A method and system for modeling and simulation of weather resistant sealant structures

Abstract

The application discloses a kind of weather-resistant sealing strip structure's modeling simulation method and system, comprising: the initial three-dimensional grid model of the sealing strip to be analyzed is constructed and is given initial material attribute, sequentially applied to model precipitation simulation working condition and dry environment simulation working condition, extract the surface microcrack node in open state and input to local corrosion evolution prediction model to output local corrosion degradation coefficient, after replacing initial corrosion rate parameter, Markov model for surface microcrack node is constructed and degradation coefficient is converted into state transition probability matrix, subsequently, the grid cell degradation state is calculated by Markov model, the surface microcrack node corresponding to the grid cell that occurs perforation failure is marked as high-risk perforation node.The application solves the problem that the prior art cannot effectively simulate and accurately predict the in-situ local punctiform perforation failure caused by microcrack water evaporation concentration.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and more specifically, to a modeling and simulation method and system for weather-resistant sealing strip structures. Background Technology

[0002] In modern industry and automobile manufacturing, weather-resistant sealing strips are widely used in car doors, windows, and building curtain walls, primarily serving crucial functions of waterproofing, sound insulation, and dust prevention. These sealing strips are exposed to complex natural environments for extended periods, needing to withstand the combined challenges of wind, sun, extreme temperature differences, and rain erosion. To assess their service life and optimize cross-sectional design during the research and development phase, computer finite element simulation technology is commonly used in engineering to simulate the mechanical deformation and aging process of sealing strips under complex environments, avoiding time-consuming and labor-intensive physical testing. Existing technologies have disclosed several solutions for computer simulation testing of sealing strips. For example, Chinese patent document CN116305899B discloses a "Method, System, Medium, and Equipment for Obtaining Simulation Parameters of Car Door Sealing Strip Materials." This solution establishes a three-dimensional mesh analysis model of the sealing strip and applies constraint loading to the simulation model according to experimental conditions, thereby calculating and obtaining the simulation parameters and deformation data of the sealing strip under compression.

[0003] However, weatherstripping exhibits a highly destructive yet easily overlooked hidden failure mechanism during actual long-term service. Under repeated pressure on car doors, the surface of the weatherstripping inevitably develops microscopic cracks that are difficult to detect with the naked eye. When exposed to rain or car washes, these microcracks act like sponges, absorbing acidic liquids deep into the surface through capillary action. As the weather clears and temperatures rise, the moisture evaporates rapidly, but the acidic chemical solutes cannot, causing the acid concentration at the bottom of the microcracks to surge dramatically, by tens of times. This extremely high concentration of acid aggressively corrodes the rubber from the inside out, creating point-like perforations and ultimately hollowing out the weatherstripping from the inside. Existing environmental simulation tests and mechanical simulations typically only treat corrosion as a uniform peeling of the surface or can only predict macroscopic elasticity loss, completely failing to capture the malignant physicochemical reactions of moisture evaporation and purification, as well as the internal pitting, occurring within the microcracks. This leads to a significant discrepancy between the predicted service life and the actual localized, sudden perforation and fracture failure. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a modeling and simulation method and system for weather-resistant sealing strip structures, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A modeling and simulation method for weather-resistant sealing strip structures includes the following steps: Construct an initial three-dimensional mesh model of the sealing strip to be analyzed, and assign initial material properties to the initial three-dimensional mesh model; Precipitation simulation and arid environment simulation are applied sequentially to the initial three-dimensional mesh model; Extract the surface microcrack nodes of the initial three-dimensional mesh model that are in an open state under the simulated dry environment conditions; The surface microcrack nodes are input into a pre-trained local corrosion evolution prediction model, and the local corrosion degradation coefficient of the surface microcrack nodes is output by the local corrosion evolution prediction model. Replace the corrosion rate parameter in the initial material properties of the surface microcrack nodes with the local corrosion degradation coefficient; A Markov model is constructed for the surface microcrack nodes, and the local corrosion degradation coefficient is transformed into the state transition probability matrix of the Markov model; The degradation state of the mesh element containing the surface microcrack node is calculated using the Markov model, and the surface microcrack node corresponding to the mesh element that has experienced perforation failure is marked as a high-risk perforation node.

[0006] Specifically, the step of extracting the surface microcrack nodes of the initial three-dimensional mesh model in the open state under the simulated dry environment includes: An assembly compression load is applied to the initial three-dimensional mesh model, and the stress-strain nodes in the initial three-dimensional mesh model are extracted and the deformation data of the stress-strain nodes are obtained. The deformation data is input into a support vector machine classification model, and the crack opening probability of each stress strain node is output by the support vector machine classification model. The stress-strain nodes with a crack opening probability greater than or equal to a preset opening probability threshold are defined as the surface microcrack nodes in an open state. The specific steps of inputting the surface microcrack nodes into the pre-trained local corrosion evolution prediction model include: Extract the node coordinates, stress tensor, and simulation duration under the precipitation simulation conditions of the surface microcrack nodes; The node coordinates, the stress tensor, and the simulation duration are input into the local corrosion evolution prediction model.

[0007] Specifically, the localized corrosion evolution prediction model is a long short-term memory neural network model; the training steps of the localized corrosion evolution prediction model include: Obtain physical test data of historical sealing strip samples under known alternating environments, and extract the actual microscopic observation coordinates of the pitting failure location, physical stress tensor, and physical precipitation duration from the physical test data as training input features; The actual material loss rate of the historical sealing strip samples at the location of pitting failure is extracted as a label for the expected local corrosion degradation coefficient. The long short-term memory neural network model to be trained is subjected to supervised training using the training input features and the expected local corrosion degradation coefficient labels until the loss function converges, thereby obtaining the local corrosion evolution prediction model.

[0008] Specifically, the steps of constructing a Markov model for the surface microcrack nodes, converting the local corrosion degradation coefficient into the state transition probability matrix of the Markov model, calculating the degradation state of the mesh element containing the surface microcrack node using the Markov model, and marking the surface microcrack node corresponding to the mesh element that has experienced perforation failure as a high-risk perforation node specifically include: In the Markov model, a discrete state set of grid cells is defined, which includes an intact state, a pitting state, and a perforation failure state. The local corrosion degradation coefficient is converted into the transition probability between each state in the discrete state set, and the transition probabilities are combined to form the state transition probability matrix. Within each simulation time step, the state transition probability matrix is ​​multiplied with the current state vector of the current simulation time step to obtain the target state vector of the next simulation time step. Extract the probability value representing the mesh cell in the perforation failure state from the target state vector. If the probability value is greater than or equal to a preset failure threshold, determine that the corresponding mesh cell has perforated and mark the surface microcrack node corresponding to the perforated mesh cell as the high-risk perforation node.

