Mesoscopic Modeling and Multi-scale Simulation Method of Coal and Rock Materials Based on Active Deep Learning

Abstract

The present application relates to a method for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning. The method comprises: first, formulating a complete unlabeled data sample space, and establishing a labeled data sample set based on the complete unlabeled data sample space; constructing a multi-scale analysis offline proxy model based on a deep learning model, and training the multi-scale analysis offline proxy model based on the labeled data sample set to obtain a basic predictor; inputting unlabeled data samples into the basic predictor to obtain sample pseudo-labels, and determining the unlabeled data information metric based on the sample pseudo-labels using an improved active learning strategy; expanding the labeled data sample set based on the unlabeled data information metric, and performing iterative training to obtain a final offline proxy metamodel; finally, embedding the final offline proxy metamodel into a finite element system to obtain the strain-stress response of coal and rock under multi-scale simulation of loading. This significantly improves the speed of multi-scale simulation.

Description

Technical Field

[0001] The present application relates to the field of constitutive modeling and numerical calculation technology of coal and rock materials, and in particular to a method for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning. Background Art

[0002] As a typical underground engineering medium, the mechanical properties and damage and failure laws of coal rock are directly affected by the microscopic particle state and contact conditions in the coal rock mass. Accurately evaluating the mechanical response of coal rock mass is crucial to ensuring safe mining in coal mines.

[0003] Numerical simulation (finite element method) is a key tool for evaluating the mechanical response of materials under such complex conditions, with constitutive models directly determining the reliability and accuracy of simulations. However, establishing accurate phenomenological constitutive models of coal and rock is extremely difficult. Rigorous mathematical derivation and extensive theoretical knowledge limit the development of traditional models. Discrete element method, or DEM, can reproduce the mechanical properties of rock masses at the microscopic level without complex model derivation and necessary assumptions. This technique is a reliable numerical simulation technique, but its massive computational resources are often prohibitive. Multiscale simulation offers a way to balance this situation. It analyzes the mechanical behavior of complex materials or structures by coupling the microscale and macroscales. This method segments the network at the macroscale and calculates variables such as displacements and stresses at the nodes. At the integration point of each macrocell, a microscopic discrete element model (commonly referred to as a representative volume element) is established to capture the material's microstructural behavior. Stress and other response quantities obtained by decoupling the equations in the micromodel are then transferred back to the macromodel. However, in practical applications, the computational power required to calculate such large-scale microscopic discrete element models remains prohibitive, limiting the development and application of multiscale simulation.

[0004] Machine learning, particularly deep learning, as an offline proxy model provides a simpler and faster approach for multiscale numerical simulation. As a core component, the construction of offline deep learning proxy models inevitably requires a large amount of data, which is often difficult to obtain experimentally. Discrete simulation, triggered by microscopic contact, can more directly reproduce the macroscopic mechanical properties of materials, making it possible to collect large-scale mechanical data. However, the stress states of underground coal and rock masses are complex, and even using a single unit model, fully realizing these complex stress states remains extremely difficult.

[0005] Therefore, in the relevant technology, there is an urgent need for a method that can achieve low-cost, high-precision multi-scale simulation based on as little high-cost data as possible and improve simulation accuracy and efficiency. Summary of the Invention

[0006] Based on this, it is necessary to provide a coal and rock material microscopic modeling and multiscale simulation method based on active deep learning that can achieve low-cost, high-precision multiscale simulation based on as little high-cost data as possible and improve simulation accuracy and efficiency in response to the above technical problems.

[0007] In a first aspect, the present application provides a method for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning. The method comprises:

[0008] Formulate a complete unlabeled data sample space, and establish a labeled data sample set based on the complete unlabeled data sample space;

[0009] Constructing a multi-scale analysis offline proxy model based on the deep learning model, and training the multi-scale analysis offline proxy model based on the labeled data sample set to obtain a basic predictor;

[0010] Inputting unlabeled data samples into the basic predictor to obtain sample pseudo labels, and determining unlabeled data information metrics based on the sample pseudo labels using an improved active learning strategy, wherein the improved active learning strategy includes a distance-weighted diversity metric query strategy of the input space and a gradient-based data difference metric query strategy;

[0011] Expanding the labeled data sample set based on the unlabeled data information metric, and performing iterative training to obtain a final offline agent meta-model;

[0012] The final offline proxy element model is embedded into the finite element system to obtain the strain-stress response of coal and rock under multi-scale simulation of loading.

