor and nodal vibration displacement data of each location of the drainage network under the composite dynamic load based on the composite dynamic load spectrum, and generating the full-field dynamic response simulation results of the pipeline structure under the combined action of fluid-structure interaction and traffic load.

[0035] In this embodiment, based on the time series data of dynamic water pressure and cross-sectional velocity, the fluid dynamic state can be jointly analyzed under a unified time reference. The dynamic water pressure time series data represents the transient pressure change sequence within the pipe over time, while the cross-sectional velocity time series data represents the velocity change sequence of the fluid passing through the pipe cross-section per unit time. By performing high-frequency differential analysis and phase space reconstruction on these two types of sequences, the transient pressure fluctuation values ​​and velocity gradient values ​​of the fluid in the pipe under unsteady flow conditions are determined. The transient pressure fluctuation values ​​are used to characterize the magnitude of pressure abrupt changes caused by valve opening and closing, pump station disturbances, or local blockages. The velocity gradient values ​​are used to characterize the rate of velocity change and shear strength distribution characteristics of the fluid in spatial locations. Through the fluid-structure interaction transfer function between dynamic water pressure and pipe wall stress, the transient pressure fluctuation values ​​and velocity gradient values ​​are transformed into a dynamically distributed force system acting on the inner wall of the pipe. This dynamically distributed force system equates the fluid action to a time-varying load distributed along the pipe wall, thus obtaining a spatially continuous force density expression, enabling fluid disturbances to be transformed into structural mechanics inputs. It should be noted that the dynamic distributed force system, in a physical sense, is the equivalent force distribution that changes continuously with space and time along the inner wall of the pipe. Its essence comes from the coupling mapping of transient pressure fluctuations and flow velocity shear effects on the pipe wall.

[0036] Based on the vibration transmission path and amplitude-frequency characteristics of the track structure during the operation of the transportation hub, the corresponding traffic load excitation is transformed into periodic additional dynamic loads acting on the pipe supports and surrounding soil. The vibration transmission path represents the energy path of track structure vibration propagating through the foundation soil to the underground pipe network. The amplitude-frequency characteristics describe the attenuation and amplification of different frequency components during propagation, enabling the periodic loads caused by train passage to form an equivalent dynamic input within the pipe support system. The dynamically distributed force system and the periodic additional dynamic loads are vector-superimposed to generate a composite dynamic load spectrum for the pipe structure. Vector superposition is used to synthesize fluid forces and external structural excitations within the same spatial coordinate system, allowing loads of different directions and frequencies to be expressed as a time-varying load set within a unified framework, thus forming a composite excitation input covering both the time and spatial domains.

[0037] Based on the composite dynamic load spectrum, the stress tensor and nodal vibration displacement data at various locations in the drainage pipe network under composite dynamic loads are determined. This generates full-field dynamic response simulation results for the pipe structure under the combined action of fluid-structure interaction and traffic loads. The stress tensor describes the distribution characteristics of normal and shear stresses within the pipe wall under multi-directional stress states, while the nodal vibration displacement data characterizes the spatial displacement response of discrete structural nodes under time-varying loads. The simulation process is based on finite element discrete modeling, dividing the pipe network geometry into continuous elements and assigning material property parameters. After loading the composite dynamic load spectrum, the structural dynamic equilibrium equations are solved to obtain the stress and displacement fields evolving over time.

[0038] In one optional implementation, the step of fusing multi-source pipeline health data and dynamic response simulation results to generate a comprehensive input feature tensor includes: extracting nodal vibration displacement data corresponding to the spatial location of the vibration acceleration time series data from the full-field dynamic response simulation results; aligning the vibration acceleration time series data and the nodal vibration displacement data with timestamps and matching their spatial coordinates to establish a correspondence between the two sets of data in the time and spatial dimensions; and performing feature-level concatenation on the time-stamp-aligned and spatially matched vibration acceleration time series data and nodal vibration displacement data to generate a comprehensive input feature tensor.

