Platform door optimization method and system and storage medium

Abstract

The invention provides a platform door optimization method and system and a storage medium, and the method comprises the steps: obtaining a pneumatic pressure change curve of a basic platform door under a train station-crossing working condition according to a pre-trained pneumatic effect calculation model, performing modal analysis, statics structural strength analysis and fatigue strength analysis on the basic platform door according to the pneumatic pressure change curve, and then converting the basic platform door structure into density variable distribution through a variable density method; the target platform door structure is obtained by iteratively updating density variable distribution according to a preset stress constraint condition by taking structural lightweight, rigidity maximization and fatigue life maximization as targets through a genetic algorithm; and target material performance parameters of the target platform door structure are selected in a preset material database through a Bayesian optimization algorithm and a finite element analysis result under the minimum cost constraint condition.

Description

Technical Field

[0001] The present invention relates to the technical field of platform door structures, and in particular, to a method and system for optimizing platform doors and a storage medium. Background Art

[0002] With the rapid development of science and technology in the railway industry, there have been rich changes in the operating speeds and operating organizations of intercity railways, suburban railways, and high-speed railways. As a result, relatively severe aerodynamic effects such as high-speed train passing through stations have had a significant impact on the surrounding environment. This poses potential safety risks to the safety of platform structures, waiting and duty personnel, and ground facilities. Drawing on similar treatment measures in the subway, platform doors are installed.

[0003] In the prior art, due to the relatively low operating speed of urban subways, the relatively unified types of vehicle models, stations, etc., the aerodynamic effects generated are relatively consistent, and there are corresponding standards to constrain the corresponding platform door structure forms; for intercity railways, suburban railways, high-speed railways, etc., due to the large number of vehicle models, formation types, station forms, and train operating speed conditions, it is difficult to determine the installation positions of platform doors and the requirements for the structural strength of platform doors. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide a method and system for optimizing platform doors and a storage medium to eliminate or improve one or more defects existing in the prior art and solve the problem that it is difficult to optimize platform doors for different lines in the prior art.

[0005] One aspect of the present invention provides a method for optimizing platform doors, the method comprising the following steps:

[0006] Obtain the aerodynamic pressure change curve of the basic platform door under the condition of train passing through the station according to a pre-trained aerodynamic effect calculation model;

[0007] After performing modal analysis, static structural strength analysis, and fatigue strength analysis on the basic platform door according to the aerodynamic pressure change curve, convert the basic platform door structure into a density variable distribution by the variable density method, and use the genetic algorithm to iteratively update the density variable distribution with the goals of structural lightweighting, maximization of stiffness, and maximization of fatigue life, and obtain the target platform door structure according to the preset stress constraint conditions. Select the target material performance parameters of the obtained target platform door structure from a preset material database by the Bayesian optimization algorithm and the finite element analysis results under the lowest cost constraint condition.

[0008] In some embodiments, the pre-training process of the aerodynamic effect calculation model includes:

[0009] Obtain a training sample set, where the training sample set includes a source data set and a target data set. The source data set includes subway passing-station working conditions and their corresponding true values of pneumatic pressure data, and the target data set includes railway vehicle passing-station working conditions and their corresponding true values of pneumatic pressure data;

[0010] Use the source data set to train an initial pneumatic effect calculation model. The initial pneumatic effect calculation model includes a convolutional neural network for extracting spatial features of pneumatic pressure, a long short-term memory network for obtaining time series features of pneumatic pressure, and a fully connected layer. After taking the subway passing-station working conditions as input, it outputs predicted values of pneumatic pressure data under subway passing-station working conditions; In the initial pneumatic effect calculation model trained by the source data set, retain the convolutional layer weights of the convolutional neural network and change the long short-term memory network and the fully connected layer according to preset configuration parameters to obtain a migrated pneumatic effect calculation model. Use the target data set to train the migrated pneumatic effect calculation model. After taking the railway vehicle passing-station working conditions as input, it outputs predicted values of pneumatic pressure data under railway vehicle passing-station working conditions;

[0011] Construct a first loss function based on the predicted values and true values of pneumatic pressure data under subway passing-station working conditions, and construct a second loss function based on the predicted values and true values of pneumatic pressure data under railway vehicle passing-station working conditions. Taking the minimization of the first loss function and the second loss function as the goal, use the training sample set to update the parameters of the initial pneumatic effect calculation model until convergence to obtain the pneumatic effect calculation model.