[0009] Specifically, after marking the surface microcrack nodes corresponding to the mesh cells that have experienced perforation failure as high-risk perforation nodes, the method further includes: Determine whether the area where the high-risk perforated node is located has permission to modify its internal structure; When the area where the high-risk perforation node is located has the permission to modify the internal structure, extract the spatial coordinates of the high-risk perforation node and the mechanical extrusion frequency parameters of the area where the high-risk perforation node is located. The spatial location coordinates and the mechanical compression frequency parameters are input into the airbag topology generation model. The airbag topology generation model generates a three-dimensional topology structure of a closed airbag cavity in the internal solid mesh region in the normal projection direction of the high-risk perforation node in the initial three-dimensional mesh model. The three-dimensional topology of the sealed airbag cavity is compared with the initial three-dimensional mesh model using a Boolean difference operation to generate the first optimized sealing strip model.

[0010] Specifically, the step of generating a closed airbag cavity three-dimensional topology structure within the internal solid mesh region in the normal projection direction of the high-risk perforation node in the initial three-dimensional mesh model using the airbag topology generation model includes: Obtain the maximum water accumulation volume of the high-risk perforation node under the simulated precipitation conditions; In the airbag topology generation model, a shape matching algorithm is called to match a three-dimensional cavity model that can accommodate the maximum water volume within the internal solid mesh area in the normal projection direction. A three-dimensional topological structure of microporous channels connecting the outer surface is generated between the high-risk perforation node and the three-dimensional model of the cavity. The three-dimensional topological structure of the microporous channels and the three-dimensional model of the cavity are then merged into the three-dimensional topological structure of the sealed airbag cavity.

[0011] Specifically, after generating the first optimized sealing strip model, the method further includes: In the first optimized sealing strip model, the interior of the three-dimensional topological structure of the sealed airbag cavity is defined as a compressible fluid domain; Apply a periodic opening and closing displacement load to the first optimized sealing strip model; Calculate the positive pressure of the internal fluid generated by the compression of the compressible fluid domain under the action of the periodic opening and closing displacement load; Obtain the capillary suction negative pressure generated at the high-risk perforation node under the simulated precipitation conditions; By comparing the positive pressure of the internal fluid with the negative pressure of capillary suction, when the positive pressure of the internal fluid is greater than or equal to the negative pressure of capillary suction, it is determined that the first optimized sealing strip model meets the drainage requirements.

[0012] Specifically, after marking the surface microcrack nodes corresponding to the mesh cells that have experienced perforation failure as high-risk perforation nodes, the method further includes: Determine whether the area where the high-risk perforation node is located has restrictions on external morphological modification; When the region where the high-risk perforation node is located is subject to external topography modification restrictions, the normal vector of the high-risk perforation node on the initial three-dimensional mesh model is extracted. The subsurface modification area is defined by translating a set distance value inward along the normal vector into the sealing strip. Match and obtain the property parameters of porous rubber materials from the material template library; The initial material properties in the sub-surface modification area are replaced with the material property parameters of the porous rubber to generate a second optimized sealing strip model.

[0013] Specifically, the steps for matching and obtaining the material property parameters of porous rubber from the material template library include: Calculate the target diffusion area of ​​the acidic solute at the high-risk perforated node under the simulated dry environment conditions; The capillary permeability parameter of each candidate material attribute record is traversed and queried in the material template library; The capillary permeability parameter is input into the porous diffusion area calculation function, and the porous diffusion area calculation function outputs the pore network diffusion area recorded by each candidate material property. The candidate material property record with the smallest difference between the diffusion area of ​​the pore network and the target diffusion area of ​​the acidic solute is selected, and the capillary permeability parameter corresponding to the selected candidate material property record is used as the property parameter of the porous rubber material.

[0014] This invention also discloses a modeling and simulation system for weather-resistant sealing strip structures, used to implement the above method, comprising: The model building module is used to build an initial three-dimensional mesh model of the sealing strip to be analyzed and to assign initial material properties to the initial three-dimensional mesh model; The boundary application module is used to sequentially apply precipitation simulation conditions and arid environment simulation conditions to the initial three-dimensional mesh model. The node filtering module is used to extract the surface microcrack nodes of the initial three-dimensional mesh model that are in an open state under the simulated dry environment conditions. The coefficient prediction module is used to input the surface microcrack nodes into a pre-trained local corrosion evolution prediction model, and output the local corrosion degradation coefficient of the surface microcrack nodes through the local corrosion evolution prediction model. The perforation determination module is used to replace the corrosion rate parameter in the initial material properties of the surface microcrack node with the local corrosion degradation coefficient; construct a Markov model for the surface microcrack node, and convert the local corrosion degradation coefficient into the state transition probability matrix of the Markov model; calculate the degradation state of the mesh cell where the surface microcrack node is located through the Markov model, and mark the surface microcrack node corresponding to the mesh cell that has experienced perforation failure as a high-risk perforation node.

[0015] The advantages of this invention compared to existing technologies lie in its modeling and simulation method, which breaks through the technical blind spots of traditional macroscopic uniform corrosion simulation and effectively solves the problem that existing technologies cannot accurately simulate and predict the rapid point-like perforation failure of sealing strips caused by local acid concentration under alternating climates. This invention extracts surface microcrack nodes in an open state by sequentially applying precipitation and drying conditions to an initial three-dimensional mesh model, and accurately outputs the local degradation coefficient using a local corrosion evolution prediction model. This invention creatively constructs a Markov model for surface microcrack nodes, transforming the local corrosion degradation coefficient into a state transition probability matrix of the Markov model. This unique technical feature of using the probability matrix for continuous calculation of mesh degradation states avoids the computationally intensive micro-hydrodynamic calculations, reducing the extremely complex micro-physicochemical corrosion process to discrete probabilistic state transition calculations. This not only accurately hollows out high-risk perforation areas caused by evaporation and concentration in the model but also greatly saves the computational burden on the computer, achieving efficient location and early warning detection of hidden micro-failures.

[0016] Building upon this, and addressing the issue that conventionally thickened surfaces increase door closing resistance and fail to actively drain infiltrated liquid after high-risk perforation nodes are discovered in simulations, this invention achieves an active drainage mechanism by generating a three-dimensional topological structure of a closed airbag cavity within the internal solid mesh region along the normal projection direction of the high-risk perforation node in the initial three-dimensional mesh model. This unique technical feature cleverly utilizes the periodic displacement load applied by the daily opening and closing of the car door to compress the compressible fluid domain, thereby generating a microscopic positive pressure from the inside out within the rubber to directly counteract the capillary suction negative pressure, directly blowing the acid water drawn into the microcracks back to the surface before the moisture completely evaporates and concentrates. Simultaneously, for exposed lips where absolute flatness must be maintained and modification of the external morphology is strictly prohibited, this invention further solves the engineering challenge of addressing the issue of extremely concentrated acid at single points that cannot be resolved by external geometric drainage channels. By defining a subsurface modification region along the normal vector inside the sealing strip and replacing the initial material properties of this region with the properties of porous rubber material, this invention constructs a microscopic capillary drainage network inside the sealing strip. This special technical feature utilizes the vast pore network diffusion area of ​​porous materials to instantly draw away the deadly acid water absorbed by surface microcracks and spread it horizontally, successfully diluting the deadly single-point high-concentration pit into a completely tolerable planar slight attenuation in a broad subsurface network, thus resolving the crisis of point perforation without changing any macroscopic shape or overall rigidity of the sealing strip. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the process of this invention; Figure 2 This is a schematic diagram of the initial three-dimensional mesh model of the sealing strip and the structure assigned by its material properties; Figure 3 It is a simulation diagram of alternating environmental conditions and the application of assembly compressive loads in the model; Figure 4 It is a data flow discriminant map that uses a support vector machine classification model to identify the opening state of surface microcracks; Figure 5 This is the input-output architecture diagram of the local corrosion evolution prediction model; Figure 6 It is a map showing the location and marking of high-risk perforation nodes on the surface of a 3D mesh model; Figure 7 This is a diagram showing the topological structure of the enclosed airbag cavity generated under the structural modification permission of this invention. Figure 8 This is a schematic diagram of the distribution of subsurface modification regions under the external morphology modification restrictions of the present invention. Detailed Implementation

