Method and device for identifying risk section of catenary on high-speed railway bridge

Abstract

This application proposes a method and equipment for identifying risk sections of high-speed railway bridges and overhead contact lines. Through regional system parameterization, structural network modeling, construction of seismic influence fields along the line, integrated algorithm calculation of bridge-net coupling probability prediction and progressive screening of the most unfavorable sections based on risk upper bound, risk mapping and progressive screening, and bridge-net risk dominance classification, the most unfavorable sections can be quickly identified and sorted.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method and equipment for identifying risk sections of overhead contact lines on high-speed railway bridges. Background Technology

[0002] High-speed railway bridges and overhead contact systems are distributed over long distances along the railway line, and the structural response of the bridges can affect the stress and deformation of the overhead contact supports and positioning points through bridge-to-net connections and constraints. Post-earthquake residual displacement of bridge piers and geometric misalignment of the contact wire can lead to a reduction in train operation safety and current collection quality. Therefore, it is necessary to quickly identify high-risk sections at the regional scale and formulate differentiated treatment plans.

[0003] Existing technologies have laid a certain foundation for regional bridge risk assessment or catenary local performance analysis, but the following shortcomings still exist:

[0004] First, the regional bridge-over-cruise system has a complex layout and variable scale, and lacks a rapid assessment framework for "unified characterization and unified inference" at the regional level;

[0005] Second, the impact of earthquakes under complex earthquake scenarios exhibits spatial differences along the line, especially affected by site zoning and topography (such as slope toe distribution), making it difficult to accurately characterize them through a single scalar intensity input.

[0006] Third, the parameter space for "identifying the most unfavorable section" is usually high in dimensionality and has many combinations. If exhaustive training or simulation of working conditions is carried out, the amount of computation will be huge, which is difficult to meet the needs of rapid decision-making after the earthquake.

[0007] Fourth, the risks of bridges and overhead contact lines have coupled effects and relatively dominant differences under the same earthquake scenario. If only two types of risks are given separately without dominant judgment, the recommendations for classification reinforcement and emergency repair will be insufficient.

[0008] Therefore, it is necessary to propose a technical method that can quickly generate regional risk distribution under complex seismic scenarios and reliably identify the most unfavorable sections at a lower operating cost. Summary of the Invention

[0009] This application proposes a method and equipment for identifying risk sections of overhead contact lines on high-speed railway bridges, which can solve one of the problems existing in the background art.

[0010] To achieve the above objectives, this application adopts the following technical solution:

[0011] Firstly, a method for identifying risk sections of high-speed railway bridge-overhead catenary system is provided, including:

[0012] Obtain the structural and locational parameters of the overhead contact system for high-speed railway bridges, as well as seismic scenario parameters;

[0013] The structural parameters and location parameters are converted into engineering object units and relational connection units that characterize the relationships between the engineering object units in the structural network model, and the seismic influence field is constructed from the seismic scenario parameters.

[0014] Furthermore, the structural network model and the seismic influence field are input into the constructed progressive screening model to obtain the risk section identification results of the high-speed railway bridge-caten network. The progressive screening model is guided by a risk upper bound and iteratively executes the following steps: for each corresponding resolution parameter combination and each section at the current progressive level, feature extraction is performed on the structural network model and the seismic influence field; prediction processing is performed based on the extracted features to obtain the probability distribution parameters and risk upper bound calibration amount of the key response indicators of the high-speed railway bridge-caten network; the risk indicators of the high-speed railway bridge-caten network for each section are determined based on the probability distribution parameters, and the risk set of the high-speed railway bridge-caten network is determined based on the risk indicators; the joint risk upper bound of the section is obtained based on the probability distribution parameters and the risk upper bound calibration amount; pruning is determined based on the pruning margin constructed from the joint risk upper bound, and processing corresponding to the pruning determination results is performed.

[0015] Based on the above technical solutions, the most unfavorable sections are rapidly identified and ranked through regional system parameterization, structural network modeling, construction of seismic influence fields along the route, integrated algorithm calculation of bridge-network coupling probability prediction and progressive screening of the most unfavorable sections based on risk upper bound guidance, risk mapping and progressive screening, and bridge-network risk-dominant classification.

[0016] In one possible design approach of the first aspect, the input feature vector of the progressive screening model is:

[0017] in, This represents the input feature vector corresponding to parameter combination θ for segment j under the progressive filtering at the r-th progressive level; This represents the structural features of segment j extracted by the structural network model. θ represents the seismic action characteristics of segment j extracted from the seismic influence field, and θ represents the parameter combination at the corresponding resolution under the current progressive level.

[0018] In one possible design approach of the first aspect, the risk indicators of the high-speed railway bridge-overhead catenary for each section are determined by the probability distribution parameters, specifically including:

[0019] Based on the probability distribution parameters, the damage exceedance probability P under the key response indicators of high-speed railway bridge-overhead catenary is determined. z,j :

[0020] Where Φ(·) is the standard normal distribution function, b represents the high-speed railway bridge object, c represents the overhead contact line object, and y th,z,k μ represents the response threshold corresponding to object category z under damage state k; lnY,z,j σ represents the mean of the natural logarithmic values ​​of the key response indicators for segment j within object category z; lnY,z,j The standard deviation of the natural logarithmic value of the key response index of object category z under parameter combination θ represents the key response index of segment j.

