End-to-end computational method for real-time prediction of explosion damage effects on RC components

Abstract

The present invention discloses an end-to-end calculation method for real-time prediction of explosion damage effect of RC components, comprising the following steps: establishing a numerical solution method for explosion damage effect of RC components, obtaining numerical simulation data of explosion damage effect of RC components; associating the numerical simulation data of explosion damage effect of RC components with explosion conditions and geometric shape, size and material properties of RC components, obtaining an explosion damage effect database of RC components including a training data set and a verification data set; using the training data set and the verification data set to train and optimize the constructed neural network, obtaining an end-to-end calculation model of explosion damage effect of RC components; using the end-to-end calculation model of explosion damage effect of RC components, completing the real-time prediction of explosion damage effect of target RC components. The method for predicting explosion damage effect of RC components of the present invention has a simple process, fast speed and good suitability.

Description

Technical Field

[0001] The invention relates to the technical field of data processing, in particular to an end-to-end calculation method for real-time prediction of explosion damage effects of RC components. Background Art

[0002] Blasting at the weak points of the target reinforced concrete (RC) components is a more efficient blasting method with controllable explosion effects. RC component explosion damage refers to the destruction and damage to the building structure and its components caused by the physical effects of the explosion, such as shock waves, high temperature, and high pressure.

[0003] The prediction of explosion damage effect on RC components refers to the analysis and deduction of the damage degree of RC components (such as columns, beams, floor slabs, etc.) in explosion events through the establishment of mathematical models and algorithms, and the identification of component weak points accordingly.

[0004] At present, regarding the prediction of RC component damage, as described in the Chinese invention patent application "Engineering Assessment Method, System, Equipment and Medium for Building Explosion Damage Effect" (Publication No.: CN116029156A, Publication Date: 2023-04-28), it discloses the following steps: According to the building structural parameters, component structural parameters and material performance parameters, the dynamic response formula of the building components in adjacent rooms and a single room is obtained through numerical simulation; based on the dimensional analysis method and the principle of consistent dynamic response of components, the engineering calculation formula of the equivalent amount of explosive propagation coefficient in different directions is established; then, according to the TNT equivalent of the explosion load on the building and the location of the explosion, the explosion equivalent TNT equivalent in different directions in different rooms is obtained and converted into the dynamic response of the building components in each room, and combined with the building damage level standard, the damage level of the building components, rooms and the entire building is finally determined. However, the process of obtaining the dynamic response formula through numerical simulation in this patent is complicated, and the process of establishing the engineering calculation formula based on the dimensional analysis method and the principle of consistent dynamic response of components is cumbersome. This method requires a comprehensive and complete professional technical reserve. In addition, the calculation process of converting the explosion load on the building into the dynamic response of the building components is slow. When the explosion equivalent and explosion position change, the dynamic response of the building components needs to be recalculated. This method is difficult to adaptively and quickly complete the calculation of the explosion damage effect of building components. Considering the real-time requirements for post-explosion rescue, the existing methods are difficult to respond quickly, especially in emergency response scenarios. The calculation efficiency of the existing methods is low and it is impossible to give an accurate prediction of the explosion damage effect of building components in a short time.

[0005] In summary, the problems existing in the prior art are that the prediction process of explosion damage effect of RC components is complex, slow and has poor adaptability. Summary of the invention

[0006] The purpose of the present invention is to provide an end-to-end calculation method for real-time prediction of explosion damage effect of RC components, which has simple process, fast speed and good adaptability.

[0007] The technical solution for achieving the purpose of the present invention is:

[0008] In a first aspect, an end-to-end calculation method for real-time prediction of explosion damage effect of RC components is provided, comprising the following steps:

[0009] Establish a numerical solution method for the explosion damage effect of RC components and obtain numerical simulation data of the explosion damage effect of RC components;

[0010] Associating the numerical simulation data of explosion damage effect of the RC component with the explosion condition and the geometric shape, size and material properties of the RC component to obtain an explosion damage effect database of the RC component including a training data set and a verification data set;

[0011] The constructed neural network is trained and optimized using the training data set and the validation data set to obtain an end-to-end calculation model of explosion damage effects of RC components;

[0012] The end-to-end calculation model of explosion damage effect of RC components is used to complete the real-time prediction of explosion damage effect of target RC components.