[0039] In this embodiment, in the full-field dynamic response simulation results, the nodal vibration displacement data represents the time-varying displacement response of each computational node in the finite element discrete model of the pipeline under the combined action of fluid-structure interaction and traffic load. This node corresponds to a spatial discrete point in the actual pipeline structure. To establish a correspondence with the vibration acceleration time series data, it is necessary to first spatially map the simulation mesh according to the sensor layout coordinate system, that is, transform the coordinates of the finite element nodes to the engineering coordinate system consistent with the distributed acceleration sensors, so that the simulation nodes and physical measurement points form a one-to-one correspondence or interpolation correspondence in the spatial dimension. In the time dimension, the vibration acceleration time series data is a continuously sampled signal, while the nodal vibration displacement data comes from the discrete time step output of the simulation solution process. There is a problem that the sampling frequency and time starting point are inconsistent between the two. Therefore, a unified time reference is established through a timestamp alignment mechanism. This mechanism takes the structural dynamics simulation time step as a reference, maps the sensor sampling time series to the same time axis, and obtains the simulation displacement value corresponding to the same moment through time interpolation, so that the monitoring signal and simulation response under the same physical state strictly correspond in the time dimension.

[0040] After completing the spatiotemporal consistency mapping, the vibration acceleration time series data and the nodal vibration displacement data are concatenated at the feature level. The vibration acceleration time series data reflects the intensity of the local inertial response, while the nodal vibration displacement data reflects the overall deformation trend of the structure. The two types of data are complementary in terms of feature dimensions. Feature-level concatenation generates a comprehensive input feature tensor by cascading the frequency domain features of the acceleration signal (such as the dominant frequency component and energy spectral density) and the time domain features of the displacement response (such as displacement amplitude and phase change) in the same feature vector space.

[0041] In one optional implementation, before the step of inputting the comprehensive input feature tensor into a preset intelligent damage localization model and outputting the predicted pipeline damage location of the drainage network, the method further includes: segmenting the vibration acceleration time series data and nodal vibration displacement data into slices according to the combination of different damage locations, damage degrees, and flow conditions, and labeling each slice with corresponding damage location coordinates to generate a training sample set; inputting the training sample set into a neural network model, processing the input sequence in the training sample set according to the encoder of the neural network model to obtain a spatiotemporal feature tensor, and performing position regression decoding on the spatiotemporal feature tensor through the decoder of the neural network model to generate the predicted damage location coordinates; generating a localization error loss function based on the difference between the labeled damage location coordinates and the predicted damage location coordinates in the training sample set; iteratively updating the weight parameters of the neural network model according to the localization error loss function until the localization error loss function converges to a preset threshold to obtain a trained and optimized intelligent damage localization model.

[0042] In this embodiment, when segmenting the vibration acceleration time-series data and nodal vibration displacement data according to different combinations of damage location, damage degree, and flow conditions, the essence is to construct physically consistent local sample segments in a multi-dimensional working condition space. The damage location characterizes the spatial distribution coordinates of defects in the pipeline structure, the damage degree describes the intensity of the defect's impact on structural stiffness, and the flow condition describes the dynamic state changes of the fluid inside the pipeline under different flow and pressure conditions. By combining and encoding these three types of conditions, the continuous operation process can be divided into multiple time windows with clear physical semantics, ensuring that each slice simultaneously corresponds to a specific structural state and fluid excitation state, thereby guaranteeing the consistency of samples in the spatiotemporal dimension. After segmentation, each slice is labeled with corresponding damage location coordinates, which are derived from the spatial localization results of defects obtained from prior phased array ultrasonic testing. These coordinates indicate the precise spatial location of the damage in the three-dimensional structure of the pipeline, giving each training sample a supervised learning label.

[0043] like Figure 2 As shown, when the training sample set is input into the neural network model, the encoder performs multi-layer feature transformation processing on the input sequence. The encoder uses a self-attention mechanism to model the spatiotemporal dependency between vibration acceleration time-series data and nodal vibration displacement data. The multi-head self-attention mechanism is used to calculate the correlation weights between different time segments in long-sequence data, thereby generating a spatiotemporal feature tensor with global dependencies. This tensor integrates the coupling features of fluid excitation and structural response. The decoder performs position regression decoding based on this spatiotemporal feature tensor. Position regression decoding maps high-dimensional features into continuous spatial coordinates. By performing a nonlinear mapping on the feature distribution, predicted damage location coordinates are generated. These coordinates represent the model's estimation of the spatial location of defects inside the pipe.

