Mechanical optimization analysis and design system and method for steel-concrete connectors of concrete poles

Abstract

The present invention provides a mechanical optimization analysis and design system and method for steel-concrete connectors of concrete poles, which relates to the technical field of concrete optimization design, including a multi-scale modeling module, a nonlinear material model module, an adaptive grid refinement module, a parallel computing module, a dynamic boundary condition module, and a user interface module. The multi-scale modeling module is used to accurately couple microstructure information to a macro model, which significantly improves the accuracy of the model and makes the simulation results closer to the actual situation. The high-precision modeling is particularly outstanding in capturing the micromechanical properties and overall mechanical behavior of steel-concrete connectors. The system realizes multi-level iterative optimization of the simulation model, ensuring the high accuracy and high reliability of the simulation results. In addition, the simulation result prediction and optimization method based on deep learning significantly shortens the simulation time while maintaining a high prediction accuracy.

Description

Technical Field

[0001] The invention relates to the technical field of concrete optimization design, and in particular to a mechanical optimization analysis and design system and method for a concrete pole steel-concrete connector. Background Art

[0002] According to the Chinese patent publication number "CN107885938B", a concrete mesoscopic model analysis method based on APDL language is disclosed, which belongs to the field of concrete mechanical analysis, and specifically relates to a concrete mesoscopic model analysis method. The present invention uses APDL parametric design language as a platform, inputs the material properties of each component of concrete and the loading method of the specimen, performs modeling and intelligent meshing, then solves the model, judges the unit information read after loading, and if the unit destruction conditions are met, the unit is eliminated, and finally the output results are obtained, including the data of concrete macroscopic indicators and the mechanical behavior and destruction process of concrete.

[0003] According to the simulation method and simulation system for active crack control of continuously reinforced concrete pavement disclosed in Chinese patent publication number "CN114818045B", a simulation method and simulation system for active crack control of continuously reinforced concrete pavement are disclosed. The early damage and cracking behavior of the continuously reinforced concrete pavement is described by non-local theory, and the early change process of the continuously reinforced concrete pavement is regarded as the result of the time-varying influence of environmental loads and concrete material properties. A near-field dynamic model including environmental loads, time-varying mechanical properties of early concrete materials, interaction between longitudinal steel bars and concrete, and fracture damage of concrete materials is established.

[0004] The above patent documents and prior art have the following technical problems when used:

[0005] 1. Insufficient analysis of mechanical behavior under complex loads. Traditional models cannot accurately simulate the actual mechanical behavior of steel and concrete under complex loads. Existing methods often ignore the influence of microstructure on macroscopic mechanical behavior.

[0006] 2. Traditional methods do not adequately simulate the dynamically changing boundary conditions in actual working conditions. Since existing simulations have no iterative solutions, the simulation results may differ from the actual experimental data, and there is no real-time correlation between the two. Summary of the invention

[0007] Technical issues solved:

[0008] In view of the shortcomings of the prior art, the present invention provides a mechanical optimization analysis and design system and method for a concrete pole steel-concrete connector, which solves the following problems:

[0009] 1. Insufficient analysis of mechanical behavior under complex loads. Traditional models cannot accurately simulate the actual mechanical behavior of steel and concrete under complex loads. Existing methods often ignore the influence of microstructure on macroscopic mechanical behavior.

[0010] 2. Traditional methods do not adequately simulate the dynamically changing boundary conditions in actual working conditions. Since existing simulations have no iterative solutions, the simulation results may differ from the actual experimental data, and there is no real-time correlation between the two.

[0011] Technical solution:

[0012] To achieve the above objectives, the present invention is implemented through the following technical solutions: a mechanical optimization analysis and design system for concrete pole steel-concrete connectors, the system comprising the following modules:

[0013] A multi-scale modeling module, used to perform multi-scale modeling of the steel-concrete connector by combining microstructural characteristics with macroscopic mechanical behavior, wherein the multi-scale modeling module establishes a finite element model of the microstructure;

[0014] A nonlinear material model module, for simulating the mechanical behavior of steel and concrete under complex loads by applying a nonlinear material model, wherein the nonlinear material model module responds to the damage mechanical behavior of concrete and the plastic deformation characteristics of steel;

[0015] An adaptive mesh refinement module is used to refine the mesh in the stress concentration area. The adaptive mesh refinement module identifies the stress concentration area and refines the mesh distribution based on the real-time stress analysis results to optimize the simulation results.

[0016] A parallel computing module is used to accelerate the mechanical simulation calculation process. The parallel computing module integrates a parallel computing algorithm based on a multi-core processor and a cloud computing platform to support real-time calculation of large-scale models and multi-field coupling analysis;

[0017] A dynamic boundary condition module is used to simulate the dynamically changing boundary conditions in the actual working conditions. The dynamic boundary condition module collects the actual working condition data from the outside through the real-time sensor data input interface and converts it into the dynamic boundary conditions in the simulation model;

[0018] The user interface module provides an intuitive operation interface to simplify the user's simulation operation. The user interface module integrates intelligent recommendation and adaptive optimization functions to recommend optimal operation processes and parameter settings based on the user's operating habits and simulation requirements.