[0013] Furthermore, the step of acquiring the numerical simulation data of the explosion damage effect of the RC component includes:

[0014] According to the principle of explosion mechanics, the energy release function of explosives at different equivalents and densities over time is established, and the initial conditions of the explosion center are determined in combination with the physical and chemical properties of the explosives to obtain the physical model of the explosion source; the initial conditions of the explosion center include the initial pressure, temperature, energy density and initial expansion velocity of the explosion center;

[0015] According to the type of explosion source and its surrounding environment information, a nonlinear equation based on fluid mechanics is introduced to describe the reflection, scattering, diffraction and attenuation effects of shock waves during propagation. Combined with the physical law of explosion shock wave propagation in the air, the mathematical equation of explosion shock wave propagation is obtained.

[0016] After the explosion source physical model and the explosion shock wave propagation mathematical equation are combined, they are converted into finite element equations using the Galerkin method, and the field variables in the calculation domain are interpolated and calculated using the three-dimensional first-order four-node element shape function;

[0017] The finite element equation is iterated in a discrete manner to form a shock wave-component fluid-solid coupling calculation form, and the component stress state formed by the shock wave on the surface of the RC component is obtained, including the overpressure peak value, impulse, and duration;

[0018] According to the material characteristics and mechanical properties of RC components, the damage thresholds of different RC components under different stress states are determined;

[0019] According to the damage threshold of RC components, the component damage mode is analyzed and a numerical solution method for the explosion damage effect of RC components is established;

[0020] A three-dimensional structured grid of RC components is made, the material properties of each RC component in the grid are specified, and the numerical solution method of RC component explosion damage effect is used to numerically simulate the damage effect of RC components under explosion loads to generate numerical simulation data of RC component explosion damage effect.

[0021] Furthermore, the step of acquiring the RC component explosion damage effect database includes:

[0022] The numerical simulation data of the explosion damage effect of the RC component is divided according to the shock waves generated under different explosion sources and explosion equivalent conditions;

[0023] Correlate the partitioning results with the corresponding explosion conditions and the geometry, dimensions, and material properties of the RC components;

[0024] The explosion conditions in the correlation results and the geometric shape, size, and material properties of the RC components are trimmed, normalized, and feature extracted to obtain the explosion vector of multi-source information fusion;

[0025] According to the explosion conditions and RC components involved in different scenarios, the explosion vector of multi-source information fusion in each scenario is obtained, and the numerical simulation data of explosion damage effect of all scenarios and corresponding RC components are stacked. The explosion vector of multi-source information fusion is used as input, and the numerical simulation data of explosion damage effect of RC components is used as the output set, which is entered into the library system;

[0026] The set is divided to obtain a RC component explosion damage effect database including a training data set and a verification data set.

[0027] Furthermore, the neural network includes a graph autoencoder layer, a multi-stream graph neural network layer, and a multi-stream merge prediction network layer:

[0028] In the graph self-encoding layer, six layers of graph convolutional networks GCN are cascaded, and a ReLU activation function is connected after each layer of GCN to output the encoding result G0;

[0029] In the multi-stream graph neural network layer, the multi-head attention mechanism in the Transformer encoder model is improved to form a multi-head local attention mechanism. In the multi-head local attention mechanism, the query vector is only matched with the key-value data pairs of the adjacent nodes in the graph structure; the three heads of the multi-head local attention mechanism are strain attention mechanism, stress attention mechanism and material property attention mechanism, and the inputs are all G0, and the outputs are the graph node abstract strain vector G n1 , graph node abstract stress vector G n2 、Graph node abstract material characteristic vector G n3 ;

[0030] In the multi-stream merging prediction network layer, a three-layer fully connected neural network is used to integrate the G output of the multi-stream graph neural network layer. n1 , G n2 and G n3 , and then the explosion damage effect of RC components is obtained through two layers of cascaded GCN, and a ReLU activation function is connected after each layer of GCN.