[0044] Based on the difference between the labeled damage location coordinates and the predicted damage location coordinates in the training sample set, a localization error loss function is generated. This function quantifies the spatial offset between the predicted and true coordinates, expressed using Euclidean distance or weighted distance, thus forming a continuously differentiable error evaluation index. The weight parameters of the neural network model are iteratively updated based on the localization error loss function. These weight parameters control the intensity of feature transformation at each layer of the network. The error gradient is propagated layer by layer to the encoder and decoder structures through backpropagation, allowing the model to gradually adjust its feature extraction and location mapping capabilities until the localization error loss function converges to a preset threshold, resulting in a trained and optimized intelligent damage localization model.

[0045] In one optional implementation, the step of obtaining the maximum principal stress value corresponding to the predicted pipeline damage location from the dynamic response simulation results and determining the health status rating of the drainage network based on the maximum principal stress value includes: extracting the maximum principal stress value at the location corresponding to the predicted pipeline damage location from the full-field dynamic response simulation results, as well as the stress gradient distribution data within a preset area of ​​the predicted pipeline damage location; determining the allowable stress benchmark value of the material based on the yield strength limit and tensile strength limit of the pipeline material, and comparing the maximum principal stress value with the allowable stress benchmark value of the material to generate a stress level index; extracting the defect size data corresponding to the predicted pipeline damage location from the multi-source pipeline health data, and extracting the distortion features of the vibration spectrum corresponding to the predicted pipeline damage location in the vibration acceleration time series data to generate a spectral distortion feature value; and using the stress level index, the defect size level constituted by the defect size data, and the spectral distortion feature value as a joint judgment index to determine the health status rating of the drainage network.

[0046] In one optional implementation, the maximum principal stress value at the location corresponding to the predicted pipeline damage location, and the stress gradient distribution data within a preset area of ​​the predicted pipeline damage location are extracted from the full-field dynamic response simulation results. Based on the yield strength limit and tensile strength limit of the pipeline material, the allowable stress benchmark value of the material is determined, and the maximum principal stress value is compared with the allowable stress benchmark value to generate a stress level index. Defect size data corresponding to the predicted pipeline damage location is extracted from multi-source pipeline health data, and distortion features are extracted from the vibration spectrum corresponding to the predicted pipeline damage location in the vibration acceleration time series data to generate spectral distortion feature values. The stress level index, the defect size level constituted by the defect size data, and the spectral distortion feature values ​​are used as joint judgment indicators to determine the health status rating of the drainage network.

[0047] In this embodiment, when extracting the maximum principal stress value at the location corresponding to the predicted pipeline damage location from the full-field dynamic response simulation results, the maximum principal stress value represents the largest eigenvalue of the three-dimensional stress tensor at that spatial point. This value characterizes the most dangerous tensile stress level that the material experiences under complex stress conditions, directly corresponding to the dominant mechanical factors of crack propagation or material instability. Simultaneously, stress gradient distribution data within a preset area of ​​the predicted pipeline damage location is extracted. This stress gradient distribution data describes the rate of change of stress within the spatial neighborhood, i.e., the increasing or decreasing trend of stress intensity per unit spatial distance, reflecting the expansion range and non-uniformity characteristics of the stress concentration area. Based on the yield strength limit and tensile strength limit of the pipe material, the allowable stress benchmark value of the material is determined. The yield strength limit represents the critical stress value at which the material transitions from elastic deformation to plastic deformation, and the tensile strength limit represents the maximum ultimate stress that the material can withstand during tension. The allowable stress benchmark value of the material is formed by subtracting a safety factor from both values. This value serves as a unified reference benchmark for structural safety assessment. The maximum principal stress value is compared with the allowable stress benchmark value of the material to generate a stress level index. This index is used to quantify the degree to which the current structural stress state approaches the safety boundary.

[0048] Defect size data corresponding to predicted pipeline damage locations are extracted from multi-source pipeline health data. This defect size data, derived from phased array ultrasonic imaging results, characterizes geometric parameters such as pit depth, crack length, or pore diameter. Distortion features are extracted from the vibration spectrum corresponding to the predicted pipeline damage locations in the vibration acceleration time-series data, generating spectral distortion feature values. These spectral distortion feature values ​​characterize the degree of abnormal shift in the frequency domain structure of the vibration signal, such as dominant frequency drift, harmonic enhancement, or energy bandwidth expansion. These changes reflect the impact of structural stiffness degradation on dynamic response. The stress level index, defect size level (composed of defect size data), and spectral distortion feature values ​​are used as joint judgment indicators. The defect size level discretizes continuous geometric defect parameters into a structural risk stratification characterization; the stress level index reflects the structural load-bearing safety margin; and the spectral distortion feature values ​​reflect the degree of abnormal dynamic response. These three factors are fused and analyzed within the same judgment space to determine the health status rating of the drainage network.