[0019] Preferably, the multi-scale modeling module further comprises:

[0020] A microstructure analysis unit, used for generating a microstructure model of concrete by image analysis technology or microscope scanning, wherein the model includes aggregate distribution, porosity and crack information;

[0021] The macroscopic mechanical behavior coupling unit is used to apply the output of the microstructure model as boundary condition input parameters to the macroscopic finite element model to couple the microscopic and macroscopic mechanical behaviors.

[0022] Preferably, the nonlinear material model module further comprises:

[0023] The concrete damage model generation unit generates a model for describing the concrete damage accumulation process through experimental data and existing parameters. The model parameters include stress-strain relationship, fracture toughness, and damage evolution.

[0024] The steel plasticity model optimization unit combines experimental data under high temperature or corrosive environment to accurately simulate the plastic deformation behavior of steel, including yield strength, plastic hardening and stress relaxation characteristic parameters.

[0025] Preferably, the adaptive grid refinement module further comprises:

[0026] Real-time stress monitoring unit, which monitors the stress distribution in the model in real time during the simulation process and identifies areas with large changes in stress gradients;

[0027] The adaptive mesh adjustment unit automatically adjusts the mesh density according to the data provided by the real-time stress monitoring unit, refines the stress concentration area, and dynamically optimizes the number and distribution of meshes.

[0028] Preferably, the parallel computing module further comprises:

[0029] The distributed computing scheduling unit uses a load balancing algorithm to distribute simulation computing tasks to multiple computing nodes, and monitors the computing status of each node to ensure load balancing and resource optimization during parallel computing.

[0030] The cloud computing-based simulation collaboration platform supports multiple users in different locations to collaborate on mechanical simulation of large-scale models through the cloud, realizing real-time sharing of simulation data and integration of results.

[0031] Preferably, the dynamic boundary condition module further comprises:

[0032] The sensor data real-time processing unit obtains real-time data of the external environment, including temperature, humidity, and vibration, through an integrated sensor network, and converts it into boundary conditions of the simulation model using data processing algorithms;

[0033] The dynamic condition adaptive optimization unit automatically adjusts and optimizes boundary conditions according to the feedback data in the simulation model.

[0034] Preferably, the user interface module further comprises:

[0035] Intelligent recommendation unit, based on the user's historical operation data and simulation requirements, uses machine learning algorithms to provide intelligent recommendations for parameter settings, improving the efficiency and accuracy of simulation operations;

[0036] The adaptive optimization engine automatically adjusts the simulation process and interface layout based on real-time user input and feedback.

[0037] Preferably, the method for mechanical optimization analysis and design system of the concrete pole steel-concrete connector comprises the following steps:

[0038] Sp1, create a mechanical model of steel-concrete connectors containing microscopic and macroscopic information through a multi-scale modeling module;

[0039] Sp2. Apply nonlinear material models, including concrete damage model and steel plasticity model, to the model to conduct mechanical analysis under complex loads;

[0040] Sp3, using adaptive mesh refinement algorithm, based on real-time stress monitoring data, to optimize the mesh of stress concentration areas in the model;

[0041] Sp4, using parallel computing modules, through cloud computing platforms and distributed computing scheduling units, to accelerate the simulation calculation of large-scale models;

[0042] Sp5, dynamically adjust the boundary conditions in the simulation model according to the external environment data obtained by the sensor;

[0043] Sp6. In the user interface module, intelligent recommendation and adaptive optimization functions are used to simplify user operation processes and improve simulation efficiency.

[0044] Preferably, the method further comprises performing multi-level iterative optimization on the simulation model through an experiment-simulation joint verification platform, comprising the following steps:

[0045] Sp1. Perform preliminary simulation based on the initial simulation model to obtain preliminary result data;

[0046] Sp2, conduct actual load tests on steel-concrete connectors on the experimental platform and collect key mechanical indicators, including stress distribution, crack propagation, and fatigue performance;

[0047] Sp3, compare the experimental data with the preliminary simulation results, and determine the optimization direction of the simulation model through difference analysis;

[0048] Sp4. Use the model parameter calibration module based on the Bayesian optimization algorithm to adjust the material parameters, geometric dimensions and boundary conditions in the simulation model according to the difference analysis results;

[0049] Sp5. Simulate again on the optimized simulation model to obtain improved simulation results;

[0050] Sp6. Repeat the above process until the simulation results and experimental data reach the predetermined consistency standard, thereby ensuring the high accuracy and high reliability of the simulation model.