[0031] Furthermore, the real-time prediction step of the explosion damage effect of the target RC component includes:

[0032] Collect the geometry, size, material properties and on-site explosion conditions of RC components to obtain on-site test data;

[0033] The field test data is input into the end-to-end calculation model of the explosion damage effect of the RC component, and the prediction result of the explosion damage effect of the target RC component is output.

[0034] In a second aspect, an end-to-end computing system for real-time prediction of explosion damage effects of RC components is also provided, comprising:

[0035] A solution unit is used to establish a numerical solution method for explosion damage effect of RC components and obtain numerical simulation data of explosion damage effect of RC components;

[0036] A data processing unit is used to associate the numerical simulation data of explosion damage effect of the RC component with the explosion condition and the geometric shape, size and material properties of the RC component to obtain an explosion damage effect database of the RC component including a training data set and a verification data set;

[0037] A model training unit, used to train and optimize the constructed neural network using the training data set and the verification data set to obtain an end-to-end calculation model of explosion damage effect of RC components;

[0038] The output unit is used to complete the real-time prediction of the explosion damage effect of the target RC component by using the end-to-end calculation model of the explosion damage effect of the RC component.

[0039] According to a third aspect, a computer-readable storage medium storing one or more programs is provided. The one or more programs include instructions. When the instructions are executed by a computing device, the computing device executes the method as described above.

[0040] In a fourth aspect, an electronic device is also provided, comprising one or more processors, one or more memories and one or more programs, wherein the one or more programs are stored in the one or more memories and are configured to be executed by the one or more processors, and the one or more programs include instructions for executing the method as described above.

[0041] Compared with the prior art, the present invention has the following significant advantages:

[0042] 1. Simpler process: The real-time prediction method of explosion damage effect of RC components adopts an end-to-end deep learning model with explosion overpressure data as input and RC component damage effect as output. It does not require complicated professional knowledge reserves or complex formula reasoning, and can directly give damage effect prediction;

[0043] 2. Faster prediction: Different from the traditional finite element analysis numerical simulation layer-by-layer reasoning method, the real-time prediction method of RC component explosion damage effect adopts an end-to-end prediction mode. After deep learning based on a large amount of numerical simulation data of the RC component explosion damage effect physical model, the prediction method of this patent can directly output the explosion overpressure data as the RC component damage effect in one step, eliminating the layer-by-layer reasoning calculation process;

[0044] 3. Good adaptability: The present invention establishes a database of multi-scenario explosion damage effects of RC components, effectively trains the graph neural network model, realizes the rapid learning and generalization capabilities of different equivalents, different explosion positions, and different component geometric features, and realizes the prediction of the damage effects of multi-scenario and multi-dimensional explosion impact target RC components.

[0045] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is the main flow chart of the present invention.

[0047] Figure 2 It is a flow chart of the steps for numerical simulation of explosion damage effects on RC components.

[0048] Figure 3 It is a flow chart of the steps for establishing a database of explosion damage effects of RC components.

[0049] Figure 4 It is a flow chart of the steps for building an end-to-end calculation model for explosion damage effects on RC components.

[0050] Figure 5 It is a structural diagram of the autoencoder network in the end-to-end calculation model structure of the explosion damage effect of RC components.

[0051] Figure 6 It is a multi-stream graph neural network.

[0052] Figure 7 It is a multi-stream merging prediction network.

[0053] Figure 8 It is a flow chart of the steps for predicting explosion damage effects on RC components. DETAILED DESCRIPTION

[0054] like Figure 1 As shown, the end-to-end calculation method for real-time prediction of explosion damage effect of RC components of the present invention comprises the following steps:

[0055] S10. Numerical simulation of explosion damage effect on RC components: Establish a physical model of the explosion source, and obtain numerical simulation data of explosion damage effect on RC components by solving the mathematical equation of shock wave propagation expressed by the finite element equation;

[0056] like Figure 2 As shown, the numerical simulation steps of the explosion damage effect of the RC component include:

[0057] S101. Establishing a physical model of the explosion source: Based on the principles of explosion mechanics, establish the energy release function of the explosive at different equivalents and densities over time, combine the physical and chemical properties of the explosive, determine the initial conditions of the explosion center, including the initial pressure, temperature, energy density and initial expansion velocity of the explosion center, and obtain the physical model of the explosion source. The physical model of the explosion source is:

[0058] Assume the total energy released by the explosion is E0 and the volume is V0, then the initial conditions are:

[0059] (1) Initial density distribution:

[0060] ,

[0061] In the formula, (r) is the local change determined by the mass distribution of the material released by the explosion, and r is the distance from the center of the explosion (the source point) to any point in space.