[0013] Optionally, in one embodiment of the present application, establishing a labeled data sample set based on the complete unlabeled data sample space includes:

[0014] Numerical calibration of discrete elements is performed to clarify the variables contained in the representative volume unit;

[0015] A complete unlabeled data sample space is proposed, typical loading paths are simulated using the representative volume unit, offline material data is acquired, and a labeled data sample set is established.

[0016] Optionally, in one embodiment of the present application, the training of the multi-scale analysis offline proxy model based on the labeled data sample set includes:

[0017] A gradient change method is used to determine the training balance judgment condition of the multi-scale analysis offline proxy model.

[0018] Optionally, in one embodiment of the present application, the iterative gradient norm calculation formula is:

[0019]

[0020] in, is the gradient norm, L is the loss function, ω i are the weights of the model.

[0021] Optionally, in one embodiment of the present application, determining the unlabeled data information metric based on the sample pseudo-label using an improved active learning strategy includes:

[0022] Calculate input data difference metrics to assess data differences;

[0023] The samples to be labeled are determined by evaluating the size or change of the gradient.

[0024] Optionally, in one embodiment of the present application, the diversity metric calculation formula of the input space is:

[0025]

[0026] Among them, x i , ε j is the strain data in the input space, x i is the data in the unlabeled sample space, is the unlabeled sample space, ε j is the labeled sample data, D i is the labeled sample space, Cosine Similarity(x i , ε j ) is the cosine similarity calculation.

[0027] Optionally, in one embodiment of the present application, the gradient-based data difference metric calculation formula is:

[0028]

[0029] in, is the gradient norm of the root mean square error of the basic predictor, is the maximum gradient norm.

[0030] In a second aspect, the present application also provides a device for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning. The device comprises:

[0031] A sample set establishment module is used to formulate a complete unlabeled data sample space and establish a labeled data sample set based on the complete unlabeled data sample space;

[0032] A multi-scale analysis offline proxy model construction and initial training module is used to construct a multi-scale analysis offline proxy model based on a deep learning model, train the multi-scale analysis offline proxy model based on the labeled data sample set, and obtain a basic predictor;

[0033] a data difference measurement module, configured to input unlabeled data samples into the basic predictor to obtain sample pseudo labels, and determine the unlabeled data information metric based on the sample pseudo labels using an improved active learning strategy, wherein the improved active learning strategy includes a distance-weighted diversity measurement query strategy of the input space and a gradient-based data difference measurement query strategy;

[0034] A multi-scale analysis offline proxy model iterative training module is used to expand the labeled data sample set based on the unlabeled data information metric and perform iterative training to obtain a final offline proxy meta-model;

[0035] The multi-scale analysis offline proxy model application module is used to embed the final offline proxy element model into the finite element to obtain the strain-stress response of coal and rock under multi-scale simulation of loading.

[0036] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program and the processor executes the steps of the method described in each of the above embodiments.

[0037] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in each of the above embodiments.

[0038] The above-mentioned coal and rock material microscopic modeling and multi-scale simulation method based on active deep learning first proposes a complete unlabeled data sample space, and establishes a labeled data sample set based on the complete unlabeled data sample space; then, a multi-scale analysis offline proxy model is constructed based on the deep learning model, and the multi-scale analysis offline proxy model is trained based on the labeled data sample set to obtain a basic predictor; then, the unlabeled data samples are input into the basic predictor to obtain sample pseudo labels, and an improved active learning strategy is used to determine the unlabeled data information metric based on the sample pseudo labels. The improved active learning strategy includes a distance-weighted diversity metric query strategy of the input space and a gradient-based data difference metric query strategy; then, the labeled data sample set is expanded based on the unlabeled data information metric, and iterative training is performed to obtain the final offline proxy metamodel; finally, the final offline proxy metamodel is embedded in the finite element to obtain the strain-stress response of coal and rock under multi-scale simulation of loading. In other words, by using data optimization methods to intelligently select the data with the most learning value in the data space, an offline proxy model of coal and rock materials is constructed with the minimum computing resource cost, which significantly reduces the human and material resources consumed in data collection or generation; secondly, compared with traditional multi-scale numerical simulation, the offline proxy model replaces the micro-scale analysis process in the multi-scale simulation, which significantly improves the multi-scale simulation speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A diagram illustrating an application environment of a coal and rock material microscopic modeling and multi-scale simulation method based on active deep learning in one embodiment;