[0012] In some embodiments, the pre-training process of the pneumatic effect calculation model further includes:

[0013] Set monitoring metrics during the training process and early stopping conditions corresponding to the monitoring metrics;

[0014] Use a preset early stopping function to check the monitoring metrics after each iteration. When the early stopping conditions are met, the iteration terminates.

[0015] In some embodiments, the method further includes:

[0016] Convert the basic platform door structure into a density variable distribution by the variable density method and construct a digital twin model of the basic platform door and an observation equation describing the stress data and pneumatic pressure data collected by the preset sensors and their changes;

[0017] Predict the stress data and pneumatic pressure data at the current moment based on the stress data, pneumatic pressure data, platform door structure, and material property parameters at the previous moment, and correct the stress data and pneumatic pressure data at the current moment according to the observation equation and the Kalman filtering algorithm;

[0018] Calculate the structural stress and strain nephogram based on the corrected stress data and pneumatic pressure data;

[0019] Adjust the basic platform screen door according to the target platform screen door structure and the target material property parameters to obtain the target platform screen door, and perform visual display on the target platform screen door.

[0020] In some embodiments, the method further includes:

[0021] Adjust the components in the platform screen door structure that do not meet the fatigue strength requirements by means of arc transition, patching and bar planting.

[0022] In some embodiments, the preset material database is constructed through a Bayesian optimization framework, and the process includes:

[0023] Collect the performance parameter data of various materials, and obtain the processed material performance parameters after cleaning, sorting and standardizing the collected material performance parameters;

[0024] Establish a material performance probability model with the material performance parameters as samples through Bayesian statistical methods;

[0025] Train the material performance probability model according to the collected material performance parameters and continuously adjust the parameters of the material performance probability model to obtain the target material performance probability model;

[0026] Integrate the processed material performance parameters and the target material performance probability model to obtain the preset material database.

[0027] On the other hand, the present invention also provides a platform screen door optimization system, which is used to execute the platform screen door optimization method described in any one of the above, and the system includes:

[0028] A pneumatic effect calculation module, which is used to obtain the pneumatic pressure change curve of the basic platform screen door under the condition of train passing through the station according to a pre-trained pneumatic effect calculation model;

[0029] A structural strength analysis module, including a modal analysis sub-module, a static structural strength analysis sub-module and a fatigue strength analysis sub-module; which is used to perform modal analysis, static structural strength analysis and fatigue strength analysis on the basic platform screen door according to the pneumatic pressure change curve;

[0030] A parametric structure optimization module, which is used to convert the basic platform door structure into a density variable distribution by the variable density method, and iteratively update the density variable distribution by the genetic algorithm with the goals of structural lightweighting, maximum stiffness, and maximum fatigue life, and obtain the target platform door structure according to the preset stress constraint conditions. Then, according to the lowest cost constraint condition and the finite element analysis results, select the target material performance parameters of the obtained target platform door structure from the preset material database.

[0031] In some embodiments, the system further includes:

[0032] An aerodynamic effect calculation result library, which is used to store the aerodynamic pressure change curves under the passing conditions of subway and railway vehicles, so as to select the corresponding aerodynamic pressure change curves according to the design requirements of the platform door structure for modal analysis, static structural strength analysis, and fatigue strength analysis.

[0033] In some embodiments, the parametric structure optimization module further includes:

[0034] A component and material data management library, which is used to uniformly manage the components and materials in the target platform door structure and the target material performance parameters.

[0035] On the other hand, the present invention also provides a computer-readable storage medium, on which a computer program / instructions are stored. When the program / instructions are executed by a processor, the steps of the method described in any one of the above are implemented.

[0036] In the platform door optimization method and system of the present invention, a combined aerodynamic effect calculation model of a convolutional neural network and a long short-term memory network is used to replace the fluid dynamics simulation. The migration technology is used to migrate the aerodynamic data of the subway platform door as a pre-trained model to the intercity / high-speed rail scenario, significantly reducing the simulation calculation time and improving the generalization ability of the aerodynamic effect calculation model. Through the digital twin model simulation and analysis, it helps to predict potential failures or performance degradation, so as to perform maintenance in advance, reduce downtime and maintenance costs. Design and test on the digital twin model before the actual manufacturing or construction of the product or system to optimize performance and reduce development costs. By various constraint conditions, the update process of the platform door is constrained to obtain the target platform door structure and the target material performance parameters that meet the requirements, thereby obtaining the target platform door.

[0037] The additional advantages, objectives, and features of the present invention will be partially described below, and will become partially obvious to those of ordinary skill in the art after studying the following text, or can be learned from the practice of the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the structures specifically pointed out in the description and the drawings.

