Composite beam bridge deck slab intelligent design method based on GEP + TCN + GA

Abstract

The invention provides an intelligent design method for a composite beam bridge deck slab based on GEP + TCN + GA, and relates to the technical field of composite beam bridge deck slab design. An improved gene expression programming algorithm is innovatively combined with finite element simulation and a static force loading test, high-precision mathematical modeling of boundary conditions of the composite beam bridge deck slab is achieved, and the design efficiency of the composite beam bridge deck slab is improved. The limitation of a traditional empirical formula is broken through, and the automation and accuracy of boundary condition expression are remarkably improved. A closed-loop intelligent design system is formed by combining the efficient time sequence prediction capability of the time convolutional network algorithm and the intelligent optimization framework of the genetic algorithm, and the panel mechanical response can be rapidly and accurately predicted and the design scheme can be optimized under the complex multi-parameter working condition. The design efficiency and precision are improved, the consumption of simulation calculation resources is reduced, the generalization ability and the practical engineering applicability of the model are enhanced, and the requirements of the modern composite beam bridge deck slab structure design for intellectualization, refinement and high efficiency are fully met.

Description

Technical Field

[0001] The present invention relates to the technical field of composite beam bridge deck design, and in particular to an intelligent design method for composite beam bridge decks based on GEP+TCN+GA. Background Art

[0002] With the rapid development of transportation infrastructure, composite beam bridges, as a common form of bridge structure, have panels whose design directly affects the bearing performance and service life of the bridge. Traditional composite beam bridge panel design methods mostly rely on empirical formulas or finite element simulations, which make it difficult to take into account complex boundary conditions and nonlinear mechanical behavior under multiple working conditions. In addition, the simulation calculation cost is high and the design cycle is long. In actual engineering, the panel force involves a variety of influencing factors, such as boundary constraint stiffness, rotational stiffness, and pressure film effect. There is a complex coupling relationship between these factors, which is difficult to accurately describe with traditional modeling methods. In addition, the existing design lacks a closed-loop optimization mechanism that synergistically integrates simulation data and experimental data, making it impossible to achieve intelligent and efficient design processes, limiting the accuracy of the composite beam bridge panel structure optimization design and the promotion of engineering applications.

[0003] Prior art publication CN116776427A discloses a parametric BIM modeling method and system for the longitudinal stiffeners of a spatial curved deck in a steel box girder bridge. The method includes the following steps: S10: Opening the main window of the Tekla Structure platform and entering the parameters for the spatial geometric arrangement of the longitudinal ribs in the curved deck; S20: Entering the cross-sectional shape of the longitudinal stiffeners not found in the form definition library into the Tekla platform database; S30: Obtaining the coordinate library for the intersection of each longitudinal stiffener and the diaphragm structure; S40: Entering the longitudinal stiffener opening shape parameters corresponding to the longitudinal stiffener cross-section from the form. While this method enables intelligent design of beam bridge decks, it cannot describe the coupling relationships under complex working conditions, resulting in insufficient intelligence and flexibility, making it difficult to meet the requirements of intelligent and efficient design.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent design method for composite beam bridge deck based on GEP+TCN+GA to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] An intelligent design method for composite beam bridge deck based on GEP+TCN+GA, the specific steps include:

[0008] S1: Construct a finite element model of the composite beam bridge, simulate it, build a boundary condition database of the composite beam bridge deck based on the simulation results, and build a calculation model of the composite beam bridge deck boundary conditions based on the GEP algorithm and the boundary condition database;

[0009] S2: Construct composite beam bridge deck specimens and conduct multi-parameter static loading tests to reflect the pressure membrane effect of the composite beam bridge deck during the load process. Based on the test data, the calculation model of the composite beam bridge deck boundary conditions is modified;

[0010] S3: Simulate the stress process of the composite beam bridge deck based on the revised calculation model, and build a mechanical performance database of the composite beam bridge deck based on the simulation results;

[0011] S4: Construct a TCN algorithm model. Based on the TCN algorithm model and the mechanical properties database of composite beam bridge decks, simulate and predict the stress process of composite beam bridge decks.