[0062] (2) Initial pressure distribution:

[0063] Taking spherical charge as an example, its initial pressure is symmetrically distributed:

[0064] ,

[0065] Where R0 is the characteristic radius of the explosion source.

[0066] (3) Initial velocity distribution:

[0067] The radial velocity at the moment of explosion is:

[0068] ,

[0069] In the formula, is the normalized radial distance.

[0070] (4) Explosive state equation:

[0071] ,

[0072] In the formula, is the detonation pressure; is the relative volume; , , , and is the material constant. The specific parameters are shown in Table 1.

[0073] Table 1 Specific parameters

[0074] .

[0075] S102. Mathematical description of shock wave propagation: According to the type of explosion source and its surrounding environment information, a nonlinear equation based on fluid mechanics is introduced to describe the reflection, scattering, diffraction and attenuation effects of shock waves during propagation. Combined with the physical law of explosion shock wave propagation in the air, the mathematical equation of explosion shock wave propagation is obtained; the shock wave control equation is:

[0076] (1) Mass conservation equation:

[0077] ,

[0078] where ρ is the air density, =( ) is the velocity vector and t is the time.

[0079] (2) Momentum conservation equation:

[0080] ,

[0081] Where p is the pressure.

[0082] (3) Energy conservation equation:

[0083] ,

[0084] in, , total energy density, e is the internal energy per unit mass.

[0085] (4) Gas state equation:

[0086] The state equation of air as an ideal gas is:

[0087] ,

[0088] in, ~ is a constant related to the properties of air; is the air pressure; is the relative density, , and are the density and initial density of air respectively;

[0089] S103, mathematical equation finite element calculation: after the explosion source physical model and the explosion shock wave propagation control equation are combined, the continuous mathematical equation is converted into a finite element equation using the Galerkin method, and the field variables in the calculation domain are interpolated and calculated using a three-dimensional first-order four-node element shape function;

[0090] S104, determining component stress state: iterating the finite element equation in a discrete manner to form a shock wave-component fluid-solid coupling calculation form, and obtaining the component stress state formed by the shock wave on the component surface, including overpressure peak value, impulse, and duration;

[0091] S105. Determine component damage threshold: determine the damage threshold of different components under different stress states according to the material characteristics and mechanical properties of the RC components;

[0092] Study the material properties of the main components of the facility (such as beams, columns, wall panels, etc.), including their basic mechanical properties such as compression, tension, and shear. Based on the material mechanical parameters, determine the damage threshold of the components under different stress states to provide a physical basis for subsequent damage analysis;

[0093] S106. Component damage analysis: Based on the damage threshold of RC components, combined with information such as explosion equivalent, explosion location, component stress, etc., according to the safety standards and design specifications in actual projects, the component damage mode is analyzed. The calculation formula is briefly as follows:

[0094] ,

[0095] Where D is the damage factor, E is the explosion equivalent, R is the explosion position, T is the proportional distance, is the component damage threshold, and is the stress state of the component. The corresponding component damage criterion is shown in Table 2.

[0096] Table 2 Component damage criteria

[0097] .

[0098] S107, establishing a numerical solution method for RC component damage effect: combining the component stress state determination step S104 and the RC component damage analysis step S106, a numerical solution method for RC component explosion damage effect is established.

[0099] S108. Acquisition of component explosion damage effect simulation data: Create a three-dimensional structured grid of RC components, specify the material properties of each RC component in the three-dimensional grid, perform numerical simulation on the damage response of RC components under explosion loads using an established numerical solution method, and carry out parallel calculation to generate simulation data on the explosion damage effect of RC components under different explosion conditions.