[0038] Those skilled in the art will understand that the objectives and advantages achievable by the present invention are not limited to those specifically described above, and the above and other objectives achievable by the present invention will be more clearly understood according to the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The drawings described herein are used to provide a further understanding of the present invention, form a part of this application, and do not limit the present invention. In the drawings:

[0040] Figure 1 It is a schematic flow chart of the platform door optimization method according to an embodiment of the present invention.

[0041] Figure 2 It is a schematic structural diagram of the platform door optimization system according to an embodiment of the present invention.

[0042] Figure 3 It is a schematic structural diagram of the platform door optimization method according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0043] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with the embodiments and the drawings. Herein, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but do not limit the present invention.

[0044] Herein, it also needs to be noted that in order to avoid obscuring the present invention with unnecessary details, only the structures and / or processing steps closely related to the solution of the present invention are shown in the drawings, while other details less related to the present invention are omitted.

[0045] It should be emphasized that the term "comprising / including" when used herein refers to the presence of features, elements, steps, or components, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0046] Herein, it also needs to be noted that if not specifically stated, the term "connection" in this article can not only refer to direct connection, but also represent indirect connection with an intermediate.

[0047] Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the drawings, the same reference numerals represent the same or similar components, or the same or similar steps.

[0048] In the prior art, for urban subways, due to the relatively low train operation speed, unified types such as train models and stations, the generated aerodynamic effects are relatively consistent, and there are corresponding standards to constrain the corresponding platform screen door structure forms; for intercity railways, suburban railways, high-speed railways, etc., due to the large number of train models, formation types, station forms, and train operation speeds, it is difficult to determine the installation position of the platform screen door and the requirements for the structural strength of the platform screen door; the present invention proposes an optimization method, system, and storage medium for platform screen doors. After obtaining the aerodynamic pressure change curve of the basic platform screen door under the train passing station condition according to the pre-trained aerodynamic effect calculation model, modal analysis, static structural strength analysis, and fatigue strength analysis are performed on the basic platform screen door, and then the basic platform screen door structure is transformed into a density variable distribution by the variable density method. The genetic algorithm is used to iteratively update the density variable distribution with the goals of structural lightweighting, maximization of stiffness, and maximization of fatigue life according to the preset stress constraint conditions to obtain the target platform screen door structure. The Bayesian optimization algorithm and the finite element analysis results are used to select the target material performance parameters of the target platform screen door structure in the preset material database under the lowest cost constraint condition.

[0049] Figure 1 It is a schematic flow chart of the platform screen door optimization method according to an embodiment of the present invention. One aspect of the present invention provides a platform screen door optimization method, and the method includes the following steps S101 to S102:

[0050] Step S101: Obtain the aerodynamic pressure change curve of the basic platform screen door under the train passing station condition according to the pre-trained aerodynamic effect calculation model.

[0051] Step S102: After performing modal analysis, static structural strength analysis, and fatigue strength analysis on the basic platform screen door according to the aerodynamic pressure change curve, the basic platform screen door structure is transformed into a density variable distribution by the variable density method. The genetic algorithm is used to iteratively update the density variable distribution with the goals of structural lightweighting, maximization of stiffness, and maximization of fatigue life according to the preset stress constraint conditions to obtain the target platform screen door structure. The Bayesian optimization algorithm and the finite element analysis results are used to select the target material performance parameters of the target platform screen door structure in the preset material database under the lowest cost constraint condition.

[0052] In step S101, the present invention optimizes platform screen doors for various rail transit lines including but not limited to urban subways, intercity railways, suburban railways, and high-speed railways. When various transportation means including but not limited to subways and railway vehicles pass through stations at high speeds, they generate aerodynamic effects that affect the surrounding environment, relevant people, and the safety of ground facilities. The pre-trained aerodynamic effect calculation model obtains the aerodynamic pressure change curves of different rail transit lines when transportation means pass through stations through transfer technology. The aerodynamic pressure change curve is the pressure change curve of the aerodynamic effect. Further, according to the platform screen door structure design requirements, the corresponding aerodynamic pressure change curve is selected from the aerodynamic effect calculation result library for storing the aerodynamic pressure change curves under the passing conditions of subways and railway vehicles for modal analysis, static structural strength analysis, and fatigue strength analysis.

[0053] In some embodiments, the pre-training process of the aerodynamic effect calculation model includes steps S1011 to S1013:

[0054] Step S1011: Obtain a training sample set. The training sample set includes a source data set and a target data set. The source data set includes subway passing conditions and their corresponding true values of aerodynamic pressure data. The target data set includes railway vehicle passing conditions and their corresponding true values of aerodynamic pressure data.