[0021] The risk index R is obtained by weighted summation of the damage exceedance probabilities. z,j :

[0022] Where ω k This is a weighting coefficient that increases with the damage level k.

[0023] In one possible design approach of the first aspect, the risk set of the high-speed railway bridge-overhead catenary is determined by the risk indicators, specifically by aggregating the risk indicators to obtain the risk set of the high-speed railway bridge-overhead catenary.

[0024] In one possible design approach of the first aspect, the probability distribution parameters include: the mean μ of segment j under object category z. z,j and standard deviation σ z,j The corresponding upper bound calibration value for risk is Δ. z,j Based on the probability distribution parameters and the risk upper bound calibration, the joint risk upper bound for this segment is obtained, specifically including:

[0025] The risk upper bound response of key response indicators for high-speed railway bridge-overhead catenary is determined based on the probability distribution parameters:

[0026] in, This represents the upper bound response of segment j to the risk of the z-th type of object under parameter combination θ. The inverse distribution function represents the distribution of the corresponding response, where α is the preset quantile level; μ z,j (θ) and σ z,j (θ) represent the mean and standard deviation of segment j under parameter combination θ for the z-th type of object, respectively; b represents the bridge object; c represents the catenary object.

[0027] Based on the risk upper bound response, determine the joint risk upper bound for segment j:

[0028] in, This represents the upper bound of the joint risk of segment j under the parameter combination θ. This represents the upper bound of the bridge risk in section j under the parameter combination θ. This represents the upper bound of the risk of the overhead contact system in section j under parameter combination θ; g b (·) and g c (·) represents the vulnerability mapping function corresponding to the bridge and the overhead contact system; η represents the correlation influence coefficient; ρ j (θ) is the key response correlation parameter.

[0029] In one possible design approach of the first aspect, the pruning margin m j (θ) is: , where m j (θ) represents the pruning margin of segment j under the parameter combination θ; R th This indicates a preset risk threshold;

[0030] Pruning decisions are made based on the pruning margin constructed from the joint risk upper bound, and corresponding processing is performed based on the pruning decision results, specifically including:

[0031] When m j When (θ) < −δ, pruning is performed on the corresponding segment-parameter combination; when |m j When (θ)∣≤δ, it is judged as a critical candidate and refined preferentially; when m j When (θ)>δ, it is retained as a high-risk candidate, where δ is the critical bandwidth parameter.

[0032] In one possible design approach of the first aspect, the iteration termination condition is: the progressive level reaches the maximum level, or the size of the candidate set is lower than a preset threshold, or the top K risk rankings of the most unfavorable segment remain stable within two consecutive levels.

[0033] In one possible design approach for the first aspect, a dominant index D is constructed based on risk indicators. j For segment j that meets the preset risk threshold conditions, bridge-dominated high risk, network-dominated high risk, and bridge-network dual high risk classification output is performed.

[0034] In one possible design approach of the first aspect, the structural parameters include: pier height sequence parameters, bridge span combination parameters, support spacing parameters, positioning point layout parameters, and boundary constraint parameters, and the location parameters include: route mileage parameters and spatial coordinate parameters under a unified reference coordinate system;

[0035] The earthquake scenario parameters include: ground motion intensity index, source coordinates, site type, and bridge site toe distribution factor;

[0036] The engineering object unit includes: bridge pier unit, catenary support unit, and positioning point unit. The bridge pier unit is defined by the line mileage parameter, the pier height sequence parameter, the bridge span combination parameter, and the boundary constraint parameter. The catenary support unit is defined by the line mileage parameter, the support spacing parameter, and the boundary constraint parameter. The positioning point unit is defined by the line mileage parameter, the positioning point layout parameter, and the boundary constraint parameter.

[0037] The relational connection unit includes: a bridge-net constraint connection relationship, which is defined by the boundary constraint parameters;

[0038] The earthquake influence field is defined by the ground motion intensity index, ground motion attenuation function, site type label, and bridge site toe distribution factor, wherein the ground motion attenuation function, the site type label, and the bridge site toe distribution factor are defined by the equivalent distance from the earthquake source to the mileage point.

[0039] In a second aspect, an electronic device is provided, comprising: a processor, and a memory coupled to the processor, the memory for storing a computer program; the processor for executing the computer program stored in the memory such that the electronic device performs the high-speed railway bridge-overhead catenary risk section identification method as described in any possible implementation of the first aspect. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a flowchart of a method for rapid identification of risk sections of regional high-speed railway bridges and overhead contact lines in complex earthquake scenarios, as proposed in an embodiment of this application.

[0042] Figure 2 This is a diagram of the integrated algorithm framework for bridge-network coupling probability prediction and progressive screening of the most unfavorable segment based on risk upper bound in the embodiments of this application.

[0043] Figure 3 This is a schematic diagram of a typical high-speed railway bridge—the overhead contact system—based on an embodiment of this application.