[0051] Preferably, the method further includes a simulation result prediction and optimization method based on deep learning, comprising the following steps:

[0052] Sp1. In the process of establishing the simulation model, a multi-layer neural network model is constructed to learn the nonlinear mapping relationship between historical simulation data and simulation results;

[0053] Sp2, inputting the initial parameters and partial calculation results of the simulation model, and using the multi-layer neural network model to predict the subsequent result data required for the complete simulation;

[0054] Sp3, by introducing the generative adversarial network (GAN), the prediction results are adaptively optimized to make them more consistent with physical laws;

[0055] Sp4. During the simulation process, dynamically adjust the weights and number of layers of the neural network to ensure the prediction accuracy of the model under different working conditions;

[0056] Sp5. Compare the prediction results of the deep learning model with the intermediate results of the finite element simulation. If the difference exceeds the predetermined threshold, further refined simulation calculations are triggered. Otherwise, the simulation results are directly output, thereby greatly shortening the simulation time while maintaining the simulation accuracy.

[0057] Beneficial effects:

[0058] The present invention provides a mechanical optimization analysis and design system and method for a concrete pole steel-concrete connector. It has the following beneficial effects:

[0059] 1. The present invention uses a multi-scale modeling module to accurately couple microstructure information to the macro model, significantly improving the accuracy of the model and making the simulation results closer to the actual situation. The high-precision modeling is particularly outstanding in capturing the micromechanical properties and overall mechanical behavior of steel-concrete connectors;

[0060] 2. The present invention adopts an adaptive mesh refinement module to ensure high precision in stress concentration areas, while significantly optimizing simulation efficiency and reducing the computational burden in non-critical areas through real-time monitoring and intelligent adjustment of mesh density;

[0061] 3. The present invention adopts parallel computing modules and cloud computing platforms, and the system can effectively distribute computing tasks, realize rapid simulation calculation of large-scale models, especially in multi-field coupling analysis, greatly shorten the calculation time and improve the simulation processing capability;

[0062] 4. The present invention uses an experimental-simulation joint verification platform to systematically implement multi-level iterative optimization of the simulation model, ensuring high precision and high reliability of the simulation results. In addition, the simulation result prediction and optimization method based on deep learning significantly shortens the simulation time while maintaining a high prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a schematic diagram of the hardware architecture of the present invention;

[0064] Figure 2 It is the operation flow chart of the present invention;

[0065] Figure 3 This is a microscopic picture of a case of the present invention. DETAILED DESCRIPTION

[0066] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Specific embodiment one:

[0068] like Figure 1-2 As shown in the figure, the mechanical optimization analysis and design system of concrete pole steel concrete connectors includes the following modules:

[0069] The multi-scale modeling module is used to combine the microstructure characteristics with the macroscopic mechanical behavior to perform multi-scale modeling on the steel-concrete connector. The multi-scale modeling module establishes a finite element model of the microstructure. The multi-scale modeling module further includes:

[0070] A microstructure analysis unit is used to generate a microstructure model of concrete by image analysis technology or microscope scanning. The model includes aggregate distribution, porosity and crack information;

[0071] The macroscopic mechanical behavior coupling unit is used to use the output of the microstructure model as a boundary condition input parameter, which is applied to the macroscopic finite element model to couple the microscopic and macroscopic mechanical behaviors. The image analysis technology used in the microstructure analysis unit: the microstructure image of the concrete is obtained using digital image analysis technology including X-ray tomography (CT) or scanning electron microscopy (SEM). Aggregates, pores and cracks in concrete are identified and segmented through image processing algorithms (including watershed algorithm, threshold segmentation, region growth, etc.); Finite element modeling: by converting these microstructure features into a finite element model, a finite element mesh containing the material distribution characteristics at the microscale is constructed. Specifically, the concrete aggregate is regarded as a rigid body, the pores as a softening zone, and the cracks as a fracture zone, so as to describe the mechanical behavior of the material at the microscale. The macroscopic mechanical behavior coupling unit adopts multi-scale coupling: the output of the microstructure model (including the equivalent elastic modulus of aggregate distribution, crack density, etc.) is input into the macroscopic finite element model as a boundary condition or material parameter. The multi-scale coupling method is used to simulate the response of the connector under macroscopic mechanical behavior at the macroscale.

[0072] The nonlinear material model module is used to simulate the mechanical behavior of steel and concrete under complex loads by applying nonlinear material models. The nonlinear material model module responds to the damage mechanical behavior of concrete and the plastic deformation characteristics of steel. The nonlinear material model module further includes:

[0073] The concrete damage model generation unit generates a model for describing the damage accumulation process of concrete through experimental data and existing parameters. The model parameters include stress-strain relationship, fracture toughness, and damage evolution. Damage accumulation process: The damage model based on continuum mechanics is adopted. The Mazars model or LublinerPlas model describes the evolution of its stress state over time by defining the damage variable of concrete. The damage variable is determined by the nonlinear part of the stress-strain curve, and the expression is as follows:

[0074] ;

[0075] is the residual strength, is the maximum intensity, D is the degree of damage variable, and its final value is between 0 and 1.