[0049] In one optional implementation, the step of matching the health status rating with a preset graded early warning rule library to generate and output a graded early warning signal includes: matching the health status rating with the preset graded early warning rule library, retrieving the corresponding threshold rules and conditional triggering logic from the graded early warning rule library; comparing the stress level index, defect size level, and spectral distortion characteristic value with the corresponding early warning thresholds at each level in the graded early warning rule library using the threshold rules and conditional triggering logic to determine whether the health status rating meets the early warning triggering conditions; when the health status rating meets the early warning triggering conditions, extracting the corresponding early warning information template from the graded early warning rule library according to the triggered early warning level, filling the predicted pipeline damage location, health status rating, and stress level index into the early warning information template, and generating a graded early warning signal for early warning.

[0050] In this embodiment, when matching the health status rating with a preset hierarchical early warning rule base, the hierarchical early warning rule base stores the threshold rule set and condition triggering logic set corresponding to different health levels. The threshold rules define the boundary ranges of stress level index, defect size level, and spectral distortion characteristic value under different risk levels, while the condition triggering logic describes the early warning activation conditions when multiple indicators jointly satisfy a specific combination relationship. After matching the rule base with the health status rating, the corresponding threshold rules and condition triggering logic are retrieved from the hierarchical early warning rule base. The health status rating represents the structural safety level output result calculated by the preceding joint judgment indicators; for example, green indicates a safe state, yellow indicates potential risk, and red indicates a high-risk critical state. The threshold rules and condition triggering logic are bound to this rating result, enabling the system to execute a unified risk judgment strategy within the same level framework.

[0051] Through threshold rules and conditional triggering logic, the stress level index, defect size level, and spectral distortion feature value are compared with the corresponding warning thresholds at each level in the hierarchical warning rule library. The stress level index characterizes the degree to which the structural stress approaches the material's allowable boundary; the defect size level reflects the degree to which the geometric scale of local damage weakens the structural integrity; and the spectral distortion feature value characterizes the degree of abnormal shift in the structural dynamic response in the frequency domain. Through a multi-threshold parallel comparison mechanism, when any indicator exceeds the corresponding level threshold or multiple indicators meet the triggering relationship, the health status rating is determined to meet the warning triggering condition. When the health status rating meets the warning triggering condition, the corresponding warning information template is extracted from the hierarchical warning rule library according to the triggered warning level. The warning information template defines the structured warning output format, including risk level identifier, spatial location field, and key indicator field. The predicted pipeline damage location, health status rating, and stress level index are filled into the warning information template, giving the warning information spatial positioning and risk quantification attributes, thereby generating a hierarchical warning signal for early warning.

[0052] For example, the graded early warning rule base is pre-configured with different threshold rules and trigger logic according to four levels of status: green, yellow, red and black. When the health status rating is green, it means that the stress level index is less than 50% of the allowable stress benchmark value of the material, the defect size level corresponds to the minor corrosion level, and the spectral distortion characteristic value only shows a slight main frequency drift. At this time, the corresponding threshold rule in the rule base only enables the monitoring trigger logic, that is, it allows continuous data updates but does not generate external alarms, and only marks the elbow area as a low-risk monitoring point within the system.

[0053] When the health status rating is yellow, the stress level index enters the 50% to 90% range. At the same time, the defect size level corresponds to the medium corrosion or crack propagation stage. The spectral distortion characteristic value shows high-frequency energy enhancement but no structural main frequency fracture. At this time, the rule base trigger logic switches to the early warning preparation mode. The threshold rule determines that the stress and vibration index have a synchronous upward trend and generates an internal prompt signal.

[0054] When the health status rating is red, the stress level index is close to or exceeds 90% of the material's allowable stress benchmark value, the defect size level corresponds to obvious wall thinning or crack propagation, and the spectral distortion characteristic value shows a main frequency shift and energy bandwidth expansion. At this time, the triggering logic in the rule base enters a strong early warning mode, and the structure is confirmed to be in a critical stress state through a multi-threshold consistency judgment mechanism.