[0044] Figure 4 This is a diagram showing the reliability results of the algorithm pruning in the embodiments of this application;

[0045] Figure 5This is a graph showing the prediction efficiency and accuracy results of different algorithms in the embodiments of this application;

[0046] Figure 6 This is a diagram showing the predicted damage probability of the regional high-speed railway bridge-catenhead system in the embodiments of this application;

[0047] Figure 7 This is a diagram showing the results of the seismic risk dominance analysis of different bridge-overhead contact network structures in the embodiments of this application. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0049] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0051] Please read Figure 1 and Figure 2 , Figure 1 This invention presents a method for rapid identification of risk sections of overhead contact lines on regional high-speed railway bridges in complex seismic scenarios. Figure 2 The diagram illustrates the framework of an integrated algorithm for bridge-network coupling probability prediction and progressive screening of the most unfavorable segment based on risk upper bound, including the following steps:

[0052] Step 1: Select a typical high-speed railway section as an example. Collect and parameterize the basic information of the bridge and catenary system in this section. To unify the parameterization of the regional system and subsequent structural network modeling, establish a unified reference coordinate system along the overall extension direction of the railway line. The origin of the coordinate system is taken as the edge point of the network distribution area, and a global spatial coordinate system O-XY is established, where X is along the railway line direction and Y is the lateral direction. The positional parameters of bridge spans, piers, catenary supports, and positioning points are all recorded in the unified reference coordinate system and written into the regional feature set P.

[0053] In one embodiment, the set of region features P can be represented as:

[0054] Where s represents the route mileage parameter, x and y represent the spatial coordinate parameters under a unified reference coordinate system, χ represents the bridge span combination parameter, h represents the pier height sequence parameter, and d c The parameter d represents the spacing between the support columns. r β represents the positioning point layout parameters, and β represents the boundary constraint parameters.

[0055] Figure 3 The diagram shown is a typical high-speed railway bridge-caten network system in this embodiment. The bridge is arranged continuously along the line direction, and the catenary system is installed above the bridge section through supports and positioning points. Different bridge spans and different pier heights correspond to different bridge-caten network coupling and transmission characteristics.

[0056] Step 2: Discretize the regional system into a structural network model N=(U,L) composed of engineering object units and relational connection units. .like Figure 3 As shown, the engineering object unit U includes the pier unit U. b Contact wire support unit U c and positioning point unit U r The relational connection element L is a bridge-network constraint connection relationship L. bc .

[0057] In one embodiment, the i-th pier unit can be denoted as U. b,i The i-th contact wire support unit can be denoted as U. c,i The i-th positioning point unit can be denoted as U. r,i .

[0058] Among them, f b (P) represents the pier element mapping function, which is used to extract the line mileage parameters, pier height sequence parameters, bridge span combination parameters, and bridge boundary constraint parameters from the regional feature set P, and generate the i-th pier element U. b,i Attribute vector; s b,i h represents the mileage coordinate of the i-th pier unit along the track direction. i χ represents the height of the i-th pier; i β represents the span parameter corresponding to the i-th pier; b,i This represents the boundary constraint parameters of the i-th pier.

[0059] The overhead contact line support unit can be represented as:

[0060] Among them, f c(P) represents the catenary unit mapping function, which is used to extract line mileage parameters, pier height sequence parameters, bridge span combination parameters, and bridge boundary constraint parameters from the regional feature set P, and generate the i-th catenary unit U. c,i Attribute vector; s c,i d represents the mileage coordinates of the i-th contact wire support unit along the line direction; c,i β represents the spacing parameter of the i-th support column; c,i This represents the boundary constraint parameter of the i-th support.

[0061] The positioning point element can be represented as:

[0062] Among them, f r (P) represents the location point unit mapping function, which is used to extract the line mileage parameters, pier height sequence parameters, bridge span combination parameters, and bridge boundary constraint parameters from the regional feature set P, and generate the i-th location point unit U. r,i Attribute vector; s r,i d represents the mileage coordinates of the i-th contact wire positioning point unit along the line direction; r,i Represents the layout parameters of the i-th positioning point; β c,i This represents the boundary constraint parameters of the i-th positioning point.

[0063] Bridge-Network Constraint Transit Relations L bc,i It can be represented as:

[0064] Among them, g bc (P) represents the bridge-network constraint connection mapping function, used to extract bridge boundary constraints and bridge-network connection layout information from the regional feature set P, θ i Let Ψ be the contact wire constraint direction parameter on the i-th pier. i Let k be the constraint type parameter on the i-th pier. bc,i This represents the equivalent stiffness parameter during the transmission of bridge deck motion from the i-th pier to the catenary support. Through the construction of the aforementioned object units and relational connection units, the regional feature set P obtained in step one is uniformly written into the structural network model N to form a unified expression model of the regional bridge-catenary system.

[0065] Step 3: Input the earthquake scenario parameter set S, which includes at least the ground motion intensity index, source coordinates, site type, and bridge site toe distribution factor. Based on these parameters, generate the earthquake influence field F(s) along the route mileage coordinates s to uniformly characterize the spatial differences in earthquake effects experienced by different mileage sections.