[0012] S5: Establish an optimization design framework for the GA algorithm, and embed the calculation model of the composite beam bridge deck boundary conditions and the TCN algorithm model into the GA algorithm. Perform fitness evaluation on the boundary conditions and mechanical properties of the composite beam bridge deck, and search for the optimal design scheme based on the evaluation results to complete the intelligent design of the composite beam bridge deck.

[0013] Preferably, in step S1, the boundary conditions include lateral constraint stiffness, lateral constraint eccentricity, and rotational constraint stiffness;

[0014] When using the GEP algorithm to construct a calculation model for the boundary conditions of composite beam bridge decks, the least squares method, tournament algorithm, and population update strategy are used to improve the GEP algorithm. The specific logic is as follows:

[0015] Use the least squares method to fit the real constant in the GEP algorithm;

[0016] Use the tournament algorithm to enhance the population selection pressure and diversity in the GEP algorithm, and set the tournament size range to 3 to 7;

[0017] The population update strategy is used to improve the global search capability and convergence speed of the GEP algorithm, and the population size in the population update strategy is set in the range of 50 to 200.

[0018] Preferably, the calculation model expression of the boundary condition of the composite beam bridge deck is:

[0019] BC i =c i ·f(P i )

[0020] Where BC i Indicates the parameter size of the i-th group of boundary conditions, c i represents the real constant corresponding to the i-th set of boundary conditions, f(P i ) represents the basic design function of the i-th group of boundary conditions, P i The vector representing the i-th set of boundary conditions, where the subscript i represents the index of the boundary condition and m represents the total number of boundary conditions.

[0021] Preferably, in step S2, when performing the static loading test, the multiple parameters that are changed include lateral constraint stiffness, lateral constraint eccentricity, rotational constraint stiffness, plate thickness, and reinforcement ratio;

[0022] The pressure film effect is expressed as:

[0023]

[0024] Where σ p Represents the additional stress of the pressure film, ∈ p represents the strain generated by the pressure film, k p Indicates the sensitivity coefficient of the pressure film effect, k p ∈[0.1,0.5], ∈0 represents the critical strain threshold, and β represents the quadratic term coefficient;

[0025] The calculation model of the modified composite beam bridge deck boundary conditions is expressed as:

[0026] BC i ′=c i ·f(P i )+a i ·σ p

[0027] Where BC i ′ represents the parameter size of the modified i-th group of boundary conditions, a i represents the sensitivity coefficient of the i-th group of boundary condition parameters to the pressure film effect, a i ∈[0.05,0.3].

[0028] Preferably, in step S3, the mechanical properties database of the composite beam bridge deck is constructed according to the simulation results, including: the parameter size of the corrected boundary conditions, the additional stress of the pressure film, the strain generated by the pressure film, and the deformation rate of the composite beam bridge deck.

[0029] Preferably, in step S4, the input of the TCN algorithm model is the mechanical properties database of the composite beam bridge deck. The TCN model structure is composed of 4 to 6 convolutional layers, and the convolutional layers include causal convolution layers and dilated convolution layers. A residual module is included to alleviate the gradient vanishing problem, and the convolution kernel adopts an adaptive dilation coefficient, which is calculated as follows:

[0030]

[0031] Where d l represents the expansion coefficient of the lth convolutional layer, γ represents the growth factor, γ∈[1.5,2], d base represents the initial expansion coefficient, Indicates rounding down.

[0032] Preferably, in step S5, when the fitness evaluation is performed on the boundary conditions and mechanical properties of the composite beam bridge deck, the initial population of the GA algorithm is set based on the design parameters, wherein the design parameters include: the number of main beams, the main beam size, the cross-diaphragm spacing, and the cross-diaphragm stiffness, and the design scheme is encoded as a multidimensional parameter vector.