[0018] The specific embodiments of the present invention will now be described with reference to the accompanying drawings.

[0019] The embodiments of this invention are provided to help those skilled in the art understand the technical concept, motivation, and implementation path of this invention, and are not intended to limit the scope of protection of this invention. Without departing from the concept of this invention, those skilled in the art can make conventional substitutions to the mesh generation method, material parameter source, model training data organization method, and optimization solution strategy; all such conventional substitutions should be considered to fall within the technical concept of this invention.

[0020] This invention addresses weather-resistant sealing strip structures, particularly those in automobiles, exposed to prolonged conditions of rainfall, sunlight, alternating temperature and humidity, and repeated mechanical compression. Traditional simulations excel at describing macroscopic compression and rebound, overall stress distribution, and uniform aging. However, this invention focuses on a more subtle chain of localized failure: surface microcracks absorb acidic liquids during rainfall, and during drying, the evaporation of moisture causes a rapid concentration of acidic solutes at the crack base, ultimately leading to pitting corrosion and perforation failure. Instead of directly performing a computationally intensive fully coupled solution of micro-fluid chemistry, this invention first identifies the most likely locations of open microcracks in structural simulation. Then, a data-driven model predicts the intensity of localized corrosion at these locations. Finally, a discrete state transition method is used to rapidly deduce the degradation process at the mesh level. This design aims to explicitly locate the microscopic perforation risk, which is typically difficult to incorporate into engineering simulations, within an acceptable computational limit.

[0021] like Figure 1As shown, the overall process of this invention includes initial three-dimensional mesh modeling, application of alternating environment, screening of surface microcrack nodes, prediction of local corrosion degradation coefficient, Markov degradation calculation, and marking of high-risk perforation nodes. Based on this, two types of improved paths that can be directly used for structural optimization are further given.

[0022] In one embodiment, an initial three-dimensional mesh model of the sealing strip to be analyzed is first constructed, and initial material properties are assigned. For example... Figure 2 As shown, a three-dimensional solid model can be established first based on the two-dimensional cross-sectional design drawing of the sealing strip and the actual contour after extrusion molding. Then, the installation section, sealing lip, corner transition section, and functional areas in contact with the door frame or glass are completely preserved. The mesh can be a tetrahedral or hexahedral mesh, with local refinement in surface areas with large structural curvature, concentrated contact deformation, and where microcracks are expected to occur. In actual implementation, the mesh edge length of the general main area can be selected from 0.5mm to 2.0mm, and the mesh edge length of the surface layer near the lip, bend root, and compression contact zone can be selected from 0.05mm to 0.30mm. The reason for this is that if the surface mesh is too coarse, even if the local corrosion trend is identified, it will be difficult to accurately pinpoint the specific danger location in space; if all areas use extremely fine mesh, the computational load will increase significantly. Therefore, this invention adopts a local refinement method rather than a global refinement method.

[0023] When assigning initial material properties, the elastic modulus, Poisson's ratio, density, coefficient of thermal expansion, hygroscopic expansion-related parameters, viscoelastic relaxation parameters, and corrosion rate parameters can be set for each mesh element in the initial 3D mesh model. The corrosion rate parameter characterizes the material's basic loss tendency under acidic liquid conditions. In its initial state, it can be derived from immersion tests, accelerated aging tests, or experimental calibration values ​​in material handbooks for rubber-based formulations. To facilitate subsequent local replacements, a simple approach is to define the initial corrosion rate parameter as a normalized parameter, with a value ranging from 0 to 1. 0 indicates negligible corrosion at the current simulation time step, and 1 indicates the strongest corrosion progression at the current simulation time step. If the actual experimental results are dimensional material loss rates, such as mm / cycle or mg / (mm²·cycle), normalization can be performed using a reference loss rate before writing the result into the initial material property field.

[0024] In further implementation, to provide a candidate range for subsequent crack screening, a surface monitoring layer can be established on the exposed surface of the sealing strip, the surface of the contact compression zone, and the corner transition surface. This surface monitoring layer is not a newly added independent component, but rather a set of nodes marked in the existing surface grid for observing the evolution of microcracks. Since actual surface microcracks usually first appear on the outer layer subjected to repeated compression, springback, and environmental erosion, defining the candidate range first, and then further identifying opening crack nodes within the candidate range, can significantly reduce false positives.

[0025] After completing the initial three-dimensional mesh model and assigning initial material properties, precipitation simulation conditions and arid environment simulation conditions are sequentially applied to the initial three-dimensional mesh model. For example... Figure 3 As shown, the precipitation simulation condition can be used to simulate the process of rainwater or car wash liquid forming a liquid film on the surface and penetrating along surface microcracks. The dry environment simulation condition can be used to simulate the process of post-rain temperature rise, humidity drop, and enhanced airflow followed by water evaporation and solute retention and concentration. In one embodiment, the precipitation simulation condition can apply liquid film coverage boundary conditions, ambient temperature conditions, and contact surface wetting conditions to the outer surface. The dry environment simulation condition can apply temperature rise boundary, humidity decrease boundary, and surface evaporation flux boundary. These boundaries can be applied directly in the multiphysics simulation platform, or the environmental conditions can be first converted into a surface humidity field, temperature field, and evaporation flux field, and then mapped onto the structural model for subsequent analysis. The precipitation duration can be set from 5 min to 180 min, and the drying duration can be set from 30 min to 600 min. The order of precipitation followed by drying is adopted here because this invention focuses not on simple immersion corrosion, but on the alternating process of first absorption, then evaporation, and finally concentration. If the drying stage is missing, the key driving force of solute enrichment at the bottom of the crack cannot be reflected in the simulation logic.

[0026] The surface microcrack nodes mentioned in this invention do not refer to explicitly constructing a geometric crack in the model beforehand. Rather, they refer to surface nodes within the surface monitoring layer whose deformation patterns conform to crack opening characteristics after being subjected to environmental conditions and mechanical loads. In other words, this invention indirectly characterizes the opening and closing state of microcracks through node-level behavior recognition, rather than requiring detailed geometric depiction of all microcracks in the initial modeling stage. This is because the number of microcracks in real sealing strips is large, their morphology is discrete, and their size is extremely small; modeling each microcrack individually would be difficult to implement in engineering.