[0066] In one embodiment, the seismic influence field F(s) is composed of the source-line distance function, the site type amplification factor, and the slope toe top terrain correction factor, and its expression can be: Wherein, IM is the seismic intensity index, G(·) is the seismic attenuation function, d(s) is the equivalent distance from the source to the mileage point, τ(s) is the site type label, κ(s) is the slope toe distribution factor, which is a correction coefficient characterizing the influence of the slope toe topography near the bridge site on the spatial distribution of seismic motion. It can be determined based on the bridge site topographic survey data, digital elevation model, existing site topographic zoning method, or existing topographic correction model. A(·) is the site type amplification function, used to obtain the corresponding site amplification coefficient based on the site type label τ(s); K(·) is the slope toe topographic correction function, used to obtain the corresponding topographic correction coefficient based on the slope toe distribution factor κ(s). The specific form of its calculation is not limited in this invention. In this embodiment, the seismic intensity index adopts peak ground acceleration (PGA), and combined with the site category and slope toe topographic distribution at different bridge sites along the line, a spatially differentiated seismic influence field along the line is constructed.

[0067] Step 4: Input the structural network model N established in Step 2 and the seismic influence field F(s) generated in Step 3 into the integrated algorithm of bridge-net coupling probability prediction and most unfavorable section progressive screening based on risk upper bound (UB-PWSS) to obtain the probability distribution parameter prediction results, bridge-net key response correlation parameters, upper bound calibration amount, and pruning margin corresponding to the key response indicators of the bridge and the catenary. In this embodiment, the bridge response indicator is selected from the pier top displacement, and the catenary response indicator is selected from the residual displacement of the positioning point.

[0068] The UB-PWSS integrated algorithm adopts a structure that combines shared feature extraction and multi-head collaborative output, including a feature assembly module, a shared feature extraction module, a dual-object probability parameter output module, a correlation-upper bound calibration module, and a joint risk determination module.

[0069] The feature assembly module is used to assemble the segment structural features, segment seismic action features, and current parameter combination features to form the algorithm input feature vector.

[0070] The shared feature extraction module is used to perform group mapping and interactive fusion on the input feature vector to extract shared latent features that can characterize the bridge-network coupling response. In one embodiment, the shared feature extraction module includes a group feature mapping unit and a bridge-network interactive gating unit. The group feature mapping unit is used to extract the latent representations of structural features, seismic action features and parameter combination features respectively. The bridge-network interactive gating unit is used to perform weighted fusion of latent features from different sources to highlight feature components that are sensitive to the bridge-network coupling response.

[0071] The dual-object probability parameter output module is used to output the probability distribution parameters of the bridge critical response and the catenary critical response respectively based on the shared latent features; wherein, the probability distribution parameters of the bridge critical response include the mean parameter μ. b,j (θ) and standard deviation parameter σ b,j (θ), the probability distribution parameters of the critical response of the overhead contact system include the mean parameter μ. c,j (θ) and standard deviation parameter σ c,j (θ); In one implementation, the standard deviation parameter is output in logarithmic form to ensure that the discrete parameter takes a positive value.

[0072] The correlation-upper bound calibration module is used to output the correlation parameter ρ between the bridge critical response and the catenary critical response based on the shared implicit features. j (θ), and output the upper bound calibration amount Δ of the bridge based on the statistical results of the predicted residuals on the calibration samples. b,j (θ) and the upper boundary calibration value Δ of the contact wire c,j (θ); In one embodiment, the correlation parameter is constrained to the interval [-1,1] by a bounded mapping, and the upper bound calibration is determined based on the positive residual distribution to reduce the probability of erroneous pruning caused by underestimation of the risk upper bound.

[0073] The joint risk assessment module is used to construct the joint risk upper bound of a section and calculate the pruning margin m based on the upper bound of bridge risk, the upper bound of catenary risk, the correlation parameters of bridge-catenary key responses, and preset risk thresholds. j (θ) is used to determine the corresponding segment-parameter combination as a pruning target, a priority refinement target, or a high-risk retention target. Through the above structure, it is possible to achieve integrated processing of bridge and catenary key response probability parameter prediction, bridge-catenary correlation characterization, risk upper bound conservative calibration, and segment pruning determination within the same algorithm framework.

[0074] For segment j and the current parameter combination θ, first construct the algorithm input feature vector:

[0075] in, This represents the algorithm input feature vector corresponding to parameter combination θ under the r-th layer of progressive filtering for segment j. The structural features of segment j extracted from the structural network model N include at least pier height, bridge span combination, support spacing, location point arrangement, bridge-network equivalent stiffness, and boundary constraints. The seismic action characteristics of segment j extracted from the seismic influence field F(s) are represented, including at least the ground motion intensity index, equivalent distance, site type, and slope toe correction factor; θ represents the parameter combination at the current level.