[0033] Preferably, when performing fitness evaluation on the boundary conditions and mechanical properties of the composite beam bridge deck, the fitness function expression is:

[0034]

[0035] Where F represents the fitness score, w1, w2, and w3 represent weight coefficients, and w1+w2+w3=1, P represents the carrying capacity of the design scheme, and P req represents the bearing capacity requirement, δ represents the deformation of the design solution, v limit represents the deformation threshold, M represents the material consumption of the main beam of the design scheme, and M ref Indicates the reference main beam material usage;

[0036] After the evaluation is completed, the design with the highest fitness score will be regarded as the optimal design.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] The present invention innovatively combines the improved genetic expression programming algorithm with finite element simulation and static loading tests to achieve high-precision mathematical modeling of the boundary conditions of composite beam bridge panels, breaking through the limitations of traditional empirical formulas and significantly improving the automation and accuracy of boundary condition expression. Combining the efficient timing prediction capabilities of the time convolutional network algorithm and the intelligent optimization framework of the genetic algorithm, a closed-loop intelligent design system integrating model construction, performance prediction and design optimization is formed, which can quickly and accurately predict the mechanical response of the panel and find the optimal design solution under complex multi-parameter working conditions. This method not only improves design efficiency and accuracy, reduces simulation computing resource consumption, but also enhances the generalization ability of the model and its applicability to actual engineering projects, fully meeting the needs of modern composite beam bridge panel structure design for intelligence, refinement and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1Schematic diagram of the overall method flow of the present invention;

[0040] Figure 2 Schematic diagram of the fitting error between the calculation model of the present invention and the traditional empirical formula;

[0041] Figure 3 Schematic diagram of the prediction error between the TCN algorithm model of the present invention and the traditional RNN algorithm model. DETAILED DESCRIPTION

[0042] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0043] It should be noted that, unless otherwise defined, the technical or scientific terms used in the present invention should have the usual meanings understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative position relationships. When the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0044] Example:

[0045] See also Figures 1 to 3 , the present invention provides a technical solution:

[0046] An intelligent design method for composite beam bridge deck based on GEP+TCN+GA, the specific steps include:

[0047] S1: Construct a finite element model of the composite beam bridge, simulate it, build a boundary condition database of the composite beam bridge deck based on the simulation results, and construct a calculation model of the boundary conditions of the composite beam bridge deck based on the GEP algorithm and the boundary condition database.

[0048] In step S1, the boundary conditions include lateral constraint stiffness, lateral constraint eccentricity, and rotational constraint stiffness;

[0049] When using the GEP algorithm (genetic expression programming algorithm) to construct a calculation model for the boundary conditions of composite beam bridge decks, the least squares method, tournament algorithm, and population update strategy are used to improve the GEP algorithm. The specific logic is as follows:

[0050] Use the least squares method to fit the real constant in the GEP algorithm;

[0051] Use the tournament algorithm to enhance the population selection pressure and diversity in the GEP algorithm, and set the tournament size range to 3 to 7;

[0052] The population update strategy is used to improve the global search capability and convergence speed of the GEP algorithm, and the population size in the population update strategy is set in the range of 50 to 200.

[0053] Specifically, when constructing a finite element model, tools such as ANSYS can be used to batch generate simulation data under different boundary conditions (lateral constraint stiffness / eccentricity, etc.). The improved GEP algorithm (tournament selection + least squares fitting) is then used to mine the nonlinear relationships between parameters, encoding the design parameters into genotypes (tree structures). Genetic operations (crossover, mutation) generate candidate expressions, and ultimately the optimal model is selected to construct an expression for the calculation model of the boundary conditions of the composite beam bridge deck. This method can take into account the detailed characteristics and working conditions of different structures, ensuring the broad representativeness and accuracy of the model. Using the GEP algorithm to automatically mine boundary condition formulas instead of traditional empirical formulas can improve the automation and accuracy of boundary condition modeling. Furthermore, incorporating least squares fitting, tournament selection, and population update strategies into the GEP algorithm can improve the model's search efficiency and fitting accuracy, enhancing the model's practicality and accuracy.