[0027] In one embodiment, when extracting the surface microcrack nodes of the initial three-dimensional mesh model in the open state under the simulated dry environment condition, after completing the simulation of precipitation and dry environment conditions, an assembly compression load can be further applied to the initial three-dimensional mesh model, and the stress-strain nodes and their deformation data in the model can be extracted. The assembly compression load is analyzed after the wet-dry conditions to better reflect the actual usage state, i.e., the sealing strip remains in a pre-compression state while experiencing environmental changes. The assembly compression load is used to simulate the actual compression conditions experienced by the sealing strip after installation on the door, window frame, or other assembled components, because the sealing strip is not freely placed during actual service but is subjected to long-term pre-compression and periodic opening and closing disturbances. Assembly compression can be achieved by displacement loading, and the compression amount can be applied at 10% to 45% of the cross-sectional thickness, preferably a value consistent with the actual assembly compression rate. Stress-strain nodes can be determined jointly by the equivalent strain threshold, contact area determination, and surface monitoring layer screening. For example, nodes with an equivalent strain greater than 0.02 within the surface monitoring layer can be used as stress-strain nodes. Deformation data may include at least nodal displacement vectors, normal displacement components, maximum principal strain, minimum principal strain, equivalent strain, and changes in the distance between adjacent nodes.

[0028] The reason for not directly using a single strain threshold to determine whether a crack has opened is that some nodes, although with high strain, are in a compressive closed state; while other nodes, although not exhibiting extreme total strain, show significant local normal opening, which better matches the characteristics of a microcrack's liquid absorption inlet. Therefore, this invention further inputs the deformation data into a support vector machine classification model, which outputs the crack opening probability for each stressed strain node. Figure 4 As shown, in one embodiment, the support vector machine (SVM) classification model can adopt a radial basis function kernel function. Input features include node normal opening displacement, tangential displacement, maximum principal strain, principal stress direction change, and node local curvature change. The output is the probability value of a node belonging to an open crack state. To obtain the probability output, probability mapping can be superimposed on the original SVM classification results. The training samples for this SVM classification model can come from open and closed crack nodes labeled by microscopy, CT slices, or digital image correlation in historical sealing strip compression tests. During model training, the input features of each dimension can be standardized first, and then the penalty parameter C and kernel function parameter gamma can be determined through cross-validation. C can be selected from 1 to 100, and gamma can be selected from 0.01 to 1. The advantage of using SVM instead of simple threshold discrimination is that SVM is more stable in classifying small to medium-sized samples and nonlinear boundaries, making it suitable for engineering scenarios with a large number of nodes but a relatively limited number of labeled samples.

[0029] After obtaining the crack opening probability of each stress-strain node, stress-strain nodes with a crack opening probability greater than or equal to a preset opening probability threshold are identified as surface microcrack nodes in the opening state. The preset opening probability threshold can be selected based on the false positive and false negative tolerance; in one embodiment, it can be set to 0.60 to 0.85, preferably 0.70. If the threshold is too low, a large number of nodes that have deformed but have not actually formed liquid absorption channels will be included in the subsequent corrosion analysis; if the threshold is too high, some dangerous nodes in the early opening state may be missed. Therefore, it is usually determined in conjunction with the F1 score or recall rate on the validation set.

[0030] After identifying the surface microcrack nodes in the open state, these nodes are input into a pre-trained local corrosion evolution prediction model. For example... Figure 5 As shown, the localized corrosion evolution prediction model described in this invention is a long short-term memory (LSTM) neural network model. The reason for using the LSTM neural network model is that localized pitting corrosion is not solely determined by its spatial location or stress state at a single moment, but rather has a continuous temporal relationship with the duration of wetting, drying, residual stress from previous cycles, and the accumulation of localized damage. The LSTM neural network model is suitable for handling this historical dependency problem that unfolds over time.

[0031] In one embodiment, a time-series input is constructed for each surface microcrack node, with each time step corresponding to a sampling moment or sub-stage in a wet-dry alternation cycle. That is, the same node will repeatedly have its node coordinates, stress tensor, and simulation duration under precipitation simulation conditions collected across multiple cycles. The node coordinates typically remain constant, while the stress tensor and simulation duration change with the evolution of the conditions. The node coordinates reflect the spatial location of the node within the cross-section; different locations may correspond to different drainage conditions, stress concentration levels, and external erosion intensities. The stress tensor reflects the local mechanical state, as prolonged tension or complex shear-tension coupling at the crack base increases the likelihood of liquid retention and material failure. The simulation duration under precipitation simulation conditions reflects the time basis for the acidic liquid's residence and penetration within the crack; the longer the residence time, the greater the likelihood of severe pitting corrosion after subsequent evaporation and concentration. In a simplified implementation, the stress tensor can be input as six independent components: σxx, σyy, σzz, τxy, τxz, and τyz; the nodal coordinates can be input as x, y, and z; and the simulation duration can be input as the duration of precipitation in the current cycle or the cumulative precipitation time. Thus, the input dimension for each time step can be 10 dimensions.

[0032] The training steps of the Long Short-Term Memory Neural Network model can be implemented as follows: First, acquire physical test data of historical sealing strip samples under known alternating environments. These historical samples can come from accelerated environment chamber tests, repeated compression fatigue tests, and acidic liquid immersion evaporation coupling tests. For each location where pitting failure occurs, extract its actual microscopic observation coordinates, physical stress tensor, and physical precipitation duration as training input features. The actual microscopic observation coordinates can be determined by using a microscope, a 3D profilometer, or micro-CT to determine the spatial position of the pitting pit center on the sample surface; the physical stress tensor can be obtained by combining digital image correlation measurement results with an inverted finite element model, or it can be calculated from the local strain field of a synchronous loading test; the physical precipitation duration can be directly recorded by the test control program. Subsequently, extract the actual material loss rate of the historical sealing strip samples at the pitting failure location as a label for the expected local corrosion degradation coefficient. The actual material loss rate can be defined as the local depth loss per unit cycle, the mass loss per unit area, or expressed as a loss ratio relative to the original material thickness. To facilitate integration with subsequent Markov models, a simple approach is to further normalize the actual material loss rate to a localized corrosion degradation coefficient label ranging from 0 to 1. A normalization method could be d=min(1,r / r). ref ), where r is the actual material loss rate, r ref The reference loss rate represents a value sufficient to rapidly advance the material from its current state to a next degraded state within a simulation time step. After this treatment, the closer the localized corrosion degradation coefficient d is to 1, the more likely that node is to experience pitting corrosion propagation or even perforation in subsequent time steps. The reference loss rate can be obtained statistically from the critical material loss rate in historical samples that transitions from an intact state to a state of obvious pitting corrosion, or it can be determined from the material loss rate corresponding to the first appearance of an identifiable pit in accelerated corrosion tests.