[0076] Furthermore, the upper bound of the critical response of the bridge and the overhead contact system can be determined by the following formula:

[0077] in, Let represent the upper bound response of segment j to the z-th type of object under parameter combination θ, b represent the bridge object, and c represent the overhead contact line object; Let α represent the inverse distribution function of the corresponding response distribution, where α is a preset quantile level, preferably α ≥ 0.9, and Δ z,j Indicates the upper bound calibration value; μ b,j (θ), σ b,j (θ) represents the probability distribution parameter of the key response index of the bridge in each segment j under the current parameter combination θ; μ c,j (θ), σ c,j (θ) represents the probability distribution parameter of the key response index of the overhead contact system for each segment j under the current parameter combination θ, ρ j (θ) is the critical response correlation parameter of the bridge-network.

[0078] Furthermore, considering the coupling correlation between the critical responses of the bridge and the overhead contact system, the upper bound of the joint risk of segment j can be further expressed as:

[0079] in, This represents the upper bound of the joint risk of segment j under parameter combination θ; g b (·) and g c (·) represent the vulnerability mapping functions corresponding to the bridge and the overhead contact system, respectively; η represents the correlation influence coefficient; when the correlation between the bridge and the overhead contact system is high and the risks of both the bridge and the overhead contact system are high, the upper bound of the joint risk is increased accordingly to reduce the risk of erroneous pruning and the risk of missed detection in the most unfavorable section;

[0080] Furthermore, the pruning margin is determined by the difference between the joint risk upper bound and the preset risk threshold, and can be expressed as:

[0081] Where, m j (θ) represents the pruning margin of segment j under the parameter combination θ; R th This indicates a preset risk threshold. When m j When (θ) < −δ, pruning is performed on the corresponding segment-parameter combination; when |m j When (θ)∣≤δ, it is judged as a critical candidate and refined preferentially; when m j When (θ)>δ, it is retained as a high-risk candidate, where δ is the critical bandwidth parameter.

[0082] Step 5: Based on the probability distribution parameters of the key response indicators of the bridge and catenary output in Step 4, the damage exceedance probability under each damage state is calculated using the log-normal vulnerability mapping function, and the bridge risk set R is formed through risk aggregation. b With contact network risk set R c Based on the pruning margin, determine the corresponding segment and parameter combination to perform pruning, retention, or priority refinement.

[0083] Furthermore, the vulnerability mapping adopts a log-normal form, representing the damage exceedance probability P under the response threshold. z,j,k Represented as:

[0084] Φ(·) is the standard normal distribution function, b represents the high-speed railway bridge object, c represents the overhead contact line object, and y th,z,k μ represents the response threshold corresponding to object category z under damage state k; lnY,z,j σ represents the mean of the natural logarithmic values ​​of the key response indicators for segment j within object category z; lnY,z,j The standard deviation of the natural logarithmic value of the key response index of object category z under parameter combination θ represents the key response index of segment j.

[0085] The risk index can be obtained by weighted summation of the damage exceedance probabilities:

[0086] Where ω k This is a weighting coefficient that increases with the damage level k.

[0087] In this embodiment, the risk index for each section is obtained by further aggregating the risks of each object unit within that section. The aggregation method can be maximum value, weighted average, exceedance probability aggregation, or a combination thereof. Through this step, the bridge risk distribution and catenary risk distribution for each section of the regional line can be obtained.

[0088] Step Six: Based on the parameter prediction results, risk sets, and pruning judgment results obtained in Steps Four and Five, a progressive refinement search is performed on the unpruned segment-parameter combinations until the most unfavorable segment set Ω is output. First, at the low-resolution parameter level, coupled probability parameter prediction, joint risk upper bound construction, and pruning margin calculation are performed on each segment-parameter combination to form a candidate set. Then, the candidate set is refined level by level, and pruning, retention, or priority refinement is performed based on the pruning margin judgment results, thereby reducing the number of high-precision calculation cases, and finally outputting the most unfavorable segment set Ω. In this step, the input is the bridge risk set R obtained in Step Five. b Overhead Contact Network Risk Set R cThe results include the pruning judgment results for each section and parameter combination; the processing involves progressively updating the candidate set, refining the parameters, and determining the stability of the sorting under multi-level parameter resolution; the output is the set of the most unfavorable sections Ω, the corresponding bridge risk results, and the catenary risk results.

[0089] The UB-PWSS progressive screening algorithm based on risk upper bound specifically includes the following steps:

[0090] S1. Construct a multi-level set of parameter resolutions, from coarse to fine, within the parameter space to be searched. (r) , where r represents the progressive level number; the parameters include at least the bridge pier height h and the seismic intensity index IM, and may further include parameters such as site type and boundary constraints; let the initial candidate set C(1) = Π(1);

[0091] S2. For each parameter combination θ∈C(r) of the current level r and each segment j, extract the segment structural features obtained from the structural network model N and the segment seismic action features obtained from the seismic influence field F(s), and form the corresponding algorithm input feature vector;

[0092] S3. Input the input feature vector into the UB-PWSS integrated algorithm to simultaneously obtain the probability distribution parameters of the bridge key response indicators, the probability distribution parameters of the catenary key response indicators, and the bridge-catenary key response correlation parameter ρ. j (θ), bridge upper boundary calibration value Δ b,j (θ) and the calibration amount Δ at the upper boundary of the contact wire c,j (θ);