[0076] Fracture toughness and damage evolution: The damage evolution curve (including the relationship between crack extension and load) is fitted by test data, and the Griffith criterion in fracture mechanics is used to describe the critical conditions for crack extension.

[0077] The steel plasticity model optimization unit combines experimental data under high temperature or corrosive environment to accurately simulate the plastic deformation behavior of steel, including yield strength, plastic hardening and stress relaxation characteristic parameters;

[0078] Plastic deformation simulation of steel plastic model optimization unit: Use J2 flow theory and hardening model (isotropic hardening and anisotropic hardening model) to model the plastic behavior of steel under stress. The formula is as follows:

[0079] ;

[0080] is the plastic strain rate, is the plasticity multiplier, is the yield function, is stress, is an internal variable.

[0081] Simulation under high temperature or corrosive environment: Combined with the modified yield criterion of temperature effect, the modified vonMises criterion is adopted, and the influence of temperature on the yield strength and elastic modulus of steel is considered. The formula is as follows:

[0082] ;

[0083] is the yield strength at high temperature, is the temperature sensitivity coefficient, is the standard yield strength of steel, T is the temperature, is the reference temperature.

[0084] The adaptive mesh refinement module is used to refine the mesh in the stress concentration area. The adaptive mesh refinement module identifies the stress concentration area and refines the mesh distribution based on the real-time stress analysis results to optimize the simulation results. The adaptive mesh refinement module further includes:

[0085] The real-time stress monitoring unit monitors the stress distribution in the model in real time during the simulation process and identifies areas with large stress gradient changes. The real-time stress monitoring unit monitors the stress distribution: by real-time monitoring the stress field during the simulation process, the stress concentration area is identified using the stress gradient method or the error estimation method. The Zienkiewicz-Zhu error estimation method is used to estimate the stress error, and the degree of mesh refinement is determined based on the size of the error.

[0086] Adaptive mesh adjustment unit, which automatically adjusts mesh density according to data provided by real-time stress monitoring unit, refines stress concentration areas, and dynamically optimizes the number and distribution of meshes;

[0087] Mesh refinement algorithm of adaptive mesh adjustment unit: According to the data of real-time stress monitoring unit, the adaptive mesh generation algorithm (h adaptive method or p adaptive method) is used to refine the mesh of the model. Specifically:

[0088] ;

[0089] is the unit size after refinement, is the initial unit size, is the local error, is the target error, and p is the error order.

[0090] The parallel computing module is used to accelerate the mechanical simulation calculation process. The parallel computing module integrates parallel computing algorithms based on multi-core processors and cloud computing platforms to support real-time calculations and multi-field coupling analysis of large-scale models. The parallel computing module further includes:

[0091] The distributed computing scheduling unit uses a load balancing algorithm to distribute simulation computing tasks to multiple computing nodes, and monitors the computing status of each node to ensure load balancing and resource optimization during parallel computing.

[0092] The distributed computing scheduling unit adopts a task allocation algorithm: using a load balancing algorithm, a Hab distribution or polling scheduling algorithm, to allocate computing tasks to different computing nodes and monitor the computing status of the nodes. The mathematical model of distributed computing is as follows:

[0093] ;

[0094] is the total computation time, is the computation time of the ith node, is the communication time of the ith node.

[0095] The cloud computing-based simulation collaboration platform supports multiple users in different locations to collaborate on mechanical simulation of large-scale models through the cloud, and realizes real-time sharing of simulation data and integration of results; Cloud simulation: Using the cloud computing platform, integrating the storage and processing capabilities of large-scale data, and realizing simulation data sharing and collaboration among multiple users through RESTful API or gRPC protocol. The typical algorithm used by the platform is MapReduce, which is used to process large-scale data sets and accelerate simulation calculations.

[0096] The dynamic boundary condition module is used to simulate the dynamically changing boundary conditions in the actual working conditions. The dynamic boundary condition module collects the actual working condition data from the outside through the real-time sensor data input interface and converts it into the dynamic boundary conditions in the simulation model. The dynamic boundary condition module further includes:

[0097] The sensor data real-time processing unit obtains real-time data of the external environment, including temperature, humidity, and vibration, through an integrated sensor network, and converts it into the boundary conditions of the simulation model using a data processing algorithm; Data acquisition and processing: Collect environmental data through a sensor network (including temperature sensors, strain gauges, accelerometers, etc.), and generate dynamic boundary conditions after filtering out noise through a data fusion algorithm (Kalman filter):

[0098] ;

[0099] is the estimated state, For Carl's burst buff, is the measurement value, H is the measurement matrix, is the location where the dynamic boundary is generated, that is, The boundary conditions are generated again.

