Domain decomposition type multi-scale intelligent simulation method for geothermal development scene

Abstract

The invention provides a terrestrial heat development scene-oriented domain decomposition type multi-scale intelligent simulation method, and belongs to the field of intelligent simulation, and the method comprises the steps: obtaining terrestrial heat resource data according to a terrestrial heat development scene, constructing a multi-scale simulation model, and obtaining the deep feature representation of the multi-scale simulation model through a deep neural network; carrying out feature fusion on the deep feature representation to obtain a common subspace; constructing an unsupervised clustering neural network model, performing expansion processing on the public subspace, and constructing an inter-scale bridging model by using a transfer learning algorithm; and utilizing the inter-scale bridging model to obtain a domain decomposition type multi-scale simulation model, carrying out simulation experiment verification, and building a simulation platform oriented to a geothermal development scene. According to the method, the problems that deep features of simulation models with different scales cannot be fully analyzed and utilized by an existing simulation method, and cross-scale transition of the simulation models is difficult due to the fact that simulation areas with different scales coexist and are mutually overlapped in a domain decomposition task are solved.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent simulation, and in particular relates to a domain decomposition multi-scale intelligent simulation method for geothermal development scenarios. Background Art

[0002] Geothermal resources are widely found in the earth's crust and are a clean, stable and renewable energy source. Geothermal development mainly includes exploration, drilling, resource extraction, utilization and management of geothermal resources. Geothermal modeling refers to the mathematical simulation of the formation, distribution, flow, extraction and utilization of geothermal resources to optimize development strategies and predict development effects. Traditional modeling methods use numerical methods such as finite element method and finite difference to solve key values ​​such as temperature and pressure distribution in geothermal fields. Physical knowledge-driven deep learning is an effective tool for computer simulation, but most existing methods are limited to single-scale scenarios characterized by simple models. In the process of deep geothermal exploitation, high-pressure fluids are injected into the ground through deep boreholes, forcing the micro-nano pores in the target rock formation to expand to several millimeters. If the giant fracture network formed in the end is considered, the research scope will be expanded to several kilometers. However, observations and experimental results show that the influence of microscopic damage accumulation in rocks on fracture behavior is often not accurately described by macroscopic continuous models. Therefore, how to develop a universal "scale bridging" method for efficient communication of multi-scale information is the key to achieving multi-scale simulation through deep learning.

[0003] With the rapid development of artificial intelligence technology, the use of deep learning methods to achieve intelligent simulation of physical problems is attracting widespread attention from academia and industry. Currently, among the many attempts to apply deep learning methods to multi-scale simulation, there are mainly studies on combining neural networks with traditional numerical methods, multi-scale model coupling methods based on data-driven or dynamic importance sampling, and the application of transfer learning to simulation.

[0004] However, existing simulation methods simply rely on the knowledge-data integrated driven characteristics of deep learning methods, and do not fully analyze and utilize the deep characteristics of simulation models of different scales. At the same time, the field of deep learning simulation has not yet found a method to solve the problem of difficulty in cross-scale transition of simulation models caused by the simultaneous existence and overlap of simulation areas of different scales in multi-scale simulation tasks in domain decomposition, nor has it effectively applied multi-scale models relatively independently to simulation research. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a domain decomposition multi-scale intelligent simulation method for geothermal development scenarios, which solves the problem that the existing simulation methods are unable to fully analyze and utilize the deep characteristics of simulation models of different scales themselves, and in the domain decomposition multi-scale simulation tasks, simulation areas of different scales will exist at the same time and overlap with each other, thereby causing difficulties in cross-scale transition of simulation models.

[0006] In order to achieve the above objectives, the technical solution adopted by the present invention is: a domain decomposition multi-scale intelligent simulation method for geothermal development scenarios, comprising the following steps: S1. According to the geothermal development scenario, geothermal resource data is obtained, and a multi-scale simulation model is constructed. The deep neural network is used to perform deep feature learning on the multi-scale simulation model to obtain the deep feature representation of the multi-scale simulation model; S2. Based on the deep feature representation of the multi-scale simulation model, feature fusion is performed on the cross-scale features of the multi-scale simulation model to obtain a common subspace; S3, build an unsupervised clustering neural network model, expand the common subspace, and use the transfer learning algorithm to build an inter-scale bridging model; S4, using the inter-scale bridging model to perform domain decomposition transition on the multi-scale simulation model to obtain a domain decomposition multi-scale simulation model; S5. Based on the physical process, the domain decomposition multi-scale simulation model is verified by simulation experiments, and the verified domain decomposition multi-scale simulation model is used to build a simulation platform for geothermal development scenarios to realize domain decomposition multi-scale intelligent simulation.