[0033] In terms of network structure, one embodiment can employ a two-layer Long Short-Term Memory (LSTM) network superimposed with a one-layer fully connected output layer. The number of LSM units in the first layer can be set to 32, and the number of LSM units in the second layer can be set to 16. The fully connected layer outputs a scalar, namely the local corrosion degradation coefficient. If the output is a normalized coefficient, a Sigmoid activation function can be used in the output layer; if the output is a dimensional material loss rate, a linear activation function can be used in the output layer. During training, mean squared error can be used as the loss function, the optimizer can be Adam, the initial learning rate can be set to 0.001, the batch size can be set to 16 to 64, and the maximum number of training epochs can be set to 100 to 300. When the loss function on the validation set decreases by less than 0.001 within 10 consecutive epochs, the loss function is considered to have converged, and the local corrosion evolution prediction model is obtained. To avoid interference from different dimensional inputs during training, the node coordinates, stress tensor components, and simulation duration can be normalized separately.

[0034] After training, the surface microcrack nodes selected in the current simulation are input into the local corrosion evolution prediction model to obtain the local corrosion degradation coefficient for each node. This local corrosion degradation coefficient essentially quantifies the risk of local pitting corrosion at the node under the current mechanical state and environmental history into a value that can directly participate in degradation calculations. It is not a separate environmental parameter or a separate mechanical parameter, but rather a model output value because it incorporates the combined effects of location, stress, and time. After obtaining this coefficient, the corrosion rate parameter in the initial material properties of the surface microcrack node is replaced with the local corrosion degradation coefficient. For finite element solvers that typically store material parameters per element, the corrosion rate parameter of the mesh element containing the surface microcrack node can be updated to the average of the local corrosion degradation coefficients of all microcrack nodes within that element. If an element contains only one surface microcrack node, the local corrosion degradation coefficient corresponding to that node is directly used.

[0035] Based on this, a Markov model is constructed for the surface microcrack nodes, and the local corrosion degradation coefficient is transformed into the state transition probability matrix of the Markov model. The purpose of introducing the Markov model here is to simplify the continuous and complex micro-corrosion evolution process into a probabilistic transition process between a finite number of discrete states. In this way, it is not necessary to solve the complex micro-acid migration and reaction equations at each time step, nor is it necessary to model each micro-pit separately. Instead, a state evolution method more suitable for batch engineering calculations is used to predict which elements will enter perforation failure.

[0036] In one embodiment, the discrete state set of the mesh element in the Markov model is defined as an intact state, a pitting state, and a perforation failure state. An intact state indicates that although the element may be near an open crack, it has not yet experienced substantial localized pitting corrosion; a pitting state indicates that the element has experienced localized corrosion damage, resulting in a decrease in the localized load-bearing capacity of the material, but has not yet formed a through-hole; a perforation failure state indicates that the element has formed penetrating damage that can lead to sealing failure. To make the implementation as simple and direct as possible, this invention provides a very simple quantitative conversion method. First, ensure that the value of the localized corrosion degradation coefficient d falls between 0 and 1. If the localized corrosion evolution prediction model outputs the actual material loss rate r, then first calculate d = min(1, r / r). refNormalization is performed, where min is a function that takes the smaller value; if the model output itself is a coefficient between 0 and 1, it is used directly. Then, this local corrosion degradation coefficient can be directly used as the state transition probability, that is, let the transition probability from the intact state to the pitting state be d, the transition probability from the pitting state to the perforation failure state also be d, and the corresponding probability of maintaining the intact state is 1-d, the probability of maintaining the pitting state is 1-d, and the probability of maintaining the perforation failure state is 1. Therefore, for a certain mesh element, its state transition probability matrix can be expressed as: [1-d,d,0 0,1-d,d 0,0,1] The sum of all elements in each row of the above matrix is ​​1, satisfying the basic requirements of a state transition probability matrix. While this transformation method is simple, it meets the requirements of quantitative calculation and engineering feasibility of this invention. Its meaning is very intuitive: the larger the local corrosion degradation coefficient at a certain location, the greater the probability that the location will advance from the current state to a more severe state within a time step. No additional complex correction terms or extra factors are introduced here; instead, the same coefficient directly drives the state forward, facilitating rapid deployment in engineering software. If it is desired to further improve the fitting accuracy, the probabilities of transitioning from the intact state to the pitting state and from the pitting state to the perforation failure state can be set to different values, but this is not a necessary condition for realizing this invention.

[0037] Within each simulation time step, the state transition probability matrix is ​​multiplied with the current state vector of the current simulation time step to obtain the target state vector for the next simulation time step. If a cell is initially in an intact state, its current state vector can be written as [1,0,0]; if it has entered the pitting state, it can be written as [0,1,0]. The simulation time step can be set according to a wet-dry cycle, or according to a 1-day, 3-day, or 1-time on-off cycle. The actual selection can be consistent with the test cycle or lifetime conversion method. After matrix multiplication, the third component in the target state vector is the probability value of the grid cell being in the perforation failure state. When this probability value is greater than or equal to the preset failure threshold, the corresponding grid cell is determined to have perforation failure. The preset failure threshold can be selected from 0.70 to 0.95, preferably 0.80. Setting the threshold higher is to ensure that the marked danger areas have a high degree of failure certainty.

[0038] After completing the above determination, the surface microcrack nodes corresponding to the mesh elements that have experienced perforation failure are marked as high-risk perforation nodes. For example... Figure 6As shown, high-risk perforation nodes can be output on the surface of the 3D mesh model using color markings, node number markings, or risk level heatmaps. The term "high-risk perforation node" does not refer to a single node that has already formed a geometric perforation, but rather to a node whose surface entry path and the degradation results of its constituent elements indicate that it is most likely to become a future perforation source. By mapping high-risk locations back to the surface node level, designers can directly see the distribution of dangerous points without having to manually search through a large number of mesh elements.

[0039] like Figure 7 As shown, in a further embodiment, after marking the surface microcrack nodes corresponding to the mesh elements that have experienced perforation failure as high-risk perforation nodes, the system can automatically enter the internal drainage optimization branch based on structural design constraints. First, it is determined whether the area where the high-risk perforation node is located has internal structure modification permissions. Here, internal structure modification permissions mean that within this area, the internal solid mesh structure of the sealing strip is allowed to be changed, but the outer surface sealing fit contour is not allowed to be changed, the contact interface with the assembled object is not affected, and the assembly datum is not damaged. This permission can be pre-written into the model by design rules, or it can be marked by CAD area attributes or manually specified. If this permission is available, the spatial coordinates of the high-risk perforation node and the mechanical compression frequency parameters of the area where the high-risk perforation node is located are extracted. The mechanical compression frequency parameters can be obtained statistically from the door opening and closing frequency, the glass raising and lowering frequency, or other periodic compression actions, or they can be preset by simulation conditions. The reason for extracting this parameter is that the internal drainage cavity is not simply made into an empty cavity, but rather aims to use the periodic compression during daily opening and closing to treat the cavity as a miniature pump cavity; therefore, the compression frequency directly affects the cavity response and drainage effect.