[0040] Figure 2 1 is a flow chart of a method for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning in one embodiment;

[0041] Figure 3 A schematic diagram of the structure of a multi-scale analysis offline agent model in one embodiment;

[0042] Figure 4 A schematic diagram of iterative training of a multi-scale analysis offline agent model in one embodiment;

[0043] Figure 5 A schematic diagram of improving the comparison of the amount of labeled data for active learning in one embodiment;

[0044] Figure 6 A schematic diagram showing a comparison between the improved active deep learning model training loss and the general training loss in one embodiment;

[0045] Figure 7 A schematic diagram of an application of a multi-scale analysis offline agent model in one embodiment;

[0046] Figure 8 A schematic diagram showing a comparison between an embodiment and a conventional numerical simulation result;

[0047] Figure 9 is a schematic diagram of the relationship between stress and strain of a unit in one embodiment;

[0048] Figure 10 A structural block diagram of a device for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning in one embodiment;

[0049] Figure 11 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0051] The coal and rock material microscopic modeling and multi-scale simulation method based on active deep learning provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown, the terminal communicates with the server through the network. The data storage system can store data that the server needs to process. The data storage system can be integrated on the server or placed on the cloud or other network servers. The terminal can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can be smart speakers, smart TVs, smart air conditioners, smart car devices, etc. Portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server can be implemented as a standalone server or a server cluster consisting of multiple servers.

[0052] In one embodiment, Figure 2 As shown in the figure, a coal rock material microscopic modeling and multi-scale simulation method based on active deep learning is provided. Figure 1 The following steps are used as an example to illustrate the server in the example:

[0053] S201: Prepare a complete unlabeled data sample space, and establish a labeled data sample set based on the complete unlabeled data sample space.

[0054] In the embodiment of the present application, first, a complete unlabeled data sample space is prepared Based on the complete unlabeled data sample space, a small number of typical loading paths are selected to establish a labeled data sample set

[0055] Specifically, in one embodiment of the present application, establishing a labeled data sample set based on the complete unlabeled data sample space includes:

[0056] S301: Perform numerical calibration on discrete elements to clarify the variables contained in the representative volume unit.

[0057] S303: Prepare a complete unlabeled data sample space, use the representative volume unit to simulate a typical loading path, obtain material offline data, and establish a labeled data sample set.

[0058] In one embodiment of the present application, first, the discrete elements are numerically calibrated to clarify the number of particles, particle grading, particle density, loading amplitude and loading speed contained in the representative volume unit. In specific applications, for the sake of generality, loading is performed in plane stress to simplify mathematical expressions and computing resources. It should be noted that loading in three-dimensional space is still applicable, and the three strain and stress components under plane stress are convenient to be displayed in the form of graphs in space, so a plane stress model is adopted. According to relevant theoretical research, when the number of particle units is greater than 4000, the microscopic model can largely express the macroscopic properties of the material, so the number of unit particles is 6000, the average particle diameter is 1.2 mm, the particle density is 1.8 g / cm3, and the particle grading is uniformly distributed; the loading speed is set to 10 -2 m / s, a time step of 2e-6, and a maximum strain of 0.1. Optionally, triaxial loading can be used, which saves more computational resources.

[0059] Afterwards, in multi-scale simulations, the stress conditions of the microscopic units are unclear, and when using deep learning proxy models, as many paths as possible should be considered. Taking pure loading conditions as an example, a small number of typical loading paths are expressed in plane mechanics space as typical model loading conditions such as uniaxial compression and biaxial compression. Typical loading paths are simulated using representative volume units. Related simulations can be executed in batches using Python language to calculate and obtain offline material data. Extract the strain components ε in the two main directions during loading XX ,ε YY and the stress component σ XX ,σ YY , forming a labeled data sample set Without loss of generality, the path described is relatively simple. When multi-scale simulation is required under complex paths, the loading conditions should be expanded. It should be noted that the same applicability also applies to cyclic loading.