[0007] The beneficial effects of the present invention are as follows: the present invention utilizes physical information neural operators to fully analyze and utilize the deep features of simulation models of different scales themselves, thereby improving the pertinence and feasibility of the simulation of the present invention; by integrating deep feature representations, a multi-scale framework that takes into account both deep feature learning of simulation models and cross-scale feature fusion is realized, enriching the existing multi-scale simulation theory; and by constructing an inter-scale bridging model to perform domain decomposition on the multi-scale simulation model, further mining inter-scale features, achieving relative independence of macro-scale models and micro-scale models, and partially overlapping domain decomposition in the simulation domain, thereby improving the accuracy of the results of multi-scale intelligent simulation and the simulation efficiency.

[0008] Furthermore, the S1 comprises the following steps: S101. Obtain geothermal resource data according to geothermal development scenarios; S102, using prior knowledge of the same physical process in multiple scale models and geothermal resource data, pre-training the physical information neural operator, constructing an initial multi-scale simulation model, and embedding features into the initial multi-scale simulation model; S103, designing a targeted representation learning model based on the multi-scale simulation model with embedded features, and constructing a deep neural network based on learning and representation of deep features of the multi-scale simulation model according to the targeted representation learning model; S104. Using a deep neural network to learn the multi-scale simulation model with embedded features, the feature space of the multi-scale simulation model is obtained, and a deep feature representation of the multi-scale simulation model related to the geothermal reservoir is obtained.

[0009] Furthermore, the expression for embedding features of the multi-scale simulation model is as follows: ; in, Represented by a neural network The function represented by y The predicted value of Represents the input function The feature embedding of Indicates input coordinates y The feature embedding of q represents the number of basis functions selected in the target space, Representing a neural network Parameters of ; among them, , , express q Euclidean real number space.

[0010] The beneficial effects of the above further scheme are as follows: the present invention eliminates the semantic gap by designing a targeted representation learning model, and constructs a deep neural network to learn and represent the deep features of the pre-trained physical information neural operator simulation model, thereby improving the correlation between the deep features and the geothermal reservoir. The established simulation model feature space lays a solid foundation for the model's stratification and domain decomposition coupling.

[0011] Furthermore, the S2 comprises the following steps: S201, mapping the feature space of the multi-scale simulation model to the same common subspace by using feature fusion according to the deep feature representation of the multi-scale simulation model; S202. Using the topological manifold structure of the model data, the consistent structure and complementary information between feature representations of simulation models of different scales in the common subspace are balanced to obtain a common subspace of balanced feature representations.

[0012] The beneficial effects of the above further scheme are as follows: the present invention realizes the mapping from the feature space of models of different scales to the common subspace by using the physical information operator method for developing multi-view feature fusion, and introduces the idea of ​​topological manifold, thereby balancing the consistency and complementarity between different view representations, improving the robustness of the present invention, and bridging the heterogeneous gap.

[0013] Furthermore, the S3 comprises the following steps: S301. Based on the K-means algorithm, an unsupervised clustering neural network model is constructed, and the unsupervised clustering neural network model is used to mine and understand the manifold information of the inter-scale bridging model; S302, using an unsupervised clustering neural network model to expand the common subspace of the balanced feature representation; S303. Design a direct transfer learning algorithm to transfer knowledge from the multi-scale simulation model to the inter-scale bridging model on the expanded common subspace, and construct an inter-scale bridging model based on the manifold information mined by the unsupervised clustering neural network model.

[0014] Furthermore, the expression of the unsupervised clustering neural network model is as follows: ; in, represents the unsupervised clustering neural network model, Indicates i Clusters, Indicates i The centroid of the cluster, x represents the sample point, Representation sample x With the centroid Geodesic distance on manifolds.

[0015] Furthermore, the expression for performing domain decomposition transition on the multi-scale simulation model using the inter-scale bridging model is as follows: ; ; in, represents the transition path, represents the transition factor, represents a low-dimensional representation of the macro-scale model in a common subspace, represents the low-dimensional representation of the microscale model in a common subspace, represents the bridging matrix, Represents bridge weight information, represents the cluster center at the macro scale, Represents the cluster centers at the micro scale.