[0055] When the health status rating reaches the black level, the stress level index exceeds the safety boundary corresponding to the tensile strength limit of the material, the defect size level corresponds to penetrating damage or severe structural weakening, and the spectral distortion characteristic value exhibits broadband noise characteristics. At this time, the rule base trigger logic enters the emergency failure response mode and directly generates an irreversible structural failure warning signal.

[0056] According to the proposed solution, a method for early warning of damage to drainage pipes in transportation hubs is provided. This method acquires multi-source pipeline health data in real time through data acquisition devices deployed along the pipeline. Combined with a fluid-structure interaction simulation model and a damage intelligent location model, it automatically outputs and predicts the location of pipeline damage. Furthermore, it establishes a scientific health status classification standard and an early warning mechanism. This method changes the existing post-event maintenance and passive response to leakage accidents in pipeline networks, effectively reduces pipeline network operation and maintenance costs, and improves the risk prevention and control capabilities of transportation hub pipeline networks.

[0057] Figure 3 This application provides an embodiment of a traffic hub drainage pipe damage early warning device, which can be used to implement the traffic hub drainage pipe damage early warning method described in the foregoing embodiments. For example... Figure 3 As shown, the early warning device for drainage pipe damage at this transportation hub mainly includes: Module 10 is used to collect multi-source pipeline health data along the drainage network in transportation hubs; Simulation module 20 is used to input multi-source pipeline health data into a preset high-fidelity fluid-structure interaction simulation model, and obtain the dynamic response simulation results of the pipeline structure by performing dynamic simulation on the pipeline structure of the drainage network. Prediction module 30 is used to fuse multi-source pipeline health data and dynamic response simulation results to generate a comprehensive input feature tensor, and input the comprehensive input feature tensor into a preset intelligent damage localization model to output the predicted pipeline damage location of the drainage network. Module 40 is used to obtain the maximum principal stress value corresponding to the predicted pipeline damage location from the dynamic response simulation results, and to determine the health status rating of the drainage network based on the maximum principal stress value. The early warning module 50 is used to match the health status rating with the preset graded early warning rule base, generate and output graded early warning signals.

[0058] In one optional implementation, the acquisition module is specifically used for: determining the key inspection areas and vulnerable parts of the drainage pipe network based on the geometric wiring, material properties, and historical operation and maintenance records of the transportation hub drainage pipe network; scanning the key inspection areas and vulnerable parts of the pipe network using a phased array ultrasonic testing device, receiving the original radio frequency echo signal, performing imaging processing on the original radio frequency echo signal, extracting the equivalent size and spatial coordinates of the defects from the imaging results, and obtaining quantitative characterization parameters of local damage to the drainage pipe; synchronously collecting vibration acceleration time-series data, cross-sectional flow velocity time-series data, and dynamic water pressure time-series data of the drainage pipe network under operating conditions using distributed acceleration sensors, high-precision flow meters, and pressure transmitters deployed along the drainage pipe network; performing time alignment and filtering denoising processing on the vibration acceleration time-series data, cross-sectional flow velocity time-series data, and dynamic water pressure time-series data to obtain a multi-source dynamic monitoring data stream; and associating and storing the quantitative characterization parameters with the multi-source dynamic monitoring data stream to obtain multi-source pipe health data of the drainage pipe network.

[0059] In one optional implementation, the simulation module is specifically used to: determine the transient pressure fluctuation value and velocity gradient value of the fluid in the pipeline under unsteady flow conditions based on the time series data of dynamic water pressure and the time series data of cross-sectional velocity; convert the transient pressure fluctuation value and velocity gradient value into a dynamic distributed force system acting on the inner wall of the pipeline through the mapping relationship between dynamic water pressure and pipe wall stress; convert the corresponding traffic load excitation into periodic additional dynamic loads acting on the pipeline supports and surrounding soil based on the vibration transmission path and vibration amplitude and frequency characteristics of the track structure during the operation of the transportation hub; vector superposition of the dynamic distributed force system and the periodic additional dynamic load to generate a composite dynamic load spectrum of the pipeline structure; determine the stress tensor and nodal vibration displacement data of each location of the drainage network under the composite dynamic load based on the composite dynamic load spectrum, and generate the full-field dynamic response simulation results of the pipeline structure under the combined action of fluid-structure interaction and traffic load.