[0060] In plane strain stress space, the general expressions of strain and stress are:

[0061]

[0062] To further simplify the expression, the strain-stress components are projected into the principal strain-stress space through transformation. This transformation follows the transformation of the general strain-stress state into the principal strain-stress space, reduces the analysis dimension, and saves computing resources. The calculation method for converting the traditional stress components into the principal stress space is as follows:

[0063]

[0064] The inverse transformation formula is as follows:

[0065]

[0066] Similarly, strain components are converted in the same way.

[0067] S203: Construct a multi-scale analysis offline proxy model based on the deep learning model, and train the multi-scale analysis offline proxy model based on the labeled data sample set to obtain a basic predictor.

[0068] In the embodiment of the present application, a multi-scale analysis offline agent model NN is constructed based on a deep learning model. ini Specifically, deep learning models can be selected from the multi-layer perceptron model MLP, the recurrent neural network model RNN, and the Kolmogorov–Arnold network model KAN. Considering that data is independent of the path, the multi-layer perceptron model MLP is taken as an example. It is built using the Python deep learning package Pytorch and has undergone detailed architecture design and parameter optimization, as shown in the following example. Figure 3 As shown, the network is set to [2, 64, 64, 64, 64, 2], meaning the input layer contains 2 neurons, the output layer also contains 2 neurons, and each of the four hidden layers in the middle contains 64 neurons. This design fully considers the complexity and balance of the problem, ensuring that the model can effectively capture the nonlinear mapping relationship between input and output. The hidden layers use the ReLU (Rectified Linear Unit) activation function, which has good nonlinear expression capabilities and can avoid the gradient vanishing problem, thereby improving the model's convergence efficiency and performance. Subsequently, the labeled data sample set is input into the multi-scale analysis offline proxy model for model training to construct a basic predictor. During training, the mean squared error (MSE) is used as the loss function, defined as the average of the squared differences between the predicted value and the true value, to measure the model's fitting error. The optimizer uses Adam (Adaptive Moment Estimation), which has the advantages of fast convergence and automatically adjusts the learning rate to accommodate different gradient update requirements. The initial learning rate is set to 0.001. The loss function is calculated as follows:

[0069]

[0070] Where N is the number of samples, y i is the predicted value, is the marked value.

[0071] In addition, in one embodiment of the present application, the training of the multi-scale analysis offline agent model based on the labeled data sample set includes:

[0072] A gradient change method is used to determine the training balance judgment condition of the multi-scale analysis offline proxy model.

[0073] In one embodiment of the present application, the gradient change method is used as the criterion for judging model convergence. When the gradient norm is less than a preset threshold in 5 consecutive training rounds, the model is considered to have converged, the training is completed, and the current multi-scale analysis offline proxy model is saved. This method can effectively avoid overtraining, reduce unnecessary computational overhead, and ensure that the model's predictive ability reaches an ideal level. The preset gradient norm threshold is 10 -5 .

[0074] In one embodiment of the present application, the iterative gradient norm calculation formula is:

[0075]

[0076] in, is the gradient norm, L is the loss function, ω i are the weights of the model.

[0077] In one embodiment of the present application, the iterative gradient norm calculation formula quantifies the magnitude of the gradient change of the current model by calculating the square root of the sum of the squares of the gradients of the loss function with respect to all weight parameters. The gradient norm directly reflects the magnitude of the model parameter updates in the current training round. A smaller gradient norm indicates that the model parameters have stabilized and training is nearing convergence.

[0078] S205: Input the unlabeled data samples into the basic predictor to obtain sample pseudo labels, and determine the unlabeled data information metric based on the sample pseudo labels using an improved active learning strategy, wherein the improved active learning strategy includes a distance-weighted diversity metric query strategy of the input space and a gradient-based data difference metric query strategy.