[0100] S20. Establishment of a database of explosion damage effects of RC components: associating the numerical simulation data of explosion damage effects of RC components with explosion conditions, RC component geometry, and material information, and performing multi-source data feature extraction and fusion to obtain a database of explosion damage effects of RC components including a training data set and a verification data set.

[0101] like Figure 3 As shown, the steps of establishing the RC component explosion damage effect database include:

[0102] S201, dividing the RC component explosion damage data: dividing the RC component explosion damage effect numerical simulation data into shock waves generated under different explosion sources and explosion equivalent conditions and damage effects of corresponding RC components;

[0103] S202, data association: associating the RC component damage data with corresponding explosion conditions, component geometry, size, and material characteristic parameters; the explosion conditions include explosion equivalent and explosion location;

[0104] The preliminary association of data features was completed, forming a corresponding relationship between the explosion conditions, component geometry, size, material property input data and the RC component damage output data, ensuring that the subsequent damage prediction model can predict the damage effect based on the building structure characteristics and explosion conditions.

[0105] S203, multi-source data feature extraction and fusion: extract multi-source data on material properties, explosion equivalent, explosion location and explosion impact load under different explosion conditions, fuse multi-source data with heterogeneous features, and realize effective integration of data of different types and dimensions;

[0106] The step of S203, extracting and fusing multi-source data features, includes:

[0107] S2031. Data pruning: Perform data pruning on the vectors of component material properties, explosion equivalent, and explosion position, and integrate vectors of different dimensions into the same dimension.

[0108] S2032, data normalization: After data pruning is completed, the data needs to be normalized to the minimum and maximum values. This method scales all features to the range of 0 and 1. The calculation formula can be expressed as:

[0109] ,

[0110] In the formula Represents the original data, Represents normalized data.

[0111] S2033. Feature extraction: Use a multi-layer convolutional neural network to downsample and encode the normalized data, remove redundant information, and extract the potential invariant features of the data.

[0112] S2034, multi-source data fusion: connecting the encoded material characteristics, explosion equivalent, and explosion position vectors along one dimension to obtain an explosion vector fused with multi-source information;

[0113] S204, data set generation and entry: each scene is sequentially processed through a multi-source data fusion method to obtain a multi-source explosion scene fusion vector for the corresponding scene, and the results of all scenes are stacked to form a multi-scene multi-source information fusion explosion scene vector data set. All data are divided into a training data set and a verification data set in a ratio of 7 to 3, and then entered into the library system to establish a database of explosion damage effects on building components.

[0114] This completes the data preparation before training the end-to-end calculation model of the explosion damage effect of RC components.

[0115] S30. Construction of an end-to-end computational model for explosion damage effects of RC components: construct a neural network model structure consisting of a graph autoencoder network, a multi-stream graph neural network, and a multi-stream merging prediction network, and use the training data set and the verification data set to train and optimize the neural network model structure to obtain an end-to-end computational model for explosion damage effects of RC components.

[0116] like Figure 4 As shown, the steps of constructing the end-to-end calculation model of explosion damage effect of RC components include:

[0117] S301, neural network model construction: construct the neural network model structure of the graph autoencoder network, the multi-stream graph neural network and the multi-stream merge prediction network. The specific construction steps include:

[0118] S3011, Figure self-coding layer construction: Figure 5As shown in the figure, a graph convolutional network (GCN) is used as the basic module of the encoder. The encoder contains six layers of GCN. The first two layers are used for feature extraction, the last layer is used to capture the relationship between nodes, and the remaining layers are used to generate the mean and variance of the latent variables. After each layer, a ReLU activation function is connected to construct a graph self-encoding layer to improve the nonlinearity of the model. The encoding result obtained after the input data is calculated by the graph self-encoding layer can be represented by G0.