[0060] In one optional implementation, the prediction module is specifically used to: extract nodal vibration displacement data corresponding to the spatial location of the vibration acceleration time series data from the full-field dynamic response simulation results; perform time stamp alignment and spatial coordinate matching on the vibration acceleration time series data and the nodal vibration displacement data to establish a correspondence between the two sets of data in the time and spatial dimensions; and perform feature-level concatenation on the time stamp aligned and spatial coordinate matched vibration acceleration time series data and nodal vibration displacement data to generate a comprehensive input feature tensor.

[0061] In an optional implementation, the prediction module is further configured to: segment the vibration acceleration time series data and nodal vibration displacement data into slices based on different combinations of damage locations, damage degrees, and flow conditions, and label the corresponding damage location coordinates for each slice to generate a training sample set; input the training sample set into a neural network model, process the input sequence in the training sample set according to the encoder of the neural network model to obtain a spatiotemporal feature tensor, and perform position regression decoding on the spatiotemporal feature tensor through the decoder of the neural network model to generate predicted damage location coordinates; generate a positioning error loss function based on the difference between the labeled damage location coordinates and the predicted damage location coordinates in the training sample set; iteratively update the weight parameters of the neural network model according to the positioning error loss function until the positioning error loss function converges to a preset threshold to obtain a trained and optimized intelligent damage positioning model.

[0062] In one optional implementation, the determination module is used to: extract the maximum principal stress value at the location corresponding to the predicted pipeline damage location from the full-field dynamic response simulation results, as well as the stress gradient distribution data within the preset area of ​​the predicted pipeline damage location; determine the allowable stress benchmark value of the material based on the yield strength limit and tensile strength limit of the pipeline material, and compare the maximum principal stress value with the allowable stress benchmark value of the material to generate a stress level index; extract the defect size data corresponding to the predicted pipeline damage location from the multi-source pipeline health data, and extract the distortion features of the vibration spectrum corresponding to the predicted pipeline damage location in the vibration acceleration time series data to generate a spectral distortion feature value; and use the stress level index, the defect size level constituted by the defect size data, and the spectral distortion feature value as a joint judgment index to determine the health status rating of the drainage network.

[0063] In one optional implementation, the early warning module is used to: match a preset hierarchical early warning rule library according to the health status rating, and retrieve the corresponding threshold rules and condition triggering logic from the hierarchical early warning rule library; through the threshold rules and condition triggering logic, compare the stress level index, defect size level, and spectral distortion characteristic value with the corresponding early warning thresholds at each level in the hierarchical early warning rule library to determine whether the health status rating meets the early warning triggering conditions; when it is determined that the health status rating meets the early warning triggering conditions, extract the corresponding early warning information template from the hierarchical early warning rule library according to the triggered early warning level, fill the predicted pipeline damage location, health status rating, and stress level index into the early warning information template, and generate a hierarchical early warning signal for early warning.

[0064] According to the present application, a damage early warning device for drainage pipes in transportation hubs acquires multi-source pipeline health data in real time through acquisition devices deployed along the pipeline. It automatically outputs the predicted pipeline damage location by combining a fluid-structure interaction simulation model and a damage intelligent positioning model. It also establishes a scientific health status classification standard and an early warning mechanism, which changes the existing operation and maintenance mode of pipeline network that relies on post-event maintenance and passive response to leakage accidents. This effectively reduces the operation and maintenance cost of pipeline network and improves the risk prevention and control capabilities of transportation hub pipeline networks.

[0065] According to the scheme provided in this application Figure 4 An electronic device is provided as an embodiment of this application. This electronic device can be used to implement the traffic hub drainage pipe damage early warning method in the foregoing embodiments, and mainly includes: The system includes a memory 401, a processor 402, and a computer program 403 stored on the memory 401 and executable on the processor 402. The memory 401 and the processor 402 are communicatively connected. When the processor 402 executes the computer program 403, it implements the traffic hub drainage pipe damage early warning method described in the foregoing embodiments. The number of processors can be one or more.

[0066] The memory 401 can be a high-speed random access memory (RAM) or a non-volatile memory, such as a disk storage device. The memory 401 is used to store executable program code, and the processor 402 is coupled to the memory 401.

[0067] Furthermore, embodiments of this application also provide a computer-readable storage medium, which may be disposed in the electronic device described in the above embodiments, and the computer-readable storage medium may be as described above. Figure 4 The memory in the illustrated embodiment.