[0016] The beneficial effects of the above further scheme are as follows: the present invention mines the manifold information of the inter-scale bridging model by constructing an unsupervised clustering neural network model, and realizes the use of manifold information for the expansion of the feature common subspace; on the basis of further expanding the model feature common subspace, by using the expanded space, the migration of knowledge from the known scale simulation model to the inter-scale simulation model is realized, the domain shift problem is solved, and the multi-scale simulation of complex physical systems in the domain decomposition form of the present invention is realized; the multi-scale intelligent simulation model realizes domain decomposition multi-scale simulation through the smooth transition of the inter-scale bridging model, which solves the problem of difficulty in cross-scale transition of the simulation model caused by the simultaneous existence and overlapping of simulation areas of different scales. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 The figure is a flow chart of the method of the present invention.

[0018] Figure 2 This is the overall structural diagram of the domain decomposition multi-scale intelligent simulation method for geothermal development scenarios in this embodiment. DETAILED DESCRIPTION

[0019] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0020] Before describing this embodiment, the following terms are explained: PINO: Physical Information Neural Operator; K-means algorithm: K-means algorithm.

[0021] Example like Figure 1 As shown, the present invention provides a domain decomposition multi-scale intelligent simulation method for geothermal development scenarios, and its implementation method is as follows: S1. According to the geothermal development scenario, geothermal resource data is obtained and a multi-scale simulation model is constructed. The deep neural network is used to perform deep feature learning on the multi-scale simulation model to obtain the deep feature representation of the multi-scale simulation model. The specific steps are as follows: S101. Obtain geothermal resource data according to geothermal development scenarios; S102, using prior knowledge of the same physical process in multiple scale models and geothermal resource data, pre-training the physical information neural operator, constructing an initial multi-scale simulation model, and embedding features into the initial multi-scale simulation model; S103, designing a targeted representation learning model based on the multi-scale simulation model with embedded features, and constructing a deep neural network based on learning and representation of deep features of the multi-scale simulation model according to the targeted representation learning model; S104. Using a deep neural network to learn the multi-scale simulation model with embedded features, the feature space of the multi-scale simulation model is obtained, and a deep feature representation of the multi-scale simulation model related to the geothermal reservoir is obtained.

[0022] In this embodiment, the deep characteristics of the multi-scale simulation model are related to the long-term evolution of the geothermal reservoir, the cross-scale correlation behavior, and the coupling relationship between different physical fields; like Figure 2 As shown, the physical models in the form of differential equations at different scales in geothermal development scenarios can be divided into: macro-scale models (continuous models based on the conservation laws of mass, momentum and energy), meso-scale models (particle dynamics models that simulate the mechanical interactions between rock units) and micro-scale models (based on). In order to achieve the learning and representation of deep features of multi-scale simulation models, the prior knowledge of physical models in the form of differential equations of the same physical process at different scales is used to pre-train the physical information neural operator and establish a simulation model of the corresponding scale; according to the process of learning the parameterized partial differential equation solving operator based on the physical information neural operator model, in the process of constructing the physical information neural operator, feature embedding based on Fourier features (RFF) is used to improve the performance of the method of the present invention. The calculation formula is as follows: ; in, Represented by a neural network The function represented by y The predicted value of Represents the input function The feature embedding of Indicates input coordinates y The feature embedding of q represents the number of basis functions selected in the target space, Representing a neural network Parameters of ; among them, , , express q dimensional Euclidean real space; based on automatic differentiation, construct differential operators and convert the loss function based on partial differential equations , loss function based on initial conditions And the loss function based on boundary conditions Minimize and get the optimal parameters , complete the physical information neural operator optimization.

[0023] In this embodiment, the multi-scale simulation model of the present invention is constructed based on a physical information neural operator network; a targeted representation learning model for the PINO simulation model is designed, and a deep neural network for learning and representing deep features of the model is constructed on the basis of eliminating the semantic gap, thereby learning the feature space of the PINO simulation model at multiple scales, and obtaining the deep feature representation of the multi-scale simulation model related to the geothermal reservoir.

[0024] S2. Based on the deep feature representation of the multi-scale simulation model, the cross-scale features of the multi-scale simulation model are fused to obtain a common subspace. The specific steps are as follows: S201, mapping the feature space of the multi-scale simulation model to the same common subspace by using feature fusion according to the deep feature representation of the multi-scale simulation model; S202. Using the topological manifold structure of the model data, the consistent structure and complementary information between feature representations of simulation models of different scales in the common subspace are balanced to obtain a common subspace of balanced feature representations.