[0055] Step S1012: Use the source data set to train an initial aerodynamic effect calculation model. The initial aerodynamic effect calculation model includes a convolutional neural network for extracting spatial features of aerodynamic pressure, a long short-term memory network for obtaining time series features of aerodynamic pressure, and a fully connected layer. Taking the subway passing conditions as input, it outputs the predicted value of aerodynamic pressure data under subway passing conditions. In the initial aerodynamic effect calculation model after training with the source data set, retain the weights of the convolutional layers of the convolutional neural network and change the long short-term memory network and the fully connected layer according to preset configuration parameters to obtain a transferred aerodynamic effect calculation model. Use the target data set to train the transferred aerodynamic effect calculation model. Taking the railway vehicle passing conditions as input, it outputs the predicted value of aerodynamic pressure data under railway vehicle passing conditions.

[0056] Step S1013: Construct a first loss function based on the predicted value and the true value of aerodynamic pressure data under subway passing conditions, and construct a second loss function based on the predicted value and the true value of aerodynamic pressure data under railway vehicle passing conditions. Taking the minimization of the first loss function and the second loss function as the goal, use the training sample set to update the parameters of the initial aerodynamic effect calculation model until convergence to obtain the aerodynamic effect calculation model.

[0057] In some embodiments, the pre-training process of the aerodynamic effect calculation model further includes steps S10131 to S10132:

[0058] Step S10131: Set the monitoring metrics during the training process and the early stopping conditions corresponding to the monitoring metrics.

[0059] Step S10132: Use a preset early stopping function to check the monitoring metrics after each iteration. When the early stopping conditions are met, the iteration terminates.

[0060] Specifically, the overall aerodynamic pressure data under historical train passing-station working conditions includes the aerodynamic pressure distributions under different speeds, vehicle types, and formation conditions. After normalization, denoising, and segmentation, training sets, validation sets, and test sets are obtained. After training the initial aerodynamic effect calculation model and the transfer aerodynamic effect calculation model with the training set, the performance of the model is monitored using the validation set, and hyperparameters including the learning rate, batch size, and number of network layers are adjusted. The performance of the model is evaluated using the test set, the prediction error is calculated, and the performance of the aerodynamic effect calculation model under different working conditions is analyzed by comparing the prediction results with the actual results; the mean squared error loss function is used as the loss function; the early stopping method is used to prevent overfitting.

[0061] In step S102, the basic platform door structure is the topological form of the key components of the basic platform door, including but not limited to the door frame and connectors; the natural frequencies and vibration modes of the platform door structure are obtained through modal analysis, and the weak structural positions are judged and the connection relationships of the assemblies are checked; the load-bearing capacity of the basic platform door structure is obtained through static analysis to determine whether the strength and stiffness of the structure meet the requirements, and the fatigue strength analysis is used to obtain the fatigue resistance ability of the platform door structure under cyclic loads. The load is the spatial distribution of the forces borne on the platform door structure. The process of converting the basic platform door structure into a density variable distribution by the variable density method includes: discretizing the basic platform door structure into multiple elements and introducing a density variable for each element to represent the material filling degree. The value range of the density variable is 0-1, where 0 means no material and 1 means full material. Then, a functional relationship between the material properties and the density variable is established, and the density variables of each element are determined according to the optimization objective, thereby obtaining the density variable distribution of the platform door structure. The density variable distribution describes the topological form of the platform door structure. Further, the genetic algorithm uses the multi-objective genetic algorithm (NSGA-II) to solve multi-objective optimization problems; the preset stress constraint conditions introduce a stress concentration sensitivity factor to actively avoid high-stress areas during the iteration process to improve the reliability of the structure. In some embodiments, the method further includes: adjusting the components in the platform door structure that do not meet the fatigue strength requirements by means of arc transition, patching, and bar planting, and can also be adjusted by changing the force application method.

[0062] In some embodiments, the preset material database is constructed through the Bayesian optimization framework, and the process includes steps S1021 to S1024:

[0063] Step S1021: Collect the performance parameter data of various materials. After cleaning, sorting, and standardizing the collected material performance parameters, the processed material performance parameters are obtained.

[0064] Step S1022: Use the Bayesian statistical method to establish a material performance probability model with the material performance parameters as samples.

[0065] Step S1023: Train the material performance probability model according to the collected material performance parameters and continuously adjust the parameters of the material performance probability model to obtain the target material performance probability model.

[0066] Step S1024: Integrate the processed material performance parameters and the standard material performance probability model to obtain a preset material database.