[0119] S3012, Multi-flow graph neural network construction: Figure 6 As shown in Figure 1, the multi-head attention mechanism in the Transformer encoder model is improved to form a multi-head local attention mechanism. In the new attention mechanism, the query vector is no longer matched with all key-value data pairs, but only with the key-value data pairs of adjacent nodes in the graph structure. The three heads of the multi-head local attention mechanism output the abstract strain vector G of the graph node respectively. n1 , graph node abstract stress vector G n2 、Graph node abstract material characteristic vector G n3 .

[0120] S3013, multi-stream merging prediction network construction: Figure 7 As shown in the figure, a three-layer fully connected neural network is used to integrate the different attention vectors output by the multi-flow graph neural network; finally, the information is restored through two layers of GCN and ReLU functions to obtain the component damage effect data.

[0121] S302, neural network model training and structure optimization: use the training data set and the verification data set to train and optimize the neural network model to obtain an end-to-end real-time prediction model for explosion damage effects of RC components.

[0122] Through repeated pre-training and error analysis, the training parameters and processing layer structure of the model are continuously tuned. The training parameters mainly include training batches, loss functions, and number of iterations. The processing layer structure tuning mainly includes the number of GN blocks, the number of nodes in each GN block, and the size of the hidden layer h in the attention mechanism.

[0123] The neural network model training and structure tuning steps include:

[0124] S3021. Model pre-training: Use the Adam algorithm to pre-train the model in a supervised manner with the help of prepared input data and output data.

[0125] The model parameters are randomly generated at the beginning of training, and the back-propagation algorithm is needed to transfer the value of the loss function in reverse and calculate the gradient of the model parameters relative to the loss value to update the model parameters.

[0126] S3022, select training parameters: monitor the model pre-training status, such as computing memory usage, graphics processor memory usage, and model convergence performance information. When model training cannot be implemented normally, reselect model training parameters;

[0127] S3023. Adjust the model structure: By adjusting the number of fully connected layers and the number of heads of the multi-head attention mechanism in the model, check the mean square error loss (MSE) and mean absolute error (MAE) of the model output results to determine whether both have converged and decreased;

[0128] S3024. Repeat steps S3021, S3022 and S3023 until the MSE and MAE converge and decrease, obtain the optimal model training and structural parameters, and obtain the end-to-end calculation model of the explosion damage effect of RC components.

[0129] S40, prediction of explosion damage effect of target RC component: according to the target RC component and the on-site explosion condition, a field explosion input data set is established, and the data of the field explosion input data set is input into the optimized RC component explosion damage effect end-to-end calculation model to predict the damage effect of the target component.

[0130] like Figure 8 As shown, the target RC component explosion damage effect prediction step includes:

[0131] S401, on-site RC component information and explosion condition collection: collect on-site component geometric parameters, material parameters, and explosion conditions such as explosion equivalent and explosion position to obtain on-site test data;

[0132] S402, preprocessing of on-site collected data: Based on the on-site RC component test data, a multi-source data feature extraction and fusion method is used to preprocess the on-site test data to obtain an on-site explosion input data set for an end-to-end calculation model of explosion damage effects of RC components;

[0133] S403, predicting the damage effect of on-site RC components: inputting the data of the on-site explosion input data set into the end-to-end calculation model of the RC component explosion damage effect, and the output thereof is the prediction result of the required RC component explosion damage effect.

[0134] The present invention uses a database of explosion damage effects of RC components to train an end-to-end calculation model of explosion damage effects of RC components, and completes the prediction of the damage effects of different explosion positions and different explosion equivalents on RC components. First, a numerical solution method for the damage effects of RC components under explosion impact is established using a finite element analysis method; a database of explosion damage effects of RC components is established using a numerical solution method for the explosion damage effects of RC components, and multi-source data feature extraction and fusion are performed on the data at the same time; then, a data set (training data set, verification data set) of a graph data structure is constructed using the processed data in the database; then an end-to-end calculation model of explosion damage effects of RC components is built, and the structure and training parameters of the end-to-end calculation model of explosion damage effects of RC components are determined using the training data set and the verification data set; after the parameters obtained are loaded into the end-to-end calculation model of explosion damage effects of RC components, an end-to-end calculation method for real-time prediction of explosion damage effects of RC components with adaptive explosion equivalents and positions can be obtained, and the explosion damage effects of RC components with different explosion equivalents and explosion positions can be predicted.