[0068] The computer-readable storage medium stores a computer program that, when executed by a processor, implements the traffic hub drainage pipe damage early warning method described in the foregoing embodiments. Furthermore, the computer-readable storage medium can also be a USB flash drive, portable hard drive, read-only memory (ROM), RAM, magnetic disk, or optical disk, or any other medium capable of storing program code.

[0069] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0070] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0071] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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 this application.

Claims

1. A traffic hub drain pipe damage early warning method, characterized in that, include: Collect multi-source pipeline health data along the drainage network in transportation hubs; The multi-source pipeline health data is input into a preset high-fidelity fluid-structure interaction simulation model. By performing dynamic simulation on the pipeline structure of the drainage network, the dynamic response simulation results of the pipeline structure are obtained. The multi-source pipeline health data and the dynamic response simulation results are fused to generate a comprehensive input feature tensor. The comprehensive input feature tensor is then input into a preset intelligent damage localization model to output the predicted pipeline damage location of the drainage network. The maximum principal stress value corresponding to the predicted pipeline damage location is obtained from the dynamic response simulation results, and the health status rating of the drainage network is determined based on the maximum principal stress value. The health status rating is matched with a preset tiered early warning rule base to generate and output tiered early warning signals.

2. The transportation hub drain pipe damage early warning method according to claim 1, characterized by, The steps for collecting multi-source pipeline health data along the drainage network in transportation hubs include: Based on the geometric layout, material properties, and historical operation and maintenance records of the drainage pipe network of the transportation hub, the key inspection areas and vulnerable parts of the drainage pipe network are determined. The key inspection areas and vulnerable parts of the pipeline are scanned by phased array ultrasonic testing equipment, the original radio frequency echo signal is received, the original radio frequency echo signal is imaged and processed, and the equivalent size and spatial coordinates of the defect are extracted from the imaging results to obtain quantitative characterization parameters of local damage to the drainage pipeline. By deploying distributed accelerometers, high-precision flow meters, and pressure transmitters along the drainage pipeline, the vibration acceleration time series data, cross-sectional flow velocity time series data, and dynamic water pressure time series data of the drainage pipeline under operating conditions are collected simultaneously. Time alignment and filtering denoising are performed on the vibration acceleration time series data, the cross-sectional flow velocity time series data, and the hydrodynamic pressure time series data to obtain a multi-source dynamic monitoring data stream. The quantitative characterization parameters are associated and stored with the multi-source dynamic monitoring data stream to obtain the multi-source pipeline health data of the drainage pipeline.

3. The transportation hub drain pipe damage early warning method according to claim 2, characterized by, The step of inputting the multi-source pipeline health data into a preset high-fidelity fluid-structure interaction simulation model, and obtaining the dynamic response simulation results of the pipeline structure by performing dynamic simulation on the pipeline structure of the drainage network, includes: Based on the dynamic water pressure time series data and cross-sectional flow velocity time series data, the transient pressure fluctuation value and flow velocity gradient value of the fluid in the pipeline under unsteady flow conditions are determined. By mapping the relationship between dynamic water pressure and pipe wall stress, the transient pressure fluctuation value and flow velocity gradient value are transformed into a dynamic distributed force system acting on the inner wall of the pipe. Based on the vibration transmission path and vibration amplitude and frequency characteristics of the track structure during the operation of the transportation hub, the corresponding traffic load excitation is transformed into periodic additional dynamic loads acting on the pipe supports and surrounding soil. The dynamic distributed force system and the periodic additional dynamic load are vector superimposed to generate a composite dynamic load spectrum on the pipeline structure. Based on the composite dynamic load spectrum, the stress tensor and nodal vibration displacement data of each location in the drainage network under the composite dynamic load are determined, and the full-field dynamic response simulation results of the pipeline structure under the combined action of fluid-structure interaction and traffic load are generated.

4. The transportation hub drain pipe damage early warning method according to claim 3, characterized by, The step of fusing the multi-source pipeline health data and the dynamic response simulation results to generate a comprehensive input feature tensor includes: Extract nodal vibration displacement data corresponding to the spatial location of the vibration acceleration time series data from the full-field dynamic response simulation results; The vibration acceleration time series data and the node vibration displacement data are time-stamp aligned and spatial coordinate matched to establish a correspondence between the two sets of data in the time and spatial dimensions. The vibration acceleration time series data and the nodal vibration displacement data after time stamp alignment and spatial coordinate matching are spliced ​​at the feature level to generate a comprehensive input feature tensor.