[0067] Furthermore, establish a finite element model according to the target platform door structure and application scenario. Select different materials from the preset material database for combination to obtain multiple material combination schemes, and input each material combination scheme into the finite element model for finite element analysis and calculation. Evaluate whether each material combination scheme meets the requirements according to the finite element analysis results. The Bayesian optimization algorithm constructs a surrogate model of the objective function based on the finite element analysis results, compares the material combination schemes in combination with the acquisition function, and searches to find the optimal material combination scheme as the target material performance parameters; the preset material database contains the material performance parameters of various materials with different tensile strengths and fatigue limits.

[0068] In some embodiments, the method further includes steps S1 - S4:

[0069] Step S1: Transform the basic platform door structure into a density variable distribution by the variable density method, construct a digital twin model of the basic platform door, and an observation equation describing the stress data and pneumatic pressure data collected by the preset sensors and their changes.

[0070] Step S2: Predict the stress data and pneumatic pressure data at the current moment based on the stress data, pneumatic pressure data, platform door structure, and material performance parameters at the previous moment, and correct the stress data and pneumatic pressure data at the current moment according to the observation equation and the Kalman filter algorithm.

[0071] Step S3: Calculate and obtain the structural stress and strain nephogram according to the corrected stress data and pneumatic pressure data.

[0072] Step S4: Adjust the basic platform door according to the target platform door structure and target material performance parameters to obtain the target platform door, and perform a visual display of the target platform door.

[0073] Specifically, the digital twin model is used for data integration, mapping, and three-dimensional visualization. The data during the platform door optimization process is input into the digital twin model, and the stress data and pneumatic pressure data collected by preset sensors are used for accuracy correction. Thus, more accurate and specific platform door stress data and pneumatic pressure data are obtained, and a structural stress and strain cloud map is constructed therefrom, and the target platform door composed of the iteratively updated target platform door structure and target material performance parameters is visually displayed. The structural stress and strain diagram can be visually displayed through a preset visualization tool.

[0074] Figure 2 It is a schematic structural diagram of the platform door optimization system according to an embodiment of the present invention. On the other hand, the present invention also provides a platform door optimization system, which is used to execute the platform door optimization method of any one of the above. The system includes:

[0075] The pneumatic effect calculation module is used to obtain the pneumatic pressure change curve of the basic platform door under the condition of the train passing through the station according to the pre-trained pneumatic effect calculation model.

[0076] The structural strength analysis module includes a modal analysis sub-module, a static structural strength analysis sub-module, and a fatigue strength analysis sub-module; it is used to perform modal analysis, static structural strength analysis, and fatigue strength analysis on the basic platform door according to the pneumatic pressure change curve.

[0077] The parametric structural optimization module is used to convert the basic platform door structure into a density variable distribution by the variable density method, and use the genetic algorithm to iteratively update the density variable distribution with the goals of structural lightweighting, maximum stiffness, and maximum fatigue life according to the preset stress constraint conditions to obtain the target platform door structure, and select the target material performance parameters of the target platform door structure in the preset material database with the lowest cost constraint condition through the Bayesian optimization algorithm and the finite element analysis results.

[0078] In some embodiments, the system further includes:

[0079] The pneumatic effect calculation result library is used to store the pneumatic pressure change curves under the conditions of subway and railway vehicle passing through the station, so as to select the corresponding pneumatic pressure change curves according to the platform door structure design requirements for modal analysis, static structural strength analysis, and fatigue strength analysis.

[0080] In some embodiments, the parametric structural optimization module further includes:

[0081] The component and material data management library is used to uniformly manage the components and materials in the target platform door structure and target material performance parameters.

[0082] On the other hand, the present invention also provides a computer-readable storage medium, on which a computer program / instructions are stored, and when the program / instructions are executed by a processor, the steps of any one of the above methods are implemented.

[0083] The present invention will be described below in conjunction with a specific embodiment:

[0084] The present invention provides a platform door optimization method, system and storage medium, which perform multidisciplinary structural optimization on the design of platform doors for different lines. The theoretical methods of multidisciplinary structural optimization include a dynamic aerodynamic effect prediction algorithm based on deep learning, a multi-objective topology optimization algorithm, a real-time data-driven dynamic adjustment algorithm, and a material-structure collaborative optimization algorithm; the platform door optimization methods include, but are not limited to, adjusting the structural dimensions, changing the connection relationship, adding strengthening structural members, and upgrading the material properties; statistically analyzing the structural dimensions, changing the connection relationship, and adding strengthening structural members are used to optimize the platform door structure, and upgrading the material properties is used to optimize the material property parameters of the platform door.