[0093] S4. Calculate the upper bound responses of the critical responses of the bridge and the overhead contact system based on the probability distribution parameters output in step S3, and obtain the upper bounds of bridge risk and overhead contact system risk through the corresponding vulnerability mapping function; construct the joint upper bound of the risk for the section by combining the correlation parameters of the bridge-overhead contact system critical responses. ;

[0094] S5. Based on the upper bound of the joint risk of the section The pruning margin m is calculated using the preset risk threshold Rth. j (θ); when m j When (θ) is less than the negative threshold, the corresponding segment-parameter combination is discarded for subsequent refinement and high-precision calculation; when m j When (θ) is within the critical interval, the corresponding segment-parameter combination is marked as the priority refinement object; when m j When (θ) is greater than the positive threshold, the corresponding segment-parameter combination will be retained as a high-risk candidate;

[0095] S6. Expand the unpruned segment-parameter combination to the next level with a finer resolution parameter set, forming the next level candidate set C(r+1); repeat steps S2 to S5 until either of the following conditions is met: the maximum level R is reached. max If the size of the candidate set is lower than the preset threshold, or if the top K risk rankings of the most unfavorable segment remain stable within two consecutive layers;

[0096] S7. Output the set Ω of the segments with the highest joint risk or the top K in the final candidate set, and output the bridge risk results and the catenary risk results respectively.

[0097] Step 7: Bridge-Cable Risk Coupling Classification and Output. Construct the dominant index D based on bridge risk and overhead contact line risk. j The process involves classifying sections that meet preset risk thresholds into bridge-dominated high-risk, catenary-dominated high-risk, and bridge-catenary dual high-risk categories. In this step, the inputs are the set of most unfavorable sections Ω output from step six, along with the corresponding bridge and catenary risk indicators for each section. The processing involves risk source identification and classification based on the dominance index and preset risk thresholds. The output is a list of bridge-dominated high-risk, catenary-dominated high-risk, and bridge-catenary dual high-risk sections.

[0098] The dominant index , where ε is a very small positive number to prevent division by zero.

[0099] Furthermore, in this embodiment, the classification rules include at least:

[0100] When R b,j ≥R b,th And R c,j <R c,th At that time, it was determined that the bridge was the main high-risk area;

[0101] When R c,j ≥R c,th And R b,j <R b,th At that time, it was judged to be a high-risk network-dominated entity;

[0102] When R b,j ≥R b,th And R c,j ≥R c,th At that time, it was judged to be a high-risk area for both the bridge and the network. Among them, R b,j R b,th and R c,j R c,th These represent the risk indicators and thresholds for bridges and overhead contact lines, obtained by weighted summation of damage exceedance probabilities, respectively. This classification not only outputs the magnitude of the risk but also identifies the dominant sources of risk, thus providing a basis for subsequent differentiated reinforcement and emergency response.

[0103] Based on the above steps, a regional risk heat map and a segment ranking list are finally generated. Figure 6 The diagram shows the damage probability prediction results of the regional high-speed railway bridge-catenment system in this embodiment. B1-B16 represent 16 catenary sections on bridges at different distances from the earthquake source in the high-speed railway network of this embodiment. Pb represents the damage exceedance probability of the high-speed railway bridge at that location, and Pc represents the damage probability of the catenary at that location. The differences in risk distribution across different mileage sections under a given complex earthquake scenario can be clearly seen. Figure 7 The results of risk dominance analysis for different bridge-catenment sections are shown, which can be used to identify high-risk sections dominated by the bridge, high-risk sections dominated by the catenary, and high-risk sections dominated by both the bridge and the catenary. To verify the effectiveness of the UB-PWSS algorithm proposed in this invention, this embodiment compares it with the exhaustive full-parameter method without pruning. Figure 4 The figure shown is a reliability result diagram of the algorithm pruning. The results show that the present invention can achieve stable candidate screening effect and high recognition reliability while ensuring that the most unfavorable segment is not missed. Figure 5 The figure shows the prediction efficiency and accuracy results of different algorithms. In this embodiment, to verify the efficiency and accuracy of the proposed algorithm, RF, GBR, and MLP were selected as comparison algorithms. RF is the Random Forest algorithm, which performs risk prediction through the ensemble results of multiple decision trees; GBR is the Gradient Boosting Regression algorithm, which performs risk prediction by iteratively constructing weak learners and correcting prediction residuals; MLP is the Multilayer Perceptron algorithm, which establishes a nonlinear mapping relationship between input features and risk response through a multilayer neural network structure. All the above comparison algorithms use the same segment structure features, seismic action features, and parameter combination features as inputs, and evaluate the actuarial ratio required to achieve the high-risk segment recall rate requirement under the same risk threshold conditions. The results show that compared with the comparison algorithms such as RF, GBR, and MLP, as well as the direct exhaustive search method, the progressive most unfavorable segment search and working condition pruning method proposed in this invention can achieve the high-risk segment recall rate requirement with a lower actuarial ratio, indicating that this method can significantly reduce the number of high-precision working conditions, improve the search efficiency of the most unfavorable segment, and maintain the identification accuracy.