[0079] In the embodiment of the present application, a basic predictor is used to predict unlabeled data samples. Perform inference and obtain sample output That is, sample pseudo label According to the unlabeled data sample space I i (ε1,ε2,…,ε s ) and the basic predictor An improved active learning strategy is used to analyze the information contained in unlabeled data samples. This strategy is an improved active deep learning query strategy that combines distance-weighted diversity metric queries in the input space with gradient-based data difference metric queries. This overcomes the problem of traditional gradient-based methods being prone to falling into local optima. Specifically, datasets with similar input sample features produce similar results when calculated based on gradients. This makes gradient-based query strategies prone to batch-selecting data from concentrated areas, leading to intelligent selection of optimal data with lost information. This also provides a foundation for batch-expanding datasets.

[0080] Specifically, in one embodiment of the present application, determining the unlabeled data information metric based on the sample pseudo-label using an improved active learning strategy includes:

[0081] S401: Calculate input data difference metrics to evaluate data differences.

[0082] S403: Determine the sample to be labeled by evaluating the size or change of the gradient.

[0083] In one embodiment of the present application, the distance-weighted diversity metric query strategy for the input space is distance-weighted diversity: directly calculating the input data difference metric to evaluate data difference. The core idea is to determine which samples are most valuable for model training by measuring the differences between different input samples. Gradient-based data difference metric query strategy: intelligently select more appropriate samples for labeling by evaluating the size or change of the gradient. The core idea is that when some data can significantly affect the model gradient update, these data samples often contain more information.

[0084] In one embodiment of the present application, the diversity metric calculation formula of the input space is:

[0085]

[0086] Among them, x i , ε j is the strain data in the input space, x i is the data in the unlabeled sample space, is the unlabeled sample space, ε j is the labeled sample data, D i is the labeled sample space, Cosine Similarity(x i , ε j ) is the cosine similarity calculation.

[0087] In one embodiment of the present application, the distance weighted diversity calculation method of the input space is cosine similarity. Without loss of generality, this method can be used to evaluate the similarity between high-dimensional vectors. XX , ε YY ) As an illustration, the cosine similarity is calculated as follows:

[0088]

[0089] Among them, x i 、x j For a vector in the data space, · represents the vector dot product, and ||·|| represents the modulus of the vector.

[0090] The calculation formula for the diversity measure of the input space is:

[0091]

[0092] Among them, x i , ε j is the strain data in the input space, x i is the data in the unlabeled sample space, is the unlabeled sample space, ε j is the labeled sample data, D i is the labeled sample space, Cosine Similarity(x i , ε j ) is the cosine similarity calculation.

[0093] In this embodiment, this strategy directly calculates the differences between input samples and identifies samples with large feature value changes by measuring the distribution differences of the samples in the feature space.

[0094] In one embodiment of the present application, the gradient-based data difference metric calculation formula is:

[0095]

[0096] in, is the gradient norm of the root mean square error of the basic predictor, is the maximum gradient norm.

[0097] In one embodiment of the present application, the gradient-based data difference metric query strategy calculation process is as follows: For the basic predictor NN ini , according to the loss function in the model training process - the root mean square error MSE is defined as L(x i ); for the input sample Calculate its gradient norm

[0098]

[0099] Where, Is the loss function with respect to sample x i The partial derivative of the j-th feature of .

[0100] Furthermore, calculate its maximum norm:

[0101]

[0102] Where N is the total number of unlabeled samples.

[0103] Furthermore, in order to normalize the diversity calculation of the input space to a unified dimension, a normalization calculation is performed, and the calculation formula of the gradient-based data difference measure is expressed as:

[0104]

[0105] in, is the gradient norm of the root mean square error of the basic predictor, is the maximum gradient norm.

[0106] The overall data information metric is calculated as follows:

[0107] Score=α×NGrad(x i )+(1-α)×InputDiff(x i , ε j )

[0108] Among them, α is a parameter that controls the weight of the two, and 0.7 can be selected in specific applications.

[0109] S207: Expand the labeled data sample set based on the unlabeled data information metric, and perform iterative training to obtain a final offline agent meta-model.

