A method and system for predicting freeze-thaw damage and durability of concrete based on deep learning and physical constraints

Abstract

This invention relates to the field of structural durability prediction technology in building and civil engineering, specifically to a method and system for predicting concrete freeze-thaw damage and durability based on