[0054] The calculation model expression of the boundary condition of the composite beam bridge deck is:

[0055] BC i =c i ·f(P i )

[0056] Where BC i Indicates the parameter size of the i-th group of boundary conditions, c i represents the real constant corresponding to the i-th set of boundary conditions, f(P i ) represents the basic design function of the i-th group of boundary conditions, P i The vector representing the i-th set of boundary conditions, where the subscript i represents the index of the boundary condition and m represents the total number of boundary conditions.

[0057] It can be seen from the calculation model that it is an explicit mathematical expression based on symbolic regression, which is used to describe the nonlinear relationship between the boundary condition parameters and design variables of the composite beam bridge deck. It is an improved GEP (Gene Expression Programming) algorithm that automatically mines the mathematical expression of the boundary conditions from finite element simulation and test data. The model expresses the boundary conditions in linear combinations, but the basic design function itself can be a nonlinear function (such as a polynomial, exponential function, or trigonometric function), thereby capturing the complex coupling relationship between parameters. Where c i It can be obtained by fitting finite element data through an improved least squares method (such as Tikhonov regularization), and then reverse calibrated in combination with the results of static loading tests. The basic design function can be determined based on engineering experience or expert experience, or the optimal form can be automatically selected from the function library based on the GEP algorithm. Compared with traditional boundary condition modeling methods (such as pure empirical formulas or black box neural networks), this model is a directly output, analyzable mathematical formula that does not require repeated calls to finite element software. At the same time, due to the diversity of basic design functions, it can accurately fit complex mechanical phenomena such as the pressure film effect, thereby achieving a balance between computational efficiency and accuracy. In addition, the initial model is trained using finite element data, and the data is then corrected through subsequent static loading tests, forming a closed-loop modeling process driven by both simulation and testing. The generalization ability is stronger than that of pure data-driven methods in traditional methods.

[0058] S2: Construct composite beam bridge deck specimens and conduct multi-parameter static loading tests to reflect the pressure membrane effect of the composite beam bridge deck during the stress process, and modify the calculation model of the composite beam bridge deck boundary conditions based on the test data.

[0059] In step S2, when performing the static loading test, the multiple parameters that are changed include lateral constraint stiffness, lateral constraint eccentricity, rotational constraint stiffness, plate thickness, and reinforcement ratio;

[0060] The pressure film effect is expressed as:

[0061]

[0062] Where σ p Represents the additional stress of the pressure film, ∈ p represents the strain generated by the pressure film, k p Indicates the sensitivity coefficient of the pressure film effect, k p ∈[0.1,0.5], ∈0 represents the critical strain threshold, and β represents the quadratic term coefficient.

[0063] From the calculation formula of the pressure film, it can be seen that the first-order term in the first half of the formula is suitable for describing the small strain stage, that is, ∈ p<∈0 saturation effect, at this time the stress growth slows down, the second half of the formula is used to describe the large strain stage, that is, ∈ p The geometric nonlinear strengthening effect of ≥∈0, at which point stress increases more rapidly. The critical strain threshold is determined by the cracking threshold of concrete and is generally set between 0.002 and 0.003. The quadratic coefficient is determined by the reinforcement ratio and reflects the contribution of the steel bars. That is, the higher the reinforcement ratio, the larger the quadratic coefficient. Its specific value can be determined based on engineering experience.

[0064] The calculation model of the modified composite beam bridge deck boundary conditions is expressed as:

[0065] BC i ′=c i ·f(P i )+a i ·σ p

[0066] Where BC i ′ represents the parameter size of the modified i-th group of boundary conditions, a i represents the sensitivity coefficient of the i-th group of boundary condition parameters to the pressure film effect, a i ∈[0.05,0.3].

[0067] It can be seen from the revised calculation model that a is superimposed on the original calculation model. i ·σ p To reflect the influence of the pressure membrane effect, the initial value of the sensitivity coefficient can be determined based on engineering experience, and then updated based on the data of the static loading test to quantify the response intensity of the boundary conditions to the membrane effect.