[0104] Figure 5 The figure shows the prediction efficiency and accuracy results of different algorithms. The results show that, compared with the comparison algorithms such as RF, GBR, MLP and the direct exhaustive search method, the progressive most unfavorable segment search and working condition pruning method proposed in this invention can achieve the high-risk segment recall rate requirement with a lower precision calculation ratio. This indicates that the method can significantly reduce the number of high-precision working conditions, improve the most unfavorable segment search efficiency, and maintain the recognition accuracy.

[0105] This invention provides a rapid identification method for risk sections of high-speed railway bridges and overhead contact lines in complex seismic scenarios, which has the following advantages: The method parametrically represents parallel objects of bridges and overhead contact lines in a region and constructs a structural network model containing engineering object units and relational connection units. It uniformly incorporates the bridge-contact line constraint transmission relationship into the model, achieving a unified representation and consistent calculation of the bridge-contact line coupling system at the regional scale. Simultaneously, it generates a seismic influence field along the railway line based on factors such as seismic source location, ground motion intensity, site type, and slope toe topography. This enables a unified characterization of the differences in seismic action across different mileage sections under complex seismic scenarios, thereby improving the scenario adaptability and spatial resolution of regional risk assessment. Furthermore, this invention proposes an integrated algorithm, UB-PWSS, for bridge-net coupling probability prediction and progressive screening of the most unfavorable section, guided by a risk upper bound. This algorithm simultaneously outputs the probability distribution parameters of the critical responses of the bridge and the overhead contact system, the correlation parameters of the bridge-net critical responses, the upper bound calibration amount, and the pruning margin. It also directly completes the joint risk upper bound construction, candidate section screening, and parameter hierarchy refinement, avoiding the loose application problem caused by separating probability prediction and section screening into two independent modules. Through the upper bound calibration mechanism, correlation correction mechanism, and critical zone priority refinement mechanism, this invention can significantly reduce the number of high-precision working conditions and computational overhead while ensuring that the most unfavorable risk section is not missed, achieving efficient search of a large parameter space and rapid locking of the most unfavorable section.

[0106] This application also provides an electronic device, including: a processor, and a memory coupled to the processor, the memory being used to store a computer program; the processor being used to execute the computer program stored in the memory, so that the electronic device performs the method as described in any of the above embodiments.

[0107] Electronic devices can be computing devices such as desktop computers, laptops, handheld computers, and cloud servers. These electronic devices may include, but are not limited to, processors and memory.

[0108] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting various parts of the device via various interfaces and lines.

[0109] The memory can be used to store the computer program, and the processor implements various functions of the electronic device by running or executing the computer program stored in the memory and calling the data stored in the memory.

[0110] The memory may primarily include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a given function, etc.; the data storage area may store data created based on the use of the computer, etc. Furthermore, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart media cards (SMC), secure digital cards (SD cards), flash cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0111] This application also provides a storage medium, which is a computer-readable storage medium. The computer program is stored in the computer-readable storage medium, and when executed by a processor, the computer program can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0112] This application also provides a computer program product, including: a computer program or instructions that, when the computer program or instructions are run on a computer, cause the computer to perform any of the above possible implementation methods.

[0113] The above description is the preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications are also considered to be within the scope of protection of this application.

Claims

1. A method for identifying risk sections of high-speed railway bridge-overhead catenary systems, characterized in that, include: Obtain the structural and locational parameters of the overhead contact system for high-speed railway bridges, as well as seismic scenario parameters; The structural parameters and location parameters are converted into engineering object units and relational connection units that characterize the relationships between the engineering object units in the structural network model, and the seismic influence field is constructed from the seismic scenario parameters. Furthermore, the structural network model and the seismic influence field are input into the constructed progressive screening model to obtain the risk section identification results of the high-speed railway bridge-caten network. The progressive screening model is guided by a risk upper bound and iteratively executes the following steps: for each corresponding resolution parameter combination and each section at the current progressive level, feature extraction is performed on the structural network model and the seismic influence field; prediction processing is performed based on the extracted features to obtain the probability distribution parameters and risk upper bound calibration amount of the key response indicators of the high-speed railway bridge-caten network; the risk indicators of the high-speed railway bridge-caten network for each section are determined based on the probability distribution parameters, and the risk set of the high-speed railway bridge-caten network is determined based on the risk indicators; the joint risk upper bound of the section is obtained based on the probability distribution parameters and the risk upper bound calibration amount; pruning is determined based on the pruning margin constructed from the joint risk upper bound, and processing corresponding to the pruning determination results is performed.

2. The method for identifying risk sections of high-speed railway bridge-overhead catenary as described in claim 1, characterized in that, The input feature vector of the progressive screening model is: in, This represents the input feature vector corresponding to parameter combination θ for segment j under the progressive filtering at the r-th progressive level; This represents the structural features of segment j extracted by the structural network model. θ represents the seismic action characteristics of segment j extracted from the seismic influence field, and θ represents the parameter combination at the corresponding resolution under the current progressive level.