[0100] The dynamic condition adaptive optimization unit automatically adjusts and optimizes the boundary conditions according to the feedback data in the simulation model; boundary condition optimization: Based on the data provided by the sensor, the Bayesian optimization algorithm is used to adjust and optimize the boundary conditions in the simulation model so that the model can accurately reflect the actual working condition changes. The objective function of Bayesian optimization is defined as:

[0101] ;

[0102] is the error function of the simulation results, is a weighing parameter with a value of 0-1. : indicates the boundary condition parameters that need to be optimized. : represents all possible boundary condition parameters The feasible domain of A set of possible values, defining the range and limits of the parameter. : represents the error function, which is used to measure the given boundary conditions The difference between the output of the simulation model and the actual sensor data is express The variance of .

[0103] The user interface module provides an intuitive operation interface to simplify the user's simulation operation. The user interface module integrates intelligent recommendation and adaptive optimization functions to recommend the optimal operation process and parameter settings according to the user's operating habits and simulation requirements. The user interface module further includes:

[0104] The intelligent recommendation unit uses machine learning algorithms to provide intelligent recommendations for parameter settings based on the user's historical operation data and simulation requirements, thereby improving the efficiency and accuracy of simulation operations. The intelligent recommendation unit uses a machine learning recommendation algorithm: based on the user's historical operation data, it uses a collaborative filtering algorithm (including matrix decomposition or graph-based algorithms) to recommend simulation parameter settings. The loss function of the recommendation algorithm is expressed as:

[0105] ;

[0106] For actual rating, is the prediction score, P and Q are decomposition matrices, : the set of known user-parameter rating pairs, : Regularization parameter, which controls the influence of the regularization term, and its value is between 0 and 1.

[0107] The adaptive optimization engine automatically adjusts the simulation process and interface layout based on the user's real-time input and feedback; the adaptive optimization engine uses adaptive optimization: combining real-time input data, adaptively adjusts the simulation process and interface layout through dynamic programming or reinforcement learning methods to optimize the user experience. The Qlearning algorithm is used to continuously optimize the interface decision process. The Q function update formula is as follows:

[0108] ;

[0109] is the state action value function, is the learning rate, ranging from 0 to 1. It is an instant reward, with a value between 0 and 1. is the discount factor, with a value between 0 and 1, s: the current state, indicating the current interface layout, simulation process configuration, or the state of user-system interaction, 𝑎: the actions that can be taken in state 𝑠, including adjusting the position of interface elements and modifying the steps of the simulation process, They represent the interface layout, simulation process configuration, or the state of user-system interaction in the new state, and The actions that can be taken include adjusting the position of interface elements and modifying the steps of the simulation process.

[0110] System hardware architecture:

[0111] Central processing unit: uses high-performance multi-core processors (Intel Xeon or AMD EPYC) to support parallel computing and complex data processing.

[0112] Storage unit: Use NVMeSSD storage to support high-speed read and write operations, ensuring fast access to simulation data.

[0113] Sensor interface unit: Contains a variety of sensor interfaces (RS485, Modbus, CANbus) for real-time data acquisition and boundary condition updates.

[0114] Network communication unit: supports Gigabit Ethernet and 5G wireless networks to ensure high-speed data transmission with the cloud.

[0115] Workflow:

[0116] Initial Data Input: The user enters initial material parameters and environmental conditions.

[0117] Multi-scale modeling: The system generates micro and macro models based on input parameters and performs preliminary mechanical simulations.

[0118] Adaptive mesh refinement: monitor stress distribution in real time and dynamically adjust mesh density as needed.

[0119] Parallel computing: Tasks are assigned to different computing nodes and simulation calculations are performed in parallel.

[0120] Dynamic boundary condition update: Dynamically adjust boundary conditions and recalculate simulation results based on real-time sensor data.

[0121] User feedback and optimization: The simulation results are presented through the user interface, and the subsequent simulation process is optimized based on the user operation records. Specific embodiment 2:

[0123] like Figure 1-3 As shown, a method for mechanical optimization analysis and design system of concrete pole steel-concrete connector includes the following steps:

[0124] Sp1, create a mechanical model of steel-concrete connectors containing microscopic and macroscopic information through a multi-scale modeling module;

[0125] Sp2. Apply nonlinear material models, including concrete damage model and steel plasticity model, to the model to conduct mechanical analysis under complex loads;

[0126] Sp3, using adaptive mesh refinement algorithm, based on real-time stress monitoring data, to optimize the mesh of stress concentration areas in the model;

[0127] Sp4, using parallel computing modules, through cloud computing platforms and distributed computing scheduling units, to accelerate the simulation calculation of large-scale models;

[0128] Sp5, dynamically adjust the boundary conditions in the simulation model according to the external environment data obtained by the sensor;

[0129] Sp6. In the user interface module, intelligent recommendation and adaptive optimization functions are used to simplify user operation processes and improve simulation efficiency.