[0068] When constructing composite beam bridge deck specimens, the specimen can be set as a 2m×1.5m composite beam bridge deck specimen (for example, steel beam + concrete slab), and various parameters such as slab thickness and reinforcement ratio can be adjusted according to the orthogonal experimental table. When conducting static loading tests, an MTS hydraulic servo system can be used, equipped with a reaction frame and distribution beam to achieve four-point bending loading. A 10×10 strain gauge grid is arranged at the bottom of the composite beam bridge deck to monitor the strain distribution generated by the pressure film. Displacement sensors are then used to measure mid-span deflection and end slip, and pressure sensors are used to measure support reactions to calculate constraint stiffness.

[0069] The basic steps of the test are set up in the following order:

[0070] Preload: Apply 10% of the ultimate load to eliminate the contact gap;

[0071] Gradual loading: Loading is performed step by step in 20% increments until failure, with each level holding the load for 5 minutes;

[0072] Data collection: record strain, displacement, and crack development under each load level;

[0073] Post-processing: Extract the relationship curve between strain and additional stress generated by the pressure film, and fit k p , β and other parameters.

[0074] During actual testing, it can be adjusted according to engineering requirements or expert experience.

[0075] In this embodiment, the boundary conditions are predicted using a modified calculation model for composite beam bridge deck boundary conditions and a traditional empirical formula. Traditional empirical formulas are mostly based on mechanical theory in structural engineering and a large number of engineering test summaries. They usually use linear or simplified nonlinear functions to express the relationship between composite beam bridge deck boundary condition parameters (such as constraint stiffness) and design variables (such as geometric dimensions and material properties), which are generally obtained through empirical fitting. Here, the lateral constraint stiffness is taken as the research object, and the traditional empirical formula uses a typical linear fitting model. The specific fitting results are shown in the following table:

[0076] Table 1: Lateral constraint stiffness fitting error

[0077]

[0078]

[0079] From this table and Figure 2 It can be seen that the error of the improved GEP model is generally less than 1.5%, indicating that the model can accurately characterize complex boundary conditions and capture the influence of multi-factor coupling. It is superior to traditional empirical formulas and improves the accuracy of design parameters. The error of traditional empirical formulas is generally between 5% and 10% and is relatively large, reflecting that it is based on simplified assumptions and linear approximation, and the GEP model error remains stable, while the error of the empirical formula fluctuates greatly in the high stiffness range, indicating that the empirical formula has poor stability under extreme working conditions, indicating that it is difficult to adapt to the reality of changing boundary conditions and complex structures, and is only suitable for rough estimates under complex working conditions such as pressure membranes.

[0080] This step combines refined experimental design with physical-driven modeling, making the nonlinear thin film effect model applicable to the elastic, plastic, and failure stages simultaneously, achieving full working condition coverage. It can provide highly reliable input for subsequent TCN prediction and GA optimization, forming a complete "experiment-simulation-optimization" technology chain and completing the intelligent design closed loop.

[0081] S3: Simulate the stress process of the composite beam bridge deck based on the revised calculation model and construct a mechanical performance database for the composite beam bridge deck based on the simulation results. The mechanical performance database for the composite beam bridge deck constructed based on the simulation results includes the parameters of the revised boundary conditions, the additional stress of the pressure membrane, the strain generated by the pressure membrane, and the deformation rate of the composite beam bridge deck.

[0082] Specifically, the mechanical properties database can be constructed in the following order:

[0083] Parameter space definition: Determine the parameter range, such as lateral restraint stiffness: 500-2000 kN / mm, plate thickness: 120-250 mm, reinforcement ratio: 0.5%-2.5%, etc. This can be determined based on engineering experience, and Latin hypercube sampling (LHS) is used to generate 10^4 groups of parameter combinations to ensure uniform coverage of the space;

[0084] Parallel simulation calculation: deploy ANSYS parameterized scripts based on HPC clusters and output various parameters;

[0085] Data cleaning and labeling: Eliminate non-convergent conditions and add data labels.