[0025] In this embodiment, according to the tensor-based canonical correlation analysis method (TCCA), the study constructs a mapping from the feature space of the multi-scale model to the same common subspace, and realizes the preliminary fusion of the multi-scale simulation model features; the overall method spans multiple scales of micro, meso and macro, extracts the key features of each scale, and then integrates the extracted key features through a fusion mechanism to construct a simulation model with global consistency; Specifically, Figure 2 As shown, based on the deep feature representation of the multi-scale simulation model, the feature representation of the learned multi-scale simulation model is used ( p =1,2,…,n; n represents the total number of views, which is the total number of models of different scales in this embodiment), calculate the covariance tensor of all views , which is used to discover the relevant information between all views. By simplifying the calculation, it is transformed into solving the equivalent problem of maximizing the typical correlation of multiple views, that is, finding a set of optimal rank tensors ( k =1,2,…,r; r represents the feature dimension after dimensionality reduction, which should be less than the minimum value of the feature dimensions of different views), so that the covariance tensor can be represented as a rank tensor The weighted sum of ; the calculation expression is as follows: ; in, represents the tensor product, Represents; and ranks a tensor Integrate into a transformation matrix , the transformation matrix Acting on feature representation , initially realize the mapping from original high-dimensional features to low-dimensional common subspace; Furthermore, considering that simulation models of different scales may have large differences in feature representation due to the different physical models they use, constructing a common subspace based solely on tensor-based canonical correlation maximization cannot completely bridge the heterogeneous gap; after combining the research work on multi-view clustering, the idea of ​​topological manifold is introduced, and the topological manifold structure of the model data is used to balance the consistent structure and complementary information between the feature representations of simulation models of different scales, thereby improving the robustness of the PINO method for multi-view feature fusion and bridging the heterogeneous gap between feature spaces of different views.

[0026] S3. Build an unsupervised clustering neural network model, expand the common subspace, and use the transfer learning algorithm to build an inter-scale bridging model. The specific steps are as follows: S301. Based on the K-means algorithm, an unsupervised clustering neural network model is constructed, and the unsupervised clustering neural network model is used to mine and understand the manifold information of the inter-scale bridging model; S302, using an unsupervised clustering neural network model to expand the common subspace of the balanced feature representation; S303. Design a direct transfer learning algorithm to transfer knowledge from the multi-scale simulation model to the inter-scale bridging model on the expanded common subspace, and construct an inter-scale bridging model based on the manifold information mined by the unsupervised clustering neural network model.

[0027] In this embodiment, in constructing the inter-scale bridging simulation model, a direct push transfer learning algorithm for the common subspace of model features is studied based on the two parts of expansion and utilization; for expansion, this embodiment uses a targeted unsupervised clustering algorithm based on the K-means algorithm to mine the manifold information of the inter-scale bridging model, and applies these to the expansion of the common subspace of model features. The expression of the unsupervised clustering neural network model is as follows: ; in, represents the unsupervised clustering neural network model, Indicates i Clusters, Indicates i The centroid of the cluster, x represents the sample point, Representation sample With the centroid Geodesic distances on manifolds; After initially realizing the construction of the unsupervised clustering neural network model, improvements are made from the perspective of utilizing manifold information, and a direct transfer learning algorithm is designed to utilize the expanded model feature common subspace to realize the effective transfer of knowledge from the known scale simulation model to the inter-scale bridging model, and obtain the inter-scale bridging model to realize the multi-scale simulation of complex physical systems in the form of domain decomposition using the PINO method. The expression of the inter-scale bridging model is as follows: ; in, represents the bridging matrix, represents the cluster center at the macro scale, Represents the cluster centers at the micro scale.

[0028] S4. Use the inter-scale bridging model to perform domain decomposition transition on the multi-scale simulation model to obtain a domain decomposition multi-scale simulation model.

[0029] In this embodiment, the multi-scale intelligent simulation model based on the neural operator network realizes domain decomposition multi-scale simulation through the smooth transition of the inter-scale bridging model, and obtains a domain decomposition multi-scale simulation model, which is expressed as follows: ; in, and They represent the low-dimensional representation of the macro-scale model and the micro-scale model in the common subspace, Represents the transition path, which is used to define arrive The smooth change of represents the transition factor, and ,when hour, ,when hour, , represents the bridging matrix, Indicates bridge weight information.

[0030] S5. Based on the physical process, the domain decomposition multi-scale simulation model is verified by simulation experiments, and the verified domain decomposition multi-scale simulation model is used to build a simulation platform for geothermal development scenarios to realize domain decomposition multi-scale intelligent simulation.

[0031] In this embodiment, a domain decomposition multi-scale simulation model is used to conduct simulation experiments based on typical physical processes. Deep learning of the domain decomposition multi-scale simulation model is used as a multi-scale simulation tool. The multi-scale simulation tool with verified performance is used to build a simulation platform for geothermal development scenarios to achieve domain decomposition multi-scale intelligent simulation.