[0130] The method further includes performing multi-level iterative optimization on the simulation model through an experiment-simulation joint verification platform, including the following steps:

[0131] Sp1. Perform preliminary simulation based on the initial simulation model to obtain preliminary result data;

[0132] Sp2, conduct actual load tests on steel-concrete connectors on the experimental platform and collect key mechanical indicators, including stress distribution, crack propagation, and fatigue performance;

[0133] Sp3, compare the experimental data with the preliminary simulation results, and determine the optimization direction of the simulation model through difference analysis;

[0134] Sp4. Use the model parameter calibration module based on the Bayesian optimization algorithm to adjust the material parameters, geometric dimensions and boundary conditions in the simulation model according to the difference analysis results;

[0135] Sp5. Simulate again on the optimized simulation model to obtain improved simulation results;

[0136] Sp6. Repeat the above process until the simulation results and experimental data reach the predetermined consistency standard, thereby ensuring the high accuracy and high reliability of the simulation model.

[0137] The method further includes a simulation result prediction and optimization method based on deep learning, comprising the following steps:

[0138] Sp1. In the process of establishing the simulation model, a multi-layer neural network model is constructed to learn the nonlinear mapping relationship between historical simulation data and simulation results;

[0139] Sp2, input the initial parameters and partial calculation results of the simulation model, and use the multi-layer neural network model to predict the subsequent result data required for the complete simulation;

[0140] Sp3, by introducing the generative adversarial network (GAN), the prediction results are adaptively optimized to make them more consistent with physical laws;

[0141] Sp4. During the simulation process, dynamically adjust the weights and number of layers of the neural network to ensure the prediction accuracy of the model under different working conditions;

[0142] Sp5. Compare the prediction results of the deep learning model with the intermediate results of the finite element simulation. If the difference exceeds the predetermined threshold, further refined simulation calculations are triggered. Otherwise, the simulation results are directly output, thereby greatly shortening the simulation time while maintaining the simulation accuracy.

[0143] according to Figure 3 As can be seen in the onlookers’ images, Figure 3 The spherical structure in the figure represents the distribution of aggregate in concrete or the mechanical response of some key characteristic points of steel-concrete joints during the simulation process. The distribution of the spheres and their mutual connection reflect the stress concentration areas or microstructure interactions that occur during the mechanical analysis process.

[0144] First, the system uses a multi-scale modeling module to create a mechanical model of steel-concrete connections that contains both microscopic and macroscopic information. In this process, the microstructure of concrete (such as Figure 3 The comprehensive mechanical model of steel-concrete connectors is constructed by combining the aggregate distribution represented by the sphere in the figure and the mechanical behavior of the macrostructure. The detailed description of the microstructure can capture the internal characteristics of the material, while the macrostructure integrates these microcharacteristics to form an overall model with practical application value. The advantage of multi-scale modeling is that it can accurately reflect the mechanical behavior at different scales, making the simulation results closer to the actual situation.

[0145] In the created model, the system applied nonlinear material models, including concrete damage models and steel plastic deformation models. The simulation took into account the stress distribution, crack propagation and nonlinear response of the material under complex loads, ensuring that the model can truly simulate the mechanical behavior under actual working conditions, thereby significantly improving the simulation accuracy. On this basis, the system uses an adaptive mesh refinement algorithm to monitor the stress concentration areas during the simulation process in real time and perform fine mesh processing on these key areas. This not only provides sufficient calculation accuracy in the stress concentration areas, but also maintains a reasonable calculation load in non-critical areas, thereby improving the overall simulation efficiency.

[0146] In order to cope with complex computing needs, the system integrates parallel computing modules, and accelerates the simulation calculation of large-scale models through cloud computing platforms and distributed computing scheduling. This not only greatly shortens the simulation time, which is particularly important when dealing with complex models and large-scale data, but also supports multi-field coupling analysis. In addition, the system dynamically adjusts the boundary conditions in the simulation model based on the external environment data obtained by the sensor to ensure that the simulation results are more in line with changes in actual working conditions.

[0147] In the user interface module, the system simplifies the user operation process and further improves the simulation efficiency through intelligent recommendation and adaptive optimization functions. This not only makes complex simulation operations more user-friendly, but also reduces the occurrence of human errors by automatically optimizing parameter settings, thereby improving the convenience and reliability of the overall operation.

[0148] In the verification and optimization of simulation results, the system adopts an experimental-simulation joint verification platform to perform multi-level iterative optimization. Preliminary result data is obtained through preliminary simulation, and then actual load testing is carried out through the experimental platform to collect key mechanical indicators such as stress distribution and crack propagation. The experimental data is compared with the preliminary simulation results, and the optimization direction of the simulation model is determined through difference analysis. The model parameters are adjusted based on the Bayesian optimization algorithm. After multiple iterative optimizations, the difference between the simulation results and the experimental data gradually decreases until the predetermined consistency standard is reached, thereby ensuring the high accuracy and high reliability of the simulation model.