[0086] This step achieves high coverage of actual engineering scenarios through physical model-driven big data generation and intelligent database architecture design, improves data completeness, provides high-quality input for subsequent TCN and GA, and greatly improves the efficiency of overall solution optimization. In addition, data mining can form a reusable design experience library, thereby reducing dependence on expert experience or engineering experience.

[0087] S4: Construct a TCN algorithm model. Based on the TCN algorithm model and the mechanical properties database of composite beam bridge decks, simulate and predict the stress process of composite beam bridge decks.

[0088] In step S4, the input of the TCN algorithm model (i.e., the temporal convolutional network algorithm) is the mechanical properties database of the composite beam bridge deck. The TCN model structure is composed of 4 to 6 layers of causal convolution and dilated convolution, including a residual module to alleviate the gradient vanishing problem, and the convolution kernel uses an adaptive dilation coefficient, which is calculated as follows:

[0089]

[0090] Where d l represents the expansion coefficient of the lth layer, γ represents the growth factor, γ∈[1.5,2], d base represents the initial expansion coefficient, The use of an adaptive expansion coefficient can accurately capture the long-range temporal dependencies of bridge deck stresses (such as creep effects), helping to alleviate the problems of insufficient long-term dependency modeling and difficulty integrating multi-scale features in traditional temporal prediction methods for bridge deck stress analysis.

[0091] In this embodiment, the TCN algorithm model under the above settings and the traditional RNN algorithm model are used to predict the simulated deformation of the finite element. The traditional TCN algorithm model adopts the standard RNN algorithm model (i.e., the standard recurrent neural network model), two hidden layers, the activation function is tanh, the output layer is a fully connected layer, a single prediction value is output, there is no residual module, the loss function uses mean square error, the optimizer uses Adam, and the learning rate is set to 0.001. The specific prediction results are shown in the following table:

[0092] Table 2: Deformation prediction error of finite element simulation

[0093]

[0094]

[0095] From the above table and Figure 3 It can be seen that the error of the TCN model with the introduction of the residual module is stable within 1%, and the error remains basically unchanged as time increases, indicating that TCN effectively captures the long-term temporal dependence and multi-scale characteristics of the structural response, and the prediction results are highly close to the finite element simulation. The traditional RNN error is large (3%-7%) and tends to increase with the increase of time steps, reflecting that RNN has a gradient vanishing problem in long sequence prediction and is difficult to stably and accurately simulate complex nonlinear dynamic responses. Therefore, compared with the RNN model commonly used in the existing technology, in this complex working scenario with a long processing time, the TCN model has a small and stable error, which is better than the traditional method.

[0096] This step, by fusing the adaptive expansion TCN with multimodal data, has higher detection accuracy and faster prediction speed than traditional time series prediction methods. It can analyze and reveal the implicit relationship between design parameters and mechanical responses, and provide a closed-loop intelligent driving capability from data to decision-making for the design of composite beam bridge panels.

[0097] S5: Establish an optimization design framework for the GA algorithm (i.e., genetic algorithm), and embed the calculation model of the composite beam bridge deck boundary conditions and the TCN algorithm model into the GA algorithm. Perform fitness evaluation on the boundary conditions and mechanical properties of the composite beam bridge deck, and search for the optimal design scheme based on the evaluation results to complete the intelligent design of the composite beam bridge deck.

[0098] Specifically, the forward transmission of data between models is the transfer of candidate solutions (such as main beam size) generated by GA to the TCN model to predict mechanical responses; the reverse correction is the feedback of fitness scores to the GEP model, dynamically adjusting the boundary condition parameters, thus forming a closed loop of "optimization-prediction-correction".