5. The transportation hub drain pipe damage early warning method according to claim 4, characterized by, Before the step of inputting the comprehensive input feature tensor into the preset damage intelligent localization model and outputting the predicted pipeline damage location of the drainage network, the method further includes: Based on the combination of different damage locations, damage degrees, and flow conditions, the vibration acceleration time series data and the node vibration displacement data are sliced, and the corresponding damage location coordinates are labeled for each slice to generate a training sample set. The training sample set is input into the neural network model. The encoder of the neural network model processes the input sequence in the training sample set to obtain a spatiotemporal feature tensor. The decoder of the neural network model performs position regression decoding on the spatiotemporal feature tensor to generate the predicted damage location coordinates. A localization error loss function is generated based on the difference between the damage location coordinates labeled in the training sample set and the predicted damage location coordinates. The weight parameters of the neural network model are iteratively updated according to the positioning error loss function until the positioning error loss function converges to a preset threshold, thereby obtaining a trained and optimized intelligent damage positioning model.

6. The transportation hub drain pipe damage early warning method according to claim 3, characterized by, The step of obtaining the maximum principal stress value corresponding to the predicted pipeline damage location from the dynamic response simulation results, and determining the health status rating of the drainage network based on the maximum principal stress value, includes: Extract the maximum principal stress value at the location corresponding to the predicted pipeline damage location from the full-field dynamic response simulation results, as well as the stress gradient distribution data within the preset area of ​​the predicted pipeline damage location; Based on the yield strength limit and tensile strength limit of the pipe material, the allowable stress benchmark value of the material is determined, and the maximum principal stress value is compared with the allowable stress benchmark value of the material to generate a stress level index; Defect size data corresponding to the predicted pipeline damage location is extracted from the multi-source pipeline health data, and distortion features are extracted from the vibration spectrum corresponding to the predicted pipeline damage location in the vibration acceleration time series data to generate spectral distortion feature values. The stress level index, the defect size level constituted by the defect size data, and the spectral distortion characteristic value are used as joint judgment indicators to determine the health status rating of the drainage pipe network.

7. The transportation hub drain pipe damage early warning method according to claim 6, characterized by, The step of matching the health status rating with a preset graded early warning rule base to generate and output graded early warning signals includes: Based on the health status rating, a preset hierarchical early warning rule library is matched, and the corresponding threshold rules and condition triggering logic are retrieved from the hierarchical early warning rule library; By using the threshold rules and condition triggering logic, the stress level index, the defect size level, and the spectral distortion feature value are compared with the corresponding early warning thresholds at each level in the hierarchical early warning rule base to determine whether the health status rating meets the early warning triggering conditions. When the health status rating is determined to meet the early warning triggering conditions, the corresponding early warning information template is extracted from the graded early warning rule base according to the triggered early warning level. The predicted pipeline damage location, the health status rating, and the stress level index are then filled into the early warning information template to generate a graded early warning signal for early warning.

8. A traffic hub drain pipe damage early warning device, characterized by, The transportation hub drainage pipe damage early warning device is used to implement the transportation hub drainage pipe damage early warning method according to claim 1, and the transportation hub drainage pipe damage early warning device includes: The acquisition module is used to collect multi-source pipeline health data along the drainage network in transportation hubs; The simulation module is used to input the multi-source pipeline health data into a preset high-fidelity fluid-structure interaction simulation model, and obtain the dynamic response simulation results of the pipeline structure by performing dynamic simulation on the pipeline structure of the drainage network. The prediction module is used to fuse the multi-source pipeline health data and the dynamic response simulation results to generate a comprehensive input feature tensor, and input the comprehensive input feature tensor into a preset intelligent damage localization model to output the predicted pipeline damage location of the drainage network. The determination module is used to obtain the maximum principal stress value corresponding to the predicted pipeline damage location from the dynamic response simulation results, and determine the health status rating of the drainage network based on the maximum principal stress value; The early warning module is used to match the health status rating with a preset hierarchical early warning rule base, generate and output hierarchical early warning signals.

9. An electronic device, comprising: Includes memory and processor, of which: The processor is used to execute computer programs stored in the memory; When the processor executes the computer program, it implements the steps in the traffic hub drainage pipe damage early warning method according to any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps in the early warning method for damage to drainage pipes in transportation hubs as described in any one of claims 1 to 7.