[0149] In addition, the system also introduces a simulation result prediction and optimization method based on deep learning. During the establishment of the simulation model, a multi-layer neural network model is constructed to learn the nonlinear mapping relationship between historical simulation data and simulation results. After inputting the initial parameters of the simulation model and some calculation results, the neural network can predict the subsequent result data required for the complete simulation. In order to further optimize the prediction results, the system introduces a generative adversarial network (GAN) to adaptively optimize the prediction results to make them more in line with physical laws. During the simulation process, the system also dynamically adjusts the weights and number of layers of the neural network to ensure the prediction accuracy of the model under different working conditions. If the difference between the prediction results of the deep learning model and the intermediate results of the finite element simulation exceeds the predetermined threshold, the system will trigger further refined simulation calculations; otherwise, the simulation results will be output directly. This method can significantly shorten the simulation time while maintaining the simulation accuracy.

[0150] In summary, the entire system greatly improves the accuracy and efficiency of steel-concrete connection simulation through a series of steps and methods such as multi-scale modeling, nonlinear material models, adaptive mesh refinement, parallel computing, experimental-simulation joint verification, and deep learning prediction and optimization, ensuring a high degree of consistency between the model and actual working conditions. In large-scale engineering applications, this system has demonstrated strong advantages, can effectively solve complex mechanical problems, and provides reliable technical support for engineering design and optimization.

[0151] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to the process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprising a reference structure" do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0152] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and alterations may be made to the embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. Mechanical optimization analysis and design system for steel-concrete connectors of concrete poles, characterized by: The system includes the following modules: The multi-scale modeling module is used to perform multi-scale modeling on the steel-concrete connector by combining the microstructure characteristics and the macroscopic mechanical behavior. The multi-scale modeling module establishes a finite element model of the microstructure. The multi-scale modeling module includes: A microstructure analysis unit, used for generating a microstructure model of concrete by image analysis technology or microscope scanning, wherein the model includes aggregate distribution, porosity and crack information; Macroscopic mechanical behavior coupling unit, used to apply the output of the microstructure model as boundary condition input parameters to the macroscopic finite element model to couple microscopic and macroscopic mechanical behaviors; A nonlinear material model module is used to simulate the mechanical behavior of steel and concrete under complex loads by applying a nonlinear material model, wherein the nonlinear material model module responds to the damage mechanical behavior of concrete and the plastic deformation characteristics of steel; the nonlinear material model module further includes: The concrete damage model generation unit generates a model for describing the damage accumulation process of concrete through experimental data and existing parameters. The model parameters include stress-strain relationship, fracture toughness, and damage evolution. Damage accumulation process: The damage model based on continuum mechanics is adopted. The Mazars model or LublinerPlas model describes the evolution of its stress state over time by defining the damage variable of concrete. The damage variable is determined by the nonlinear part of the stress-strain curve, and the expression is as follows: ； is the residual strength, is the maximum intensity, D is the degree of damage variable, and its final value is between 0 and 1; Fracture toughness and damage evolution: Fitting the damage evolution curve through test data, which includes the relationship between crack growth and load, and using the Griffith criterion in fracture mechanics to describe the critical conditions for crack growth; The steel plasticity model optimization unit combines experimental data under high temperature or corrosive environment to accurately simulate the plastic deformation behavior of steel, including yield strength, plastic hardening and stress relaxation characteristic parameters; Plastic deformation simulation of steel plastic model optimization unit: Use J2 flow theory and hardening model, the hardening model is isotropic hardening and anisotropic hardening model, to model the plastic behavior of steel under stress; the formula is as follows: ； is the plastic strain rate, is the plasticity multiplier, is the yield function, is stress, is the internal variable; Simulation under high temperature or corrosive environment: Combined with the modified yield criterion of temperature effect, the modified vonMises criterion is adopted, and the influence of temperature on the yield strength and elastic modulus of steel is considered; the formula is as follows: ； is the yield strength at high temperature, is the temperature sensitivity coefficient, is the standard yield strength of steel, T is the temperature, is the reference temperature; An adaptive mesh refinement module is used to refine the mesh in the stress concentration area. The adaptive mesh refinement module identifies the stress concentration area and refines the mesh distribution based on the real-time stress analysis results to optimize the simulation results. The adaptive mesh refinement module further includes: The real-time stress monitoring unit monitors the stress distribution in the model in real time during the simulation process and identifies the areas with large stress gradient changes. The real-time stress monitoring unit monitors the stress distribution: by real-time monitoring the stress field during the simulation process, the stress concentration area is identified using the stress gradient method or the error estimation method. The Zienkiewicz-Zhu error estimation method is used to estimate the stress error, and the degree of mesh refinement is determined according to the size of the error. Adaptive mesh adjustment unit, which automatically adjusts mesh density based on data provided by real-time stress monitoring unit, refines stress concentration areas, and dynamically optimizes the number and distribution of meshes; Mesh refinement algorithm of the adaptive mesh adjustment unit: According to the data of the real-time stress monitoring unit, the model is meshed by an adaptive mesh generation algorithm, which is the h adaptive method or the p adaptive method. Specifically: ； is the unit size after refinement, is the initial unit size, is the local error, is the target error, p is the error order; A parallel computing module is used to accelerate the mechanical simulation calculation process. The parallel computing module integrates a parallel computing algorithm based on a multi-core processor and a cloud computing platform to support real-time calculation of large-scale models and multi-field coupling analysis; A dynamic boundary condition module is used to simulate the dynamically changing boundary conditions in the actual working conditions. The dynamic boundary condition module collects the actual working condition data from the outside through the real-time sensor data input interface and converts it into the dynamic boundary conditions in the simulation model; A user interface module provides an intuitive operation interface to simplify the user's simulation operation. The user interface module integrates intelligent recommendation and adaptive optimization functions to recommend optimal operation procedures and parameter settings based on the user's operating habits and simulation requirements; The method of the mechanical optimization analysis and design system further includes a simulation result prediction and optimization method based on deep learning, comprising the following steps: Sp1. In the process of establishing the simulation model, a multi-layer neural network model is constructed to learn the nonlinear mapping relationship between historical simulation data and simulation results; Sp2, inputting the initial parameters and partial calculation results of the simulation model, and using the multi-layer neural network model to predict the subsequent result data required for the complete simulation; Sp3, by introducing the generative adversarial network, the prediction results are adaptively optimized to make them more consistent with physical laws; Sp4. During the simulation process, dynamically adjust the weights and number of layers of the neural network to ensure the prediction accuracy of the model under different working conditions; Sp5. Compare the prediction results of the deep learning model with the intermediate results of the finite element simulation. If the difference exceeds the predetermined threshold, further refined simulation calculations are triggered. Otherwise, the simulation results are directly output, thereby greatly shortening the simulation time while maintaining the simulation accuracy.