[0110] In the embodiment of the present application, based on the unlabeled data information metric, unlabeled data samples containing more information are extracted, discrete element RVE offline numerical calculation is performed, the corresponding sample labeling information is obtained, and the labeled data sample set is expanded to obtain a new data set. In specific applications, the first 5% of the most information is extracted as augmented data, and RVE units are used for simulation to extract strain and stress data and expand the existing data set. Without loss of generality, the amount of information sample extraction can be selected according to the specific data space, rather than being fixed. At the same time, iterative training is performed, that is, the multi-scale analysis offline proxy model is continued to be trained based on the expanded labeled data sample set, data information measurement is performed again, the labeled data sample set is expanded, and these steps are repeated in an iterative cycle until the amount of information in the unlabeled data set is less than the threshold or the number of iterations is greater than the preset value, and the final offline proxy metamodel is saved. The iterative loop execution process is as follows Figure 4 As shown in the flowchart, the existing data space is executed cyclically until the proxy model prediction gap is less than a threshold or the maximum number of iterations is reached. The preset proxy model prediction gap threshold is 0.05; the preset maximum number of iterations is 8, that is, the maximum learning path is 40% of all paths. In this embodiment, the final number of iterations is 5, and the expanded data is 25% of the data volume of the general data-driven method. Figure 5 As shown, a comparison is made between the different data requirements of the general deep learning agent model and the improved active deep learning proposed in this invention. Figure 5 The left side shows the data requirements of the general deep learning agent model as the control group, with a preset maximum load value of 0.15 strain, and the path loading end point is considered to be ε x ,ε y ,ε xy Select evenly distributed points in space.

[0111] like Figure 6 The following figure shows the change curve of the general deep learning training loss and the change trend of the optimal loss function under different iterations of the improved active deep learning proposed in this invention. It can be seen that only one-quarter of the data is used to achieve an accuracy extremely close to that of the general deep learning method. The data after the improved active deep learning iteration is input into the multi-scale analysis offline agent model NN ini , train using the same training method as the offline agent model and obtain the final offline agent meta-model

[0112] S209: Embed the final offline proxy element model into the finite element to obtain the strain-stress response of the coal rock under multi-scale simulation of loading.

[0113] In the embodiment of this application, Figure 7 The figure shows a schematic diagram of the application of the multi-scale analysis offline agent model. Using the finite element analysis secondary development interface, Abaqus is used to build and implement a user subroutine VUMAT for explicit analysis. This program uses C++ as the basic framework and embeds the designed final offline agent metamodel by calling Python programs. Based on previous training and data fitting, this model can effectively predict and analyze the strain-stress response of coal rock materials. By embedding this subroutine, seamless connection with the finite element calculation environment can be achieved, allowing users to directly call the model in the subsequent simulation process to perform real-time stress-strain analysis and result visualization. At the same time, a loading model of coal rock materials is established in the finite element software to simulate the stress-strain behavior of coal rock under different loading conditions in actual engineering. Specifically, a two-dimensional model is constructed, uniaxial loading is implemented, the maximum strain is less than 0.15, and the inp input file is generated. Finally, the input file and the developed user subroutine VUMAT are called through the Abaqus command to complete the numerical calculation. Figure 8 As shown in FIG, the misess stress of coal rock under uniaxial loading calculated by the present invention is compared with that calculated by traditional numerical simulation. Figure 9 As shown in the figure, the stress conditions of the same unit in the computational model are compared. It is easy to see that the errors in the macroscopic mechanical response of the coal rock under compression and the stress changes of the unit are very small, fully meeting engineering requirements. Through this mechanism, the microscopic and macroscopic properties of coal rock materials can be simultaneously considered at multiple scales, capturing their complex stress-strain response. Ultimately, the strain-stress relationship of coal rock under multi-scale loading can be obtained, further providing an efficient numerical tool for the research and application of coal rock mechanical behavior.