[0040] Subsequently, the spatial coordinates and the mechanical compression frequency parameters are input into the airbag topology generation model. This model then generates a three-dimensional topology of a closed airbag cavity within the internal solid mesh region corresponding to the high-risk perforation node along the normal projection direction in the initial three-dimensional mesh model. Here, the normal projection direction refers to the direction in which the normal to the high-risk perforation node extends inwards towards the sealing strip. Since acid water typically infiltrates from surface cracks, arranging the cavity along the internal path corresponding to the normal direction is most beneficial for establishing a reverse drainage path from the inside out. In a simple implementation, the airbag topology generation model may not be a complex neural network, but rather a parametric geometry generator that automatically generates ellipsoidal, teardrop-shaped, or flattened cavity shapes based on the high-risk node location, the available internal solid space, and the mechanical compression frequency. If the mechanical compression frequency is high, smaller cavities with faster rebound can be preferentially generated; if the mechanical compression frequency is low but the displacement is large each time, larger cavities can be generated.

[0041] Furthermore, when generating the three-dimensional topological structure of the closed airbag cavity within the internal solid mesh region using the airbag topology generation model, the process also includes obtaining the maximum water accumulation volume of the high-risk perforation node under the simulated precipitation conditions. This maximum water accumulation volume can be estimated from the liquid-occupied volume of the crack channel and surface stagnant zone near the node during the precipitation simulation stage, or approximated by the product of the local liquid film thickness and the effective area of ​​the crack opening. The purpose of obtaining this maximum water accumulation volume is to ensure that the subsequently generated cavity is not of arbitrary size, but rather has at least sufficient drainage capacity and backflow capability. Then, a morphological matching algorithm is invoked in the airbag topology generation model to match a three-dimensional cavity model capable of accommodating the maximum water accumulation volume within the internal solid mesh region corresponding to the normal projection direction. The morphological matching algorithm can be implemented as follows: First, multiple candidate cavity models are read from a preset cavity template library. These candidate cavity models have different volumes, aspect ratios, and wall thickness constraints. Then, the spatial overlap between each candidate cavity model and the internal solid mesh region, as well as the difference between each candidate cavity model and the target volume, are calculated. Finally, the candidate cavity model with the smallest volume difference, without exceeding the internal solid mesh boundary, is selected as the matching result. Subsequently, a three-dimensional topology of microporous channels connecting the outer surface is generated between the high-risk perforation node and the cavity 3D model. The three-dimensional topology of the microporous channels and the cavity 3D model are then merged into the three-dimensional topology of the sealed airbag cavity. The function of the microporous channels is to establish a pressure transmission and liquid backflow path between the internal cavity and the high-risk crack inlet. The equivalent diameter of the microporous channels can be selected from 0.05mm to 0.50mm. If the diameter is too small, the channel resistance will be too large; if the diameter is too large, it may affect the local mechanical strength and the integrity of the sealing surface.

[0042] After generating the three-dimensional topology of the sealed airbag cavity, a Boolean difference operation is performed between it and the initial three-dimensional mesh model to generate the first optimized sealing strip model. After generating the first optimized sealing strip model, the interior of the sealed airbag cavity is defined as a compressible fluid domain within this model, and then a periodic opening and closing displacement load is applied to the model. This periodic opening and closing displacement load can be directly obtained from the compression displacement curve experienced by the sealing strip when a car door is repeatedly closed and opened. The medium within the compressible fluid domain can be treated as air, using an ideal gas approximation, without the need for a complex multiphase flow model. Subsequently, the positive pressure of the internal fluid generated by the compression of the compressible fluid domain under the periodic opening and closing displacement load is calculated, and the capillary suction negative pressure generated at the high-risk perforation node under simulated precipitation conditions is obtained. If the crack geometry is not explicitly established in the model, the normal opening displacement at the high-risk perforation node and the change in distance between adjacent nodes can be converted into the equivalent crack radius, and then the capillary suction negative pressure can be calculated by combining the liquid surface tension and contact angle; alternatively, it can be obtained through experimental calibration and table lookup. By comparing the positive pressure of the internal fluid with the negative pressure of capillary suction, when the positive pressure of the internal fluid is greater than or equal to the negative pressure of capillary suction, the first optimized sealing strip model is determined to meet the drainage requirements. The essential purpose of this design is to utilize the originally unavoidable opening and closing compression behavior to create a micro-pump effect from the inside out, pushing out the acidic liquid that has been sucked into the crack as early as possible, thus preventing it from concentrating during the subsequent drying process.

[0043] like Figure 8 As shown, in another embodiment, for areas where an absolutely smooth external morphology must be ensured, such as exposed lips, visible mating edges, or surface areas directly adjacent to exterior components, it is not permissible to solve the problem by adding microporous drainage channels or changing the external contour. In this case, after marking the surface microcrack nodes corresponding to the mesh cells where perforation failure occurs as high-risk perforation nodes, it can be determined whether the area where the high-risk perforation node is located has external morphology modification restrictions. Here, external morphology modification restrictions refer to the requirement that the outer surface contour, surface roughness, edge smoothing transition, and mating shape of the area must not be changed. This restriction can also be given in advance by design rules, appearance area markings, or assembly constraint documents. If this restriction exists, the normal vector of the high-risk perforation node on the initial three-dimensional mesh model is extracted, and a set distance is translated along the normal vector into the interior of the sealing strip to delineate the secondary surface modification area. This set distance value can be selected from 0.2mm to 1.5mm, preferably from 0.3mm to 0.8mm. The reason why the modification is not done directly on the surface, but rather the subsurface modification region is formed by translating inward, is that the present invention aims to construct an internal buffer layer below the surface layer that can laterally diffuse and disperse acidic solutes without changing the appearance and contact geometry.

[0044] After defining the subsurface modification region, the material properties of porous rubber are matched and obtained from the material template library. The initial material properties within the subsurface modification region are then replaced with these porous rubber material properties to generate a second optimized sealing strip model. The porous rubber material properties mentioned here can include at least porosity, capillary permeability, equivalent elastic modulus, and local compression rebound parameters. Capillary permeability is the parameter of most interest in this branch of the invention because it determines whether acidic liquid drawn into the surface crack can quickly diffuse and spread throughout the porous subsurface network, rather than remaining concentrated at a single point. The records of the various candidate material properties in the material template library can be obtained from experimental databases of different foamed rubber formulations and samples with different microporous structures. The porosity can be selected from 0.10 to 0.50, the capillary permeability can be pre-calibrated through a specialized permeation test, and the equivalent elastic modulus should ensure that the replacement does not cause a significant abnormality in the overall stiffness of the sealing strip.

[0045] When matching and obtaining the property parameters of porous rubber materials from the material template library, in one embodiment, the target diffusion area of ​​acidic solute at the high-risk perforation node under the simulated dry environment conditions is first calculated. Here, the target diffusion area refers to the minimum diffusion coverage area required to dilute the residual acidic solute near the high-risk node from a single-point high-concentration state to below a safe concentration. This target diffusion area can be obtained by simplifying the diffusion solution, that is, using the equivalent acidic solute mass remaining near the high-risk node during the drying stage after precipitation as the initial condition, establishing a local diffusion plane below the surface, and solving for the minimum diffusion area required to be covered when the acid load per unit area is reduced to a safe value. The safe value can be determined based on the material's basic corrosion test, for example, selecting the acid load per unit area below which no significant pitting corrosion expansion will occur within a preset lifespan. The advantage of this definition is that the target diffusion area directly corresponds to the engineering objective of this invention, namely, diffusing the small-area high-concentration corrosion source originally concentrated at the bottom of the microcrack into a larger area but lower-risk distribution. Subsequently, the capillary permeability parameters of each candidate material attribute record are traversed and queried in the material template library. These parameters are then input into a porous diffusion area calculation function, which outputs the pore network diffusion area of ​​each candidate material attribute record. This pore diffusion area calculation function can be a pre-calibrated monotonic function or a simplified seepage diffusion model. In a simple implementation, offline experiments can be conducted on each candidate material within a fixed time window, recording its lateral diffusion area under the same liquid volume conditions, and establishing a correspondence table between capillary permeability and diffusion area. When used online, only table lookup or interpolation is needed, eliminating the need for complex real-time simulations. Finally, the candidate material attribute record with the smallest difference between the pore network diffusion area and the target diffusion area of ​​the acidic solute is selected, and the corresponding capillary permeability parameter is used as the attribute parameter of the porous loose rubber material. The essential purpose of this matching is to ensure that the subsurface porous network neither suffers from insufficient diffusion, thus retaining the risk of high concentration at single points, nor from excessive diffusion leading to excessive structural softening.

[0046] It should be understood that the first and second optimized sealing strip models mentioned above are not mutually exclusive. For areas with internal structural modification permissions and no external morphological modification restrictions, the active drainage path with closed airbag cavities can be prioritized; for areas with strict appearance constraints but where internal material layer adjustments are still possible, a subsurface porous and loose rubber modification path can be adopted; for different hazardous areas, different optimization methods can also be used at different positions on the same sealing strip.

[0047] This invention also discloses a modeling and simulation system for weather-resistant sealing strip structures to implement the above-mentioned method. This system can be deployed on an engineering workstation or server and consists of a model building module, a boundary application module, a node selection module, a coefficient prediction module, and a perforation determination module. The model building module reads the geometric model of the sealing strip and completes mesh generation and initial material property assignment; the boundary application module sequentially applies simulated precipitation conditions, simulated dry environment conditions, and assembly compression loads; the node selection module calls a support vector machine classification model based on the deformation data of the stress-strain nodes to output surface microcrack nodes; the coefficient prediction module calls a local corrosion evolution prediction model to output the local corrosion degradation coefficient of each surface microcrack node; the perforation determination module writes the local corrosion degradation coefficient into the material field of the mesh element, establishes a state transition probability matrix, and progressively calculates the degradation state of the mesh element through a Markov model, finally outputting the distribution results of high-risk perforation nodes. Furthermore, the system may also include a structural optimization submodule, which automatically calls the airbag topology generation model or material template library matching logic based on internal structure modification permissions and external morphology modification restrictions, thereby generating a first optimized sealing strip model or a second optimized sealing strip model.

[0048] In a complete engineering implementation process, the designer first imports the cross-sectional model of the sealing strip to be analyzed, and the system automatically completes the three-dimensional tensile modeling and local fine mesh generation; then, the initial material properties in the material database are read and assigned values; subsequently, several rounds of alternating precipitation and drying boundaries are applied according to preset life conditions or accelerated conditions, while assembly compression loads are applied simultaneously; next, the support vector machine classification model identifies surface microcrack nodes, and the long short-term memory neural network model outputs the local corrosion degradation coefficient, which is then replaced in the corrosion rate parameters of the corresponding mesh element; then, a three-state Markov model is established and advanced step by step over time, outputting mesh elements whose perforation failure probability exceeds the failure threshold; finally, the surface microcrack nodes corresponding to these mesh elements are marked as high-risk perforation nodes and enter the corresponding optimization branch. Through this entire process, this invention achieves an integrated modeling and simulation closed loop from alternating environment input, local hazard identification, degradation inference to structural improvement suggestion output.

[0049] The above embodiments have provided a simple, clear, and directly programmable solution for the source of the localized corrosion degradation coefficient and its transformation into the state transition probability matrix. For those skilled in the art, even without introducing additional complex mechanistic models, efficient prediction and location of the risk of localized perforation failure of weather-resistant sealing strips can be achieved solely through node-level temporal input, the localized corrosion degradation coefficient output by the long short-term memory neural network model, and the three-state Markov state transition matrix. If further accuracy improvements are needed, the feature set of the support vector machine, the number of layers in the long short-term memory neural network, the time step division method, and the size of the cavity template library or material template library can be expanded without changing the overall concept of the invention.

Claims

1. A modeling and simulation method for a weather-resistant sealing strip structure, characterized in that, Includes the following steps: Construct an initial three-dimensional mesh model of the sealing strip to be analyzed, and assign initial material properties to the initial three-dimensional mesh model; Precipitation simulation and arid environment simulation are applied sequentially to the initial three-dimensional mesh model; Extract the surface microcrack nodes of the initial three-dimensional mesh model that are in an open state under the simulated dry environment conditions; The surface microcrack nodes are input into a pre-trained local corrosion evolution prediction model, and the local corrosion degradation coefficient of the surface microcrack nodes is output by the local corrosion evolution prediction model. Replace the corrosion rate parameter in the initial material properties of the surface microcrack nodes with the local corrosion degradation coefficient; A Markov model is constructed for the surface microcrack nodes, and the local corrosion degradation coefficient is transformed into the state transition probability matrix of the Markov model; The degradation state of the mesh element containing the surface microcrack node is calculated using the Markov model, and the surface microcrack node corresponding to the mesh element that has experienced perforation failure is marked as a high-risk perforation node.

2. The modeling and simulation method for weather-resistant sealing strip structures according to claim 1, characterized in that, The specific steps for extracting the surface microcrack nodes of the initial 3D mesh model in the open state under the simulated dry environment conditions include: An assembly compression load is applied to the initial three-dimensional mesh model, and the stress-strain nodes in the initial three-dimensional mesh model are extracted and the deformation data of the stress-strain nodes are obtained. The deformation data is input into a support vector machine classification model, and the crack opening probability of each stress strain node is output by the support vector machine classification model. The stress-strain nodes with a crack opening probability greater than or equal to a preset opening probability threshold are defined as the surface microcrack nodes in an open state. The specific steps of inputting the surface microcrack nodes into the pre-trained local corrosion evolution prediction model include: Extract the node coordinates, stress tensor, and simulation duration under the precipitation simulation conditions of the surface microcrack nodes; The node coordinates, the stress tensor, and the simulation duration are input into the local corrosion evolution prediction model.

3. The modeling and simulation method for the weather-resistant sealing strip structure according to claim 1, characterized in that, The localized corrosion evolution prediction model is a long short-term memory neural network model; the training steps of the localized corrosion evolution prediction model include: Obtain physical test data of historical sealing strip samples under known alternating environments, and extract the actual microscopic observation coordinates of the pitting failure location, physical stress tensor, and physical precipitation duration from the physical test data as training input features; The actual material loss rate of the historical sealing strip samples at the location of pitting failure is extracted as a label for the expected local corrosion degradation coefficient. The long short-term memory neural network model to be trained is subjected to supervised training using the training input features and the expected local corrosion degradation coefficient labels until the loss function converges, thereby obtaining the local corrosion evolution prediction model.

4. The modeling and simulation method for the weather-resistant sealing strip structure according to claim 1, characterized in that, The steps of constructing a Markov model for the surface microcrack nodes, converting the local corrosion degradation coefficient into the state transition probability matrix of the Markov model, calculating the degradation state of the mesh element containing the surface microcrack node using the Markov model, and marking the surface microcrack node corresponding to the mesh element that has experienced perforation failure as a high-risk perforation node specifically include: In the Markov model, a discrete state set of grid cells is defined, which includes an intact state, a pitting state, and a perforation failure state. The local corrosion degradation coefficient is converted into the transition probability between each state in the discrete state set, and the transition probabilities are combined to form the state transition probability matrix. Within each simulation time step, the state transition probability matrix is ​​multiplied with the current state vector of the current simulation time step to obtain the target state vector of the next simulation time step. Extract the probability value representing the mesh cell in the perforation failure state from the target state vector. If the probability value is greater than or equal to a preset failure threshold, determine that the corresponding mesh cell has perforated and mark the surface microcrack node corresponding to the perforated mesh cell as the high-risk perforation node.

5. The modeling and simulation method for the weather-resistant sealing strip structure according to claim 1, characterized in that, After marking the surface microcrack nodes corresponding to the mesh elements that have experienced perforation failure as high-risk perforation nodes, the method further includes: Determine whether the area where the high-risk perforated node is located has permission to modify its internal structure; When the area where the high-risk perforation node is located has the permission to modify the internal structure, extract the spatial coordinates of the high-risk perforation node and the mechanical extrusion frequency parameters of the area where the high-risk perforation node is located. The spatial location coordinates and the mechanical compression frequency parameters are input into the airbag topology generation model. The airbag topology generation model generates a three-dimensional topology structure of a closed airbag cavity in the internal solid mesh region in the normal projection direction of the high-risk perforation node in the initial three-dimensional mesh model. The three-dimensional topology of the sealed airbag cavity is compared with the initial three-dimensional mesh model using a Boolean difference operation to generate the first optimized sealing strip model.

6. The modeling and simulation method for the weather-resistant sealing strip structure according to claim 5, characterized in that, The steps of generating a closed airbag cavity three-dimensional topology structure within the internal solid mesh region of the high-risk perforation node in the normal projection direction of the airbag topology generation model in the initial three-dimensional mesh model specifically include: Obtain the maximum water accumulation volume of the high-risk perforation node under the simulated precipitation conditions; In the airbag topology generation model, a shape matching algorithm is called to match a three-dimensional cavity model that can accommodate the maximum water volume within the internal solid mesh area in the normal projection direction. A three-dimensional topological structure of microporous channels connecting the outer surface is generated between the high-risk perforation node and the three-dimensional model of the cavity. The three-dimensional topological structure of the microporous channels and the three-dimensional model of the cavity are then merged into the three-dimensional topological structure of the sealed airbag cavity.

7. The modeling and simulation method for the weather-resistant sealing strip structure according to claim 5, characterized in that, After generating the first optimized sealing strip model, the method further includes: In the first optimized sealing strip model, the interior of the three-dimensional topological structure of the sealed airbag cavity is defined as a compressible fluid domain; Apply a periodic opening and closing displacement load to the first optimized sealing strip model; Calculate the positive pressure of the internal fluid generated by the compression of the compressible fluid domain under the action of the periodic opening and closing displacement load; Obtain the capillary suction negative pressure generated at the high-risk perforation node under the simulated precipitation conditions; By comparing the positive pressure of the internal fluid with the negative pressure of capillary suction, when the positive pressure of the internal fluid is greater than or equal to the negative pressure of capillary suction, it is determined that the first optimized sealing strip model meets the drainage requirements.

8. The modeling and simulation method for the weather-resistant sealing strip structure according to claim 1, characterized in that, After marking the surface microcrack nodes corresponding to the mesh elements that have experienced perforation failure as high-risk perforation nodes, the method further includes: Determine whether the area where the high-risk perforation node is located has restrictions on external morphological modification; When the region where the high-risk perforation node is located is subject to external topography modification restrictions, the normal vector of the high-risk perforation node on the initial three-dimensional mesh model is extracted. The subsurface modification area is defined by translating a set distance value inward along the normal vector into the sealing strip. Match and obtain the property parameters of porous rubber materials from the material template library; The initial material properties in the sub-surface modification area are replaced with the material property parameters of the porous rubber to generate a second optimized sealing strip model.

9. The modeling and simulation method for the weather-resistant sealing strip structure according to claim 8, characterized in that, The specific steps for matching and obtaining the material property parameters of porous rubber from the material template library include: Calculate the target diffusion area of ​​the acidic solute at the high-risk perforated node under the simulated dry environment conditions; The capillary permeability parameter of each candidate material attribute record is traversed and queried in the material template library; The capillary permeability parameter is input into the porous diffusion area calculation function, and the porous diffusion area calculation function outputs the pore network diffusion area recorded by each candidate material property. The candidate material property record with the smallest difference between the diffusion area of ​​the pore network and the target diffusion area of ​​the acidic solute is selected, and the capillary permeability parameter corresponding to the selected candidate material property record is used as the property parameter of the porous rubber material.

10. A modeling and simulation system for implementing the weather-resistant sealing strip structure according to claim 1, characterized in that, include: The model building module is used to build an initial three-dimensional mesh model of the sealing strip to be analyzed and to assign initial material properties to the initial three-dimensional mesh model; The boundary application module is used to sequentially apply precipitation simulation conditions and arid environment simulation conditions to the initial three-dimensional mesh model. The node filtering module is used to extract the surface microcrack nodes of the initial three-dimensional mesh model that are in an open state under the simulated dry environment conditions. The coefficient prediction module is used to input the surface microcrack nodes into a pre-trained local corrosion evolution prediction model, and output the local corrosion degradation coefficient of the surface microcrack nodes through the local corrosion evolution prediction model. The perforation determination module is used to replace the corrosion rate parameter in the initial material properties of the surface microcrack node with the local corrosion degradation coefficient; construct a Markov model for the surface microcrack node, and convert the local corrosion degradation coefficient into the state transition probability matrix of the Markov model; calculate the degradation state of the mesh cell where the surface microcrack node is located through the Markov model, and mark the surface microcrack node corresponding to the mesh cell that has experienced perforation failure as a high-risk perforation node.

Images