2. The mechanical optimization analysis and design system for concrete pole steel-concrete connectors according to claim 1 is characterized in that: The parallel computing module further comprises: The distributed computing scheduling unit uses a load balancing algorithm to distribute simulation computing tasks to multiple computing nodes, and monitors the computing status of each node to ensure load balancing and resource optimization during parallel computing. The cloud computing-based simulation collaboration platform supports multiple users in different locations to collaborate on mechanical simulation of large-scale models through the cloud, realizing real-time sharing of simulation data and integration of results.

3. The mechanical optimization analysis and design system for concrete pole steel-concrete connectors according to claim 1 is characterized in that: The dynamic boundary condition module further comprises: The sensor data real-time processing unit obtains real-time data of the external environment, including temperature, humidity, and vibration, through an integrated sensor network, and converts it into boundary conditions of the simulation model using data processing algorithms; The dynamic condition adaptive optimization unit automatically adjusts and optimizes boundary conditions according to the feedback data in the simulation model.

4. The mechanical optimization analysis and design system for concrete pole steel-concrete connectors according to claim 1 is characterized in that: The user interface module further comprises: Intelligent recommendation unit, based on the user's historical operation data and simulation requirements, uses machine learning algorithms to provide intelligent recommendations for parameter settings, improving the efficiency and accuracy of simulation operations; The adaptive optimization engine automatically adjusts the simulation process and interface layout based on real-time user input and feedback.

5. A mechanical optimization analysis and design method for a concrete pole steel-concrete connector, based on the system described in any one of claims 1 to 4, characterized in that: The method comprises the following steps: Sp1, create a mechanical model of steel-concrete connectors containing microscopic and macroscopic information through a multi-scale modeling module; Sp2. Apply nonlinear material models, including concrete damage model and steel plasticity model, to the model to conduct mechanical analysis under complex loads; Sp3, using adaptive mesh refinement algorithm, based on real-time stress monitoring data, to optimize the mesh of stress concentration areas in the model; Sp4, using parallel computing modules, through cloud computing platforms and distributed computing scheduling units, to accelerate the simulation calculation of large-scale models; Sp5, dynamically adjust the boundary conditions in the simulation model according to the external environment data obtained by the sensor; Sp6. In the user interface module, intelligent recommendation and adaptive optimization functions are used to simplify user operation processes and improve simulation efficiency.

6. The mechanical optimization analysis and design method of the concrete pole steel-concrete connector according to claim 5 is characterized in that: The method further includes performing multi-level iterative optimization on the simulation model through an experiment-simulation joint verification platform, including the following steps: Sp1. Perform preliminary simulation based on the initial simulation model to obtain preliminary result data; Sp2, conduct actual load tests on steel-concrete connectors on the experimental platform and collect key mechanical indicators, including stress distribution, crack propagation, and fatigue performance; Sp3, compare the experimental data with the preliminary simulation results, and determine the optimization direction of the simulation model through difference analysis; Sp4. Use the model parameter calibration module based on the Bayesian optimization algorithm to adjust the material parameters, geometric dimensions and boundary conditions in the simulation model according to the difference analysis results; Sp5. Simulate again on the optimized simulation model to obtain improved simulation results; Sp6. Repeat the above process until the simulation results and experimental data reach the predetermined consistency standard, thereby ensuring the high accuracy and high reliability of the simulation model.

Images