[0114] In the above-mentioned coal and rock material microscopic modeling and multi-scale simulation method based on active deep learning, first, a complete unlabeled data sample space is proposed, and a labeled data sample set is established based on the complete unlabeled data sample space; then, a multi-scale analysis offline proxy model is constructed based on the deep learning model, and the multi-scale analysis offline proxy model is trained based on the labeled data sample set to obtain a basic predictor; then, the unlabeled data samples are input into the basic predictor to obtain sample pseudo labels, and an improved active learning strategy is used to determine the unlabeled data information metric based on the sample pseudo labels. The improved active learning strategy includes a distance-weighted diversity metric query strategy of the input space and a gradient-based data difference metric query strategy; then, the labeled data sample set is expanded based on the unlabeled data information metric, and iterative training is performed to obtain the final offline proxy metamodel; finally, the final offline proxy metamodel is embedded in the finite element to obtain the strain-stress response of coal and rock under multi-scale simulation of loading. In other words, by using data optimization methods to intelligently select the data with the most learning value in the data space, an offline proxy model of coal and rock materials is constructed with the minimum computing resource cost, which significantly reduces the human and material resources consumed in data collection or generation; secondly, compared with traditional multi-scale numerical simulation, the offline proxy model replaces the micro-scale analysis process in the multi-scale simulation, which significantly improves the multi-scale simulation speed.

[0115] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0116] Based on the same inventive concept, embodiments of the present application also provide an active deep learning-based coal and rock material mesoscopic modeling and multi-scale simulation device for implementing the above-mentioned active deep learning-based coal and rock material mesoscopic modeling and multi-scale simulation method. The implementation solution provided by this device is similar to the implementation solution described in the above-mentioned method. Therefore, the specific limitations of one or more embodiments of the active deep learning-based coal and rock material mesoscopic modeling and multi-scale simulation device provided below can be found in the above-mentioned limitations of the active deep learning-based coal and rock material mesoscopic modeling and multi-scale simulation method, and will not be repeated here.

[0117] In one embodiment, Figure 10 As shown, a coal and rock material microscopic modeling and multi-scale simulation device 1000 based on active deep learning is provided, comprising: a sample set establishment module 1001, a multi-scale analysis offline proxy model construction and initial training module 1003, a data difference measurement module 1005, a multi-scale analysis offline proxy model iterative training module 1007 and a multi-scale analysis offline proxy model application module 1009, wherein:

[0118] The sample set establishing module 1001 is used to formulate a complete unlabeled data sample space and establish a labeled data sample set based on the complete unlabeled data sample space.

[0119] The multi-scale analysis offline proxy model construction and initial training module 1003 is used to construct a multi-scale analysis offline proxy model based on the deep learning model, train the multi-scale analysis offline proxy model based on the labeled data sample set, and obtain a basic predictor.

[0120] The data difference measurement module 1005 is used to input unlabeled data samples into the basic predictor to obtain sample pseudo labels, and determine the unlabeled data information measurement based on the sample pseudo labels using an improved active learning strategy, wherein the improved active learning strategy includes a distance-weighted diversity measurement query strategy of the input space and a gradient-based data difference measurement query strategy.

[0121] The multi-scale analysis offline proxy model iterative training module 1007 is used to expand the labeled data sample set based on the unlabeled data information metric and perform iterative training to obtain a final offline proxy meta-model.

[0122] The multi-scale analysis offline proxy model application module 1009 is used to embed the final offline proxy element model into the finite element to obtain the strain-stress response of coal and rock under multi-scale simulation of loading.

[0123] In one embodiment of the present application, the sample set establishment module is further configured to:

[0124] Numerical calibration of discrete elements is performed to clarify the variables contained in the representative volume unit;

[0125] A complete unlabeled data sample space is proposed, typical loading paths are simulated using the representative volume unit, offline material data is acquired, and a labeled data sample set is established.

[0126] In one embodiment of the present application, the multi-scale analysis offline agent model construction and initial training module is further used to:

[0127] A gradient change method is used to determine the training balance judgment condition of the multi-scale analysis offline proxy model.

[0128] In one embodiment of the present application, the iterative gradient norm calculation formula is:

[0129]

[0130] in, is the gradient norm, L is the loss function, ω i are the weights of the model.

[0131] In one embodiment of the present application, the data difference measurement module is further configured to:

[0132] Calculate input data difference metrics to assess data differences;

[0133] The samples to be labeled are determined by evaluating the size or change of the gradient.

[0134] In one embodiment of the present application, the diversity metric calculation formula of the input space is:

[0135]

[0136] Among them, x i , ε j is the strain data in the input space, x i is the data in the unlabeled sample space, is the unlabeled sample space, ε j is the labeled sample data, D i is the labeled sample space, CosineSimilarity(x i , ε j ) is the cosine similarity calculation.