[0099] In step S5, the fitness of the composite beam bridge deck's boundary conditions and mechanical properties is evaluated. The initial population of the GA algorithm is set based on design parameters, including the number of girders, girder dimensions, diaphragm spacing, and diaphragm stiffness. The design is then encoded as a multidimensional parameter vector. The modified boundary condition model is used to rapidly calculate constraint reactions, replacing finite element methods and significantly reducing solution time.

[0100] When evaluating the fitness of the boundary conditions and mechanical properties of the composite beam bridge deck, the fitness function expression is:

[0101]

[0102] Where F represents the fitness score, w1, w2, and w3 represent weight coefficients, and w1+w2+w3=1, P represents the carrying capacity of the design scheme, and P req represents the bearing capacity requirement, δ represents the deformation of the design solution, δ limit represents the deformation threshold, M represents the material consumption of the main beam of the design scheme, and M ref Indicates the reference main beam material usage.

[0103] The fitness function uniformly quantifies the bearing capacity, deformation and material usage, and realizes dynamic trade-offs between objectives through weight coefficients. For example, in the preliminary design stage, the safety threshold is prioritized as the optimization target, and w1=0.7, w2=0.2, and w3=0.1 are taken. In the subsequent in-depth design stage, it is necessary to balance the various objectives with each other, and w1=0.4, w2=0.3, and w3=0.3 are taken.

[0104] After the evaluation is completed, the design with the highest fitness score is selected as the optimal design.

[0105] In summary, the present invention innovatively combines the improved genetic expression programming algorithm with finite element simulation and static loading test to achieve high-precision mathematical modeling of the boundary conditions of composite beam bridge panels, breaking through the limitations of traditional empirical formulas and significantly improving the automation and accuracy of boundary condition expression. Combining the efficient timing prediction capability of the time convolutional network algorithm and the intelligent optimization framework of the genetic algorithm, a closed-loop intelligent design system integrating model construction, performance prediction and design optimization is formed, which can quickly and accurately predict the mechanical response of the panel and find the optimal design solution under complex multi-parameter working conditions. This method not only improves design efficiency and accuracy, reduces simulation computing resource consumption, but also enhances the generalization ability of the model and its applicability to actual engineering projects, fully meeting the needs of modern composite beam bridge panel structure design for intelligence, refinement and efficiency.

[0106] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0107] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software depends on the specific application and design constraints of the technical solution.

[0108] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0109] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. An intelligent design method for composite beam bridge deck based on GEP+TCN+GA, characterized in that: The specific steps include: S1: Construct a finite element model of the composite beam bridge, simulate it, build a boundary condition database of the composite beam bridge deck based on the simulation results, and build a calculation model of the composite beam bridge deck boundary conditions based on the GEP algorithm and the boundary condition database; S2: Construct composite beam bridge deck specimens and conduct multi-parameter static loading tests to reflect the pressure membrane effect of the composite beam bridge deck during the load process. Based on the test data, the calculation model of the composite beam bridge deck boundary conditions is modified; S3: Simulate the stress process of the composite beam bridge deck based on the revised calculation model, and build a mechanical performance database of the composite beam bridge deck based on the simulation results; S4: Construct a TCN algorithm model. Based on the TCN algorithm model and the mechanical properties database of composite beam bridge decks, simulate and predict the stress process of composite beam bridge decks. S5: Establish an optimization design framework for the GA algorithm, and embed the calculation model of the composite beam bridge deck boundary conditions and the TCN algorithm model into the GA algorithm. Perform fitness evaluation on the boundary conditions and mechanical properties of the composite beam bridge deck, and search for the optimal design scheme based on the evaluation results to complete the intelligent design of the composite beam bridge deck.

2. The intelligent design method for composite beam bridge deck based on GEP+TCN+GA according to claim 1, characterized in that: In step S1, the boundary conditions include lateral constraint stiffness, lateral constraint eccentricity, and rotational constraint stiffness; When using the GEP algorithm to construct a calculation model for the boundary conditions of composite beam bridge decks, the least squares method, tournament algorithm, and population update strategy are used to improve the GEP algorithm. The specific logic is as follows: Use the least squares method to fit the real constant in the GEP algorithm; Use the tournament algorithm to enhance the population selection pressure and diversity in the GEP algorithm, and set the tournament size range to 3 to 7; The population update strategy is used to improve the global search capability and convergence speed of the GEP algorithm, and the population size in the population update strategy is set in the range of 50 to 200.