3. The method for identifying risk sections of high-speed railway bridge-overhead catenary as described in claim 2, characterized in that, Based on the probability distribution parameters, the risk indicators for the high-speed railway bridge-overhead catenary in each section are determined, specifically including: Based on the probability distribution parameters, the damage exceedance probability P under the key response indicators of high-speed railway bridge-overhead catenary is determined. z,j : Where Φ(·) is the standard normal distribution function, b represents the high-speed railway bridge object, c represents the overhead contact line object, and y th,z,k μ represents the response threshold corresponding to object category z under damage state k; lnY,z,j σ represents the mean of the natural logarithmic values ​​of the key response indicators for segment j within object category z; lnY,z,j The standard deviation of the natural logarithmic value of the key response index for object category z under parameter combination θ represents the value of segment j. The risk index R is obtained by weighted summation of the damage exceedance probabilities. z,j : Where, ω k This is a weighting coefficient that increases with the damage level k.

4. The method for identifying risk sections of high-speed railway bridge-overhead catenary as described in claim 2, characterized in that, The risk set of high-speed railway bridge-overhead catenary is determined by the aforementioned risk indicators, specifically by aggregating the aforementioned risk indicators to obtain the risk set of high-speed railway bridge-overhead catenary.

5. The method for identifying risk sections of high-speed railway bridge-overhead catenary as described in claim 3, characterized in that, The probability distribution parameters include: the mean μ of segment j under object category z. z,j and standard deviation σ z,j The corresponding upper bound calibration value for risk is Δ. z,j Based on the probability distribution parameters and the risk upper bound calibration, the joint risk upper bound for this segment is obtained, specifically including: The risk upper bound response of key response indicators for high-speed railway bridge-overhead catenary is determined based on the probability distribution parameters: in, This represents the upper bound response of segment j to the risk of the z-th type of object under parameter combination θ. This represents the inverse distribution function of the corresponding response distribution, where α is the preset quantile level, and μ... z,j (θ) and σ z,j (θ) represents the mean and standard deviation of segment j for the z-th class of objects under parameter combination θ, respectively; b represents the bridge object; and c represents the overhead contact line object. Based on the risk upper bound response, determine the joint risk upper bound for segment j: in, This represents the upper bound of the joint risk of segment j under the parameter combination θ. This represents the upper bound of the bridge risk in section j under the parameter combination θ. This represents the upper bound of the risk of the overhead contact system in section j under parameter combination θ; g b (·) and g c (·) represent the vulnerability mapping functions for the bridge and the overhead contact system, respectively; η represents the correlation influence coefficient; ρ j (θ) is the key response correlation parameter.

6. The method for identifying risk sections of high-speed railway bridge-overhead catenary as described in claim 5, characterized in that, The pruning margin m j (θ) is: , where m j (θ) represents the pruning margin of segment j under the parameter combination θ; R th This indicates a preset risk threshold. Pruning decisions are made based on the pruning margin constructed from the joint risk upper bound, and corresponding processing is performed based on the pruning decision results, specifically including: When m j When (θ) < −δ, pruning is performed on the corresponding segment-parameter combination; when |m j When (θ)∣≤δ, it is judged as a critical candidate and refined preferentially; when m j When (θ)>δ, it is retained as a high-risk candidate, where δ is the critical bandwidth parameter.

7. The method for identifying risk sections of high-speed railway bridge-overhead catenary as described in claim 1, characterized in that, The iteration termination conditions are: the progressive level reaches the maximum level, or the size of the candidate set is lower than the preset threshold, or the top K risk rankings of the most unfavorable segment remain stable within two consecutive layers.

8. The method for identifying risk sections of high-speed railway bridge-overhead catenary as described in claim 1, characterized in that, Constructing the dominant index D based on risk indicators j For segment j that meets the preset risk threshold conditions, bridge-dominated high risk, network-dominated high risk, and bridge-network dual high risk classification output is performed.

9. The method for identifying risk sections of high-speed railway bridge-overhead catenary as described in claim 1, characterized in that, The structural parameters include: pier height sequence parameters, bridge span combination parameters, support spacing parameters, positioning point layout parameters, and boundary constraint parameters; the location parameters include: route mileage parameters and spatial coordinate parameters under a unified reference coordinate system. The earthquake scenario parameters include: ground motion intensity index, source coordinates, site type, and bridge site toe distribution factor; The engineering object unit includes: bridge pier unit, catenary support unit, and positioning point unit. The bridge pier unit is defined by the line mileage parameter, the pier height sequence parameter, the bridge span combination parameter, and the boundary constraint parameter. The catenary support unit is defined by the line mileage parameter, the support spacing parameter, and the boundary constraint parameter. The positioning point unit is defined by the line mileage parameter, the positioning point layout parameter, and the boundary constraint parameter. The relational connection unit includes: a bridge-net constraint connection relationship, which is defined by the boundary constraint parameters; The earthquake influence field is defined by the ground motion intensity index, ground motion attenuation function, site type label, and bridge site toe distribution factor, wherein the ground motion attenuation function, the site type label, and the bridge site toe distribution factor are defined by the equivalent distance from the earthquake source to the mileage point.

10. An electronic device, characterized in that, The electronic device includes: a processor, and a memory coupled to the processor. The memory is used to store computer programs; The processor is configured to execute the computer program stored in the memory, so that the electronic device performs the high-speed railway bridge-overhead catenary risk section identification method as described in any one of claims 1-9.

Images