[0137] In one embodiment of the present application, the gradient-based data difference metric calculation formula is:

[0138]

[0139] in, is the gradient norm of the root mean square error of the basic predictor, is the maximum gradient norm.

[0140] Each module in the above-mentioned active deep learning-based coal and rock material mesoscopic modeling and multi-scale simulation device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0141] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 11As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse.

[0142] Those skilled in the art will understand that Figure 11 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0143] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0144] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0145] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0146] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0147] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processor involved in the various embodiments provided herein may be, but are not limited to, a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, and the like.

[0148] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0149] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning, characterized in that: The method comprises: Formulate a complete unlabeled data sample space, and establish a labeled data sample set based on the complete unlabeled data sample space; Constructing a multi-scale analysis offline proxy model based on the deep learning model, and training the multi-scale analysis offline proxy model based on the labeled data sample set to obtain a basic predictor; Inputting unlabeled data samples into the basic predictor to obtain sample pseudo labels, and determining unlabeled data information metrics based on the sample pseudo labels using an improved active learning strategy, wherein the improved active learning strategy includes a distance-weighted diversity metric query strategy of the input space and a gradient-based data difference metric query strategy; Expanding the labeled data sample set based on the unlabeled data information metric, and performing iterative training to obtain a final offline agent meta-model; The final offline proxy element model is embedded into the finite element system to obtain the strain-stress response of coal and rock under multi-scale simulation of loading.

2. The method for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning according to claim 1 is characterized in that: The establishing of a labeled data sample set based on the complete unlabeled data sample space includes: Numerical calibration of discrete elements is performed to clarify the variables contained in the representative volume unit; A complete unlabeled data sample space is proposed, typical loading paths are simulated using the representative volume unit, offline material data is acquired, and a labeled data sample set is established.

3. The method for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning according to claim 1 is characterized in that: The training of the multi-scale analysis offline agent model based on the labeled data sample set includes: A gradient change method is used to determine the training balance judgment condition of the multi-scale analysis offline proxy model.

4. The method for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning according to claim 3 is characterized in that: The formula for calculating the iterative gradient norm is: in, is the gradient norm, L is the loss function, ω i are the weights of the model.

5. The method for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning according to claim 1 is characterized in that: Determining the unlabeled data information metric based on the sample pseudo labels using an improved active learning strategy includes: Calculate input data difference metrics to assess data differences; The samples to be labeled are determined by evaluating the size or change of the gradient.

6. The method for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning according to claim 1 is characterized in that: The calculation formula for the diversity metric of the input space is: Among them, x i , ε j is the strain data in the input space, x i is the data in the unlabeled sample space, is the unlabeled sample space, ε j is the labeled sample data, D i is the labeled sample space, Cosine Similarity(x i ,ε j ) is the cosine similarity calculation.

7. The method for microscopic modeling and multi-scale simulation of coal and rock materials based on active deep learning according to claim 1 is characterized in that: The gradient-based data difference metric calculation formula is: in, is the gradient norm of the root mean square error of the basic predictor, is the maximum gradient norm.

8. A coal and rock material microscopic modeling and multi-scale simulation device based on active deep learning, characterized in that: The device comprises: A sample set establishment module is used to formulate a complete unlabeled data sample space and establish a labeled data sample set based on the complete unlabeled data sample space; A multi-scale analysis offline proxy model construction and initial training module is used to construct a multi-scale analysis offline proxy model based on a deep learning model, train the multi-scale analysis offline proxy model based on the labeled data sample set, and obtain a basic predictor; a data difference measurement module, configured to input unlabeled data samples into the basic predictor to obtain sample pseudo labels, and determine the unlabeled data information metric based on the sample pseudo labels using an improved active learning strategy, wherein the improved active learning strategy includes a distance-weighted diversity measurement query strategy of the input space and a gradient-based data difference measurement query strategy; A multi-scale analysis offline proxy model iterative training module is used to expand the labeled data sample set based on the unlabeled data information metric and perform iterative training to obtain a final offline proxy meta-model; The multi-scale analysis offline proxy model application module is used to embed the final offline proxy element model into the finite element to obtain the strain-stress response of coal and rock under multi-scale simulation of loading.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images