3. The intelligent design method for composite beam bridge deck based on GEP+TCN+GA according to claim 2 is characterized by: The calculation model expression of the boundary condition of the composite beam bridge deck is: BC i =c i ·f(P i ) Where BC i Indicates the parameter size of the i-th group of boundary conditions, c i represents the real constant corresponding to the i-th set of boundary conditions, f(P i ) represents the basic design function of the i-th group of boundary conditions, P i The vector representing the i-th set of boundary conditions, where the subscript i represents the index of the boundary condition and m represents the total number of boundary conditions.

4. The intelligent design method for composite beam bridge deck based on GEP+TCN+GA according to claim 3 is characterized by: In step S2, when performing the static loading test, the multiple parameters that are changed include lateral constraint stiffness, lateral constraint eccentricity, rotational constraint stiffness, plate thickness, and reinforcement ratio; The pressure film effect is expressed as: Where σ p Represents the additional stress of the pressure film, ∈ p represents the strain generated by the pressure film, k p Indicates the sensitivity coefficient of the pressure film effect, k p ∈[0.1,0.5], ∈0 represents the critical strain threshold, and β represents the quadratic term coefficient; The calculation model of the modified composite beam bridge deck boundary conditions is expressed as: BC′ i (c i ·f(P i )+a i ·σ p Where BC′ i Indicates the parameter size of the modified i-th group of boundary conditions, a i represents the sensitivity coefficient of the i-th group of boundary condition parameters to the pressure film effect, a i ∈[0.05,0.3].

5. The intelligent design method for composite beam bridge deck based on GEP+TCN+GA according to claim 1 is characterized in that: In step S3, the mechanical properties database of the composite beam bridge deck is constructed based on the simulation results, including the parameter size of the modified boundary conditions, the additional stress of the pressure film, the strain generated by the pressure film, and the deformation rate of the composite beam bridge deck.

6. The intelligent design method for composite beam bridge deck based on GEP+TCN+GA according to claim 5 is characterized by: In step S4, the input of the TCN algorithm model is the mechanical properties database of the composite beam bridge deck. The TCN model structure is composed of 4 to 6 convolutional layers. The convolutional layers include causal convolutional layers and dilated convolutional layers. A residual module is included to alleviate the gradient vanishing problem. The convolution kernel uses an adaptive dilation coefficient, which is calculated as follows: Where d l represents the expansion coefficient of the lth convolutional layer, γ represents the growth factor, γ∈[1.5,2], d base represents the initial expansion coefficient, Indicates rounding down.

7. The intelligent design method for composite beam bridge deck based on GEP+TCN+GA according to claim 1 is characterized by: In step S5, when evaluating the fitness of the boundary conditions and mechanical properties of the composite beam bridge deck, the initial population of the GA algorithm is set based on the design parameters, wherein the design parameters include: the number of main beams, the main beam size, the cross-diaphragm spacing, and the cross-diaphragm stiffness, and the design scheme is encoded as a multidimensional parameter vector.

8. The intelligent design method for composite beam bridge deck based on GEP+TCN+GA according to claim 7 is characterized in that: When evaluating the fitness of the boundary conditions and mechanical properties of the composite beam bridge deck, the fitness function expression is: Where F represents the fitness score, w1, w2, and w3 represent weight coefficients, and w1+w2+w3=1, P represents the carrying capacity of the design scheme, and P req represents the bearing capacity requirement, δ represents the deformation of the design solution, δ limit represents the deformation threshold, M represents the material consumption of the main beam of the design scheme, and M ref Indicates the reference main beam material usage; After the evaluation is completed, the design with the highest fitness score will be regarded as the optimal design.

Images