[0085] The platform door optimization system includes an aerodynamic effect calculation module, a structural strength analysis module, and a parametric structural optimization module. Among them, the structural strength analysis module includes: a modal analysis sub-module, a static structural strength analysis sub-module, and a fatigue strength analysis sub-module. In the parametric structural optimization module, a basic platform door part library is established, and structural optimization is achieved by parametric design and management of the dimensions of the main structural components; a parts database is established to uniformly manage the strengthening structures of the platform doors that need to be added.

[0086] Figure 3 It is a structural schematic diagram of the platform door structure optimization method according to an embodiment of the present invention. The working process in the present invention includes: First, the aerodynamic effect calculation module calculates the pressure change curve of the aerodynamic effect under different train passing station conditions, and stores the calculation results in the aerodynamic effect calculation result library according to the working conditions; Secondly, according to the platform door structure design requirements, the corresponding aerodynamic pressure change curve is called from the aerodynamic effect calculation result library, and the basic platform door is sequentially subjected to modal analysis and static structural strength analysis. According to the analysis results, the performance of the basic type platform door is strengthened by methods including but not limited to adjusting the structural dimensions, changing the connection relationship, upgrading the material properties, and adding strengthening structural members. The upgraded platform door is re-analyzed for modal analysis and static structural strength analysis until the design requirements are met. Next, fatigue strength analysis is carried out. According to the design requirements and the relevant specification standards of the platform door, the components that do not meet the fatigue strength requirements are optimized by methods including but not limited to arc transition, patching, implanting steel bars, and changing the force application method until the fatigue strength meets the requirements; modal analysis, static structural analysis, and fatigue strength analysis are iteratively performed multiple times, and finally the target platform door suitable for the corresponding calculation conditions is completed.

[0087] 1. Dynamic Aerodynamic Effect Prediction Algorithm Based on Deep Learning: A hybrid model combining Convolutional Neural Network (CNN) and Long Short-Term Memory Network (LSTM) is adopted. It is trained with the aerodynamic pressure data of historical train passing-station conditions to achieve rapid prediction of aerodynamic pressure distribution under different speeds, vehicle types, and formation conditions. Through transfer learning technology, the aerodynamic data of subway platform screen doors is used as a pre-trained model and transferred to the railway scenarios of intercity / high-speed railways, significantly reducing the simulation calculation time. It replaces part of the computational fluid dynamics simulation in the aerodynamic effect calculation module for rapid iteration in the preliminary design stage. Collect the aerodynamic pressure data of historical train passing-station conditions, including the aerodynamic pressure distribution under different speeds, vehicle types, and formation conditions; preprocess the data, including operations such as normalization, denoising, and segmentation; divide the data into training set, validation set, and test set according to a preset ratio; the transfer learning data includes source dataset and target dataset, and both types of data are normalized, segmented, and noise-filtered to ensure the consistency of the feature space. Construct a CNN-LSTM hybrid model as the initial aerodynamic pressure calculation model. The convolutional neural network is used to extract the spatial features of aerodynamic pressure, such as the spatial pattern of pressure distribution; the long short-term memory network is used to capture the time-series features of aerodynamic pressure, such as the change trend of pressure over time; use transfer learning technology to transfer the aerodynamic data of subway platform screen doors as a pre-trained model to the intercity / high-speed railway scenario; the transfer learning model retains the convolutional layer weights of the convolutional neural network, freezes its parameters, modifies the long short-term memory network and fully connected layer to adapt to the output dimension of the target task. Use the training set data to train the model and optimize the loss function; monitor the model performance through the validation set and adjust the hyperparameters, including but not limited to learning rate, batch size, and number of network layers; further, use the early stopping method to prevent overfitting. Use the test set to evaluate the model performance and calculate the prediction error; visualize the comparison between the prediction results and the actual results to analyze the performance of the model under different working conditions.

[0088] 2. Multi-Objective Topology Optimization Algorithm: Based on the variable density method and multi-objective genetic algorithm, with the optimization objectives of structural lightweight, maximum stiffness, and maximum fatigue life, it generates the optimal topological forms of key components of the platform screen door, including but not limited to door frames and connectors; introduces the stress concentration sensitivity factor to actively avoid high-stress areas and improve structural reliability during the optimization process, and can generate the geometric configurations of new reinforcement structures and obtain the target platform screen door structure in the parametric structural optimization module.