[0135] The present invention adopts a graph neural network, which can use explosion overpressure data as input to directly predict the damage effect of RC components without relying on complex professional knowledge or cumbersome formula derivation. Compared with traditional finite element analysis, the end-to-end deep learning model of the present invention can quickly convert explosion overpressure data into damage effect prediction results, omitting the multi-step reasoning and calculation process. By establishing a multi-scenario explosion damage effect database, the graph neural network model trained by the present invention can quickly learn and generalize, adapt to different explosion equivalents, explosion locations, and changes in component geometric characteristics, and realize the prediction of the damage effect of RC components under multi-scenario and multi-dimensional explosion impacts.

[0136] The present invention also designs an end-to-end computing system for real-time prediction of explosion damage effects of RC components, comprising:

[0137] A solution unit is used to establish a numerical solution method for explosion damage effect of RC components and obtain numerical simulation data of explosion damage effect of RC components;

[0138] A data processing unit is used to associate the numerical simulation data of explosion damage effect of the RC component with the explosion condition and the geometric shape, size and material properties of the RC component to obtain an explosion damage effect database of the RC component including a training data set and a verification data set;

[0139] A model training unit, used to train and optimize the constructed neural network using the training data set and the verification data set to obtain an end-to-end calculation model of explosion damage effect of RC components;

[0140] The output unit is used to complete the real-time prediction of the explosion damage effect of the target RC component by using the end-to-end calculation model of the explosion damage effect of the RC component.

[0141] The technical solution of the above end-to-end computing system is similar to the above-mentioned bridge end-to-end computing method, which will not be repeated here.

[0142] Based on the same technical solution, the present invention also discloses a computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, which, when executed by a computing device, enable the computing device to perform the above-mentioned end-to-end computing method for real-time prediction of explosion damage effects of RC components.

[0143] Based on the same technical solution, the present invention also discloses a computing device, including one or more processors, one or more memories and one or more programs, wherein the one or more programs are stored in the one or more memories and are configured to be executed by the one or more processors, and the one or more programs include instructions for executing the end-to-end computing method for real-time prediction of the explosion damage effect of the above-mentioned RC components.

[0144] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0145] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0146] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0147] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0148] The above embodiments are only for illustrating the technical idea of ​​the present invention, and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the present invention.

Claims

1. An end-to-end calculation method for real-time prediction of explosion damage effects on RC components, characterized in that: The steps include: Establish a numerical solution method for the explosion damage effect of RC components and obtain numerical simulation data of the explosion damage effect of RC components; The numerical simulation data of explosion damage effect of RC components are associated with explosion conditions and geometric shapes, sizes and material properties of RC components to obtain an explosion damage effect database of RC components including a training data set and a verification data set; The constructed neural network is trained and optimized using the training data set and the validation data set to obtain an end-to-end calculation model of the explosion damage effect of RC components. Using the end-to-end calculation model of RC component explosion damage effect, the real-time prediction of the target RC component explosion damage effect is completed; The steps for obtaining numerical simulation data of explosion damage effects of RC components include: According to the principle of explosion mechanics, the energy release function of explosives with different equivalents and densities over time is established. Combined with the physical and chemical properties of explosives, the initial conditions of the explosion center are determined to obtain the physical model of the explosion source; the initial conditions of the explosion center include the initial pressure, temperature, energy density and initial expansion velocity of the explosion center; According to the type of explosion source and its surrounding environment information, a nonlinear equation based on fluid mechanics is introduced to describe the reflection, scattering, diffraction and attenuation effects of shock waves during propagation. Combined with the physical law of explosion shock wave propagation in the air, the mathematical equation of explosion shock wave propagation is obtained. After the explosion source physical model and the explosion shock wave propagation mathematical equation are combined, they are transformed into finite element equations using the Galerkin method, and the field variables in the calculation domain are interpolated through the three-dimensional first-order four-node element shape function. The finite element equations are iterated in a discrete manner to form a shock wave-component fluid-solid coupling calculation form, and the stress state of the component formed by the shock wave on the surface of the RC component is obtained, including the overpressure peak, impulse, and duration; According to the material characteristics and mechanical properties of RC components, the damage thresholds of different RC components under different stress states are determined; According to the damage threshold of RC components, the component damage mode is analyzed and a numerical solution method for the explosion damage effect of RC components is established; Create a three-dimensional structured grid of RC components, specify the material properties of each RC component in the grid, use the numerical solution method of RC component explosion damage effect to numerically simulate the damage effect of RC components under the action of explosion load, and generate numerical simulation data of RC component explosion damage effect; The real-time prediction steps of the explosion damage effect of the target RC component include: Collect the target RC component geometry, size, material properties and on-site explosion conditions to obtain on-site test data; The field test data is input into the end-to-end calculation model of the explosion damage effect of RC components, and the prediction results of the explosion damage effect of the target RC components are output.