[0032] In this embodiment, the deep feature representation of the multi-scale simulation model and the scale-to-scale bridging model are integrated to develop a PINO method for domain decomposition multi-scale simulation.

[0033] 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.

[0034] 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.

[0035] 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 A step that specifies a function in one or more boxes.

Claims

1. A domain decomposition multi-scale intelligent simulation method for geothermal development scenarios, characterized in that: The following steps are involved: S1. According to the geothermal development scenario, geothermal resource data is obtained, and a multi-scale simulation model is constructed. The deep neural network is used to perform deep feature learning on the multi-scale simulation model to obtain the deep feature representation of the multi-scale simulation model; S2. Based on the deep feature representation of the multi-scale simulation model, feature fusion is performed on the cross-scale features of the multi-scale simulation model to obtain a common subspace; S3, build an unsupervised clustering neural network model, expand the common subspace, and use the transfer learning algorithm to build an inter-scale bridging model; S4, using the inter-scale bridging model to perform domain decomposition transition on the multi-scale simulation model to obtain a domain decomposition multi-scale simulation model; S5. Based on the physical process, the domain decomposition multi-scale simulation model is verified by simulation experiments, and the verified domain decomposition multi-scale simulation model is used to build a simulation platform for geothermal development scenarios to realize domain decomposition multi-scale intelligent simulation.

2. The domain decomposition multi-scale intelligent simulation method for geothermal development scenarios according to claim 1 is characterized in that: The S1 comprises the following steps: S101. Obtain geothermal resource data according to geothermal development scenarios; S102, using prior knowledge of the same physical process in multiple scale models and geothermal resource data, pre-training the physical information neural operator, constructing an initial multi-scale simulation model, and embedding features into the initial multi-scale simulation model; S103, designing a targeted representation learning model based on the multi-scale simulation model with embedded features, and constructing a deep neural network based on learning and representation of deep features of the multi-scale simulation model according to the targeted representation learning model; S104. Using a deep neural network to learn the multi-scale simulation model with embedded features, the feature space of the multi-scale simulation model is obtained, and a deep feature representation of the multi-scale simulation model related to the geothermal reservoir is obtained.

3. The domain decomposition multi-scale intelligent simulation method for geothermal development scenarios according to claim 2 is characterized in that: The expression for feature embedding of the multi-scale simulation model is as follows: in, Represented by a neural network The function represented by y The predicted value of Represents the input function The feature embedding of Indicates input coordinates y The feature embedding of q represents the number of basis functions selected in the target space, Representing a neural network Parameters of ; among them, , , express q Euclidean real number space.

4. The domain decomposition multi-scale intelligent simulation method for geothermal development scenarios according to claim 1 is characterized in that: The S2 comprises the following steps: S201, mapping the feature space of the multi-scale simulation model to the same common subspace by using feature fusion according to the deep feature representation of the multi-scale simulation model; S202. Using the topological manifold structure of the model data, the consistent structure and complementary information between feature representations of simulation models of different scales in the common subspace are balanced to obtain a common subspace of balanced feature representations.

5. The domain decomposition multi-scale intelligent simulation method for geothermal development scenarios according to claim 1 is characterized in that: The S3 comprises the following steps: S301. Based on the K-means algorithm, an unsupervised clustering neural network model is constructed, and the unsupervised clustering neural network model is used to mine and understand the manifold information of the inter-scale bridging model; S302, using an unsupervised clustering neural network model to expand the common subspace of the balanced feature representation; S303. Design a direct transfer learning algorithm to transfer knowledge from the multi-scale simulation model to the inter-scale bridging model on the expanded common subspace, and construct an inter-scale bridging model based on the manifold information mined by the unsupervised clustering neural network model.

6. The domain decomposition multi-scale intelligent simulation method for geothermal development scenarios according to claim 5 is characterized in that: The expression of the unsupervised clustering neural network model is as follows: in, represents the unsupervised clustering neural network model, Indicates i Clusters, Indicates i The centroid of the cluster, x represents the sample point, Representation sample x With the centroid Geodesic distance on manifolds.

7. The domain decomposition multi-scale intelligent simulation method for geothermal development scenarios according to claim 6 is characterized in that: The expression for performing domain decomposition transition on the multi-scale simulation model using the inter-scale bridging model is as follows: in, represents the transition path, represents the transition factor, represents a low-dimensional representation of the macro-scale model in a common subspace, represents the low-dimensional representation of the microscale model in a common subspace, represents the bridging matrix, Represents bridge weight information, represents the cluster center at the macro scale, Represents the cluster centers at the micro scale.

Images