[0089] 3. Real-time data-driven dynamic adjustment algorithm: Through the prediction-update loop, Kalman filtering fuses multi-source sensor data collected by preset sensors and dynamically corrects the parameters of the digital twin model, including data modeling: defining the digital twin model of the platform door structure and the observation equation describing the stress data and pneumatic pressure data collected by the sensors and their changes; carrying out real-time data fusion: predicting the stress data and pneumatic pressure data at the current moment based on the stress data, pneumatic pressure data, platform door structure and material property parameters at the previous moment, and using the sensor observation equation to correct the predicted values; correcting parameters: updating the load distribution according to the corrected pneumatic pressure data. The digital twin model, as the virtual mirror of the platform door, works in cooperation with Kalman filtering to form a "monitoring-simulation-optimization" closed loop, including data mapping: synchronizing sensor data to the digital twin model and data preprocessing; dynamically updating the simulation model: injecting the stress data and pneumatic pressure data corrected by Kalman filtering, calculating the structural stress and strain nephogram, and displaying the stress data, pneumatic pressure data, target platform door structure and target material property parameters.

[0090] 4. Material-structure collaborative optimization algorithm: Based on the Bayesian optimization framework, a material database is constructed. Combining the finite element analysis results, the optimal material combination is automatically matched, such as the aluminum alloy-composite material hybrid structure; introducing cost constraint conditions, and preferentially selecting low-cost and easy-to-process material solutions on the premise of meeting the mechanical properties; the parts database supports the intelligent recommendation of material upgrade solutions.

[0091] In summary, the present invention provides a platform door optimization method, system and storage medium. According to the pre-trained pneumatic effect calculation model, the pneumatic pressure change curve of the basic platform door under the train passing station condition is obtained. After performing modal analysis, static structural strength analysis and fatigue strength analysis on the basic platform door according to the pneumatic pressure change curve, the basic platform door structure is transformed into a density variable distribution by the variable density method. Through the genetic algorithm, with the goals of structural lightweight, maximum stiffness and maximum fatigue life, and according to the preset stress constraint conditions, the density variable distribution is iteratively updated to obtain the target platform door structure. Through the Bayesian optimization algorithm and the finite element analysis results, the target material property parameters of the target platform door structure are selected from the preset material database under the condition of the lowest cost constraint.

[0092] The embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the foregoing edge computing server deployment method are implemented. The computer-readable storage medium may be a tangible storage medium, such as a random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, register, floppy disk, hard disk, removable storage disk, CD-ROM, or any other form of storage medium known in the technical field.

[0093] Those of ordinary skill in the art should understand that the various exemplary components, systems, and methods described in connection with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Specifically, whether to implement in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present invention. When implemented in hardware, it can be, for example, an electronic circuit, an application-specific integrated circuit (ASIC), appropriate firmware, a plug-in, a functional card, and so on. When implemented in software, the elements of the present invention are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium or transmitted through a data signal carried in a carrier wave over a transmission medium or a communication link.

[0094] It should be clear that the present invention is not limited to the specific configurations and processes described above and illustrated in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and illustrated as examples. However, the method process of the present invention is not limited to the specific steps described and illustrated. Those skilled in the art can make various changes, modifications, and additions, or change the order between steps after understanding the spirit of the present invention.

[0095] In the present invention, the features described and / or exemplified for one embodiment can be used in the same way or in a similar way in one or more other embodiments, and / or combined with the features of other embodiments or replace the features of other embodiments.

[0096] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, various changes and variations can be made to the embodiments of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An optimization method for platform screen doors, characterized in that, The method includes the following steps: Obtain the pneumatic pressure change curve of the basic platform screen door under the train passing station condition according to the pre-trained pneumatic effect calculation model; After performing modal analysis, static structural strength analysis, and fatigue strength analysis on the basic platform screen door according to the pneumatic pressure change curve, convert the basic platform screen door structure into a density variable distribution by the variable density method. Use the genetic algorithm with the goals of structural lightweighting, maximizing stiffness, and maximizing fatigue life, and iteratively update the density variable distribution according to the preset stress constraint conditions to obtain the target platform screen door structure. Select the target material property parameters for the obtained target platform screen door structure from the preset material database by the Bayesian optimization algorithm and the finite element analysis results under the lowest cost constraint condition.

2. The platform door optimization method according to claim 1, wherein The pre-training process of the pneumatic effect calculation model includes: Obtain a training sample set, which includes a source data set and a target data set. The source data set includes subway passing station conditions and the corresponding true values of pneumatic pressure data, and the target data set includes railway vehicle passing station conditions and the corresponding true values of pneumatic pressure data; Train the initial pneumatic effect calculation model using the source data set. The initial pneumatic effect calculation model includes a convolutional neural network for extracting spatial features of pneumatic pressure, a long short-term memory network for obtaining temporal sequence features of pneumatic pressure, and a fully connected layer. Taking the subway passing station condition as the input, output the predicted value of pneumatic pressure data under the subway passing station condition; In the initial pneumatic effect calculation model after training with the source data set, retain the convolutional layer weights of the convolutional neural network and change the long short-term memory network and the fully connected layer according to the preset configuration parameters to obtain a transfer pneumatic effect calculation model. Train the transfer pneumatic effect calculation model using the target data set, and take the railway vehicle passing station condition as the input to output the predicted value of pneumatic pressure data under the railway vehicle passing station condition; Construct a first loss function based on the predicted value and the true value of pneumatic pressure data under the subway passing station condition, and construct a second loss function based on the predicted value and the true value of pneumatic pressure data under the railway vehicle passing station condition. With the goal of minimizing the first loss function and the second loss function, use the training sample set to update the parameters of the initial pneumatic effect calculation model until convergence to obtain the pneumatic effect calculation model.

3. The platform door optimization method according to claim 2, wherein The pre-training process of the pneumatic effect calculation model further includes: Set the monitoring indicators during the training process and the early stopping conditions corresponding to the monitoring indicators; Use the preset early stopping function to check the monitoring indicators after each iteration round, and terminate the iteration when the early stopping conditions are met.

4. The platform door optimization method according to claim 1, characterized in that The method further includes: Convert the basic platform screen door structure into a density variable distribution by the variable density method, construct a digital twin model of the basic platform screen door, and an observation equation describing the stress data and the pneumatic pressure data collected by the preset sensors and their changes. Predict the stress data and pneumatic pressure data at the current moment based on the stress data, pneumatic pressure data, platform door structure, and material property parameters at the previous moment, and correct the stress data and pneumatic pressure data at the current moment according to the observation equation and Kalman filtering algorithm; Calculate and obtain the structural stress and strain nephogram based on the corrected stress data and pneumatic pressure data; Adjust the basic platform door according to the target platform door structure and the target material property parameters to obtain the target platform door, and perform visual display on the target platform door.

5. The platform door optimization method according to claim 1, wherein The method further includes: Adjust the components in the platform door structure that do not meet the fatigue strength requirements by means of arc transition, patching, and bar planting.

6. The platform door optimization method according to claim 1, wherein The preset material database is constructed through a Bayesian optimization framework, and the process includes: Collect the performance parameter data of various materials, and obtain the processed material performance parameters after cleaning, sorting, and standardizing the collected material performance parameters; Establish a material performance probability model with the material performance parameters as samples through Bayesian statistical methods; Train the material performance probability model according to the collected material performance parameters and continuously adjust the parameters of the material performance probability model to obtain the target material performance probability model; Integrate the processed material performance parameters and the target material performance probability model to obtain the preset material database.

7. An optimized platform screen door system, characterized in that, The system is used to execute the platform door optimization method according to any one of claims 1 to 6, and the system includes: A pneumatic effect calculation module, which is used to obtain the pneumatic pressure change curve of the basic platform door under the train passing through the station condition according to the pre-trained pneumatic effect calculation model; A structural strength analysis module, including a modal analysis sub-module, a static structural strength analysis sub-module, and a fatigue strength analysis sub-module; which is used to perform modal analysis, static structural strength analysis, and fatigue strength analysis on the basic platform door according to the pneumatic pressure change curve; A parametric structural optimization module, which is used to convert the basic platform door structure into a density variable distribution by the variable density method, and use the genetic algorithm to iteratively update the density variable distribution with the goals of structural lightweight, maximum stiffness, and maximum fatigue life, and select the target material property parameters for the obtained target platform door structure in the preset material database according to the Bayesian optimization algorithm and the finite element analysis results under the minimum cost constraint condition.

8. The platform door optimization system according to claim 7, wherein The system further includes: A pneumatic effect calculation result library, which is used to store the pneumatic pressure change curves under the conditions of subway and railway vehicle passing through the station, so as to select the corresponding pneumatic pressure change curves for modal analysis, static structural strength analysis, and fatigue strength analysis according to the platform door structure design requirements.

9. The platform door optimization system according to claim 7, characterized in that The parametric structural optimization module further includes: A component and material data management library, which is used to uniformly manage the components and materials in the target platform door structure and the target material property parameters.

10. A computer-readable storage medium having computer programs / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images