2. The method according to claim 1, characterized in that The steps for obtaining the explosion damage effect database of RC components include: The numerical simulation data of explosion damage effect of RC components are divided according to the shock waves generated under different explosion sources and explosion equivalent conditions; Correlate the partitioning results with the corresponding explosion conditions and the geometry, dimensions, and material properties of the RC components; The explosion conditions in the correlation results and the geometric shape, size, and material properties of the RC components are trimmed, normalized, and feature extracted to obtain the explosion vector of multi-source information fusion; According to the explosion conditions and RC components involved in different scenarios, the explosion vector of multi-source information fusion in each scenario is obtained, and the numerical simulation data of explosion damage effect of all scenarios and corresponding RC components are stacked. The explosion vector of multi-source information fusion is used as input, and the numerical simulation data of explosion damage effect of RC components is used as the output set, which is entered into the library system; The set is divided to obtain a RC component explosion damage effect database including a training data set and a verification data set.

3. The method according to claim 1, characterized in that The neural network includes a graph autoencoder layer, a multi-stream graph neural network layer, and a multi-stream merge prediction network layer: In the image encoding layer, six layers of graph convolutional networks GCN are cascaded, and a ReLU activation function is connected after each layer of GCN to output the encoding result G0; In the multi-stream graph neural network layer, the multi-head attention mechanism in the Transformer encoder model is improved to form a multi-head local attention mechanism. In the multi-head local attention mechanism, the query vector is only matched with the key-value data pairs of adjacent nodes in the graph structure; the three heads of the multi-head local attention mechanism are strain attention mechanism, stress attention mechanism and material property attention mechanism, and the input is G0, and the output is the graph node abstract strain vector G n1 , graph node abstract stress vector G n2 、Graph node abstract material characteristic vector G n3 ; In the multi-stream merging prediction network layer, a three-layer fully connected neural network is used to integrate the G output of the multi-stream graph neural network layer. n1 , G n2 and G n3 , and then the explosion damage effect of RC components is obtained through two layers of cascaded GCN, and a ReLU activation function is connected after each layer of GCN.

4. An end-to-end computing system based on the method according to any one of claims 1 to 3, characterized in that: include: A solution unit is used to establish a numerical solution method for explosion damage effect of RC components and obtain numerical simulation data of explosion damage effect of RC components; A data processing unit is used to associate the numerical simulation data of explosion damage effect of RC components with the explosion conditions and the geometric shape, size and material properties of the RC components, and obtain an explosion damage effect database of RC components including a training data set and a verification data set; A model training unit is used to train and optimize the constructed neural network using the training data set and the verification data set to obtain an end-to-end calculation model of the explosion damage effect of RC components; The output unit is used to complete the real-time prediction of the explosion damage effect of the target RC component by using the end-to-end calculation model of the explosion damage effect of the RC component.

5. A computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, characterized in that When the instructions are executed by a computing device, the computing device is caused to perform the method as claimed in any one of claims 1 to 3.

6. An electronic device, characterized in that: The method comprises one or more processors, one or more memories and one or more programs, wherein the one or more programs are stored in the one or more memories and are configured to be executed by the one or more processors, and the one or more programs include instructions for executing the method as claimed in any one of claims 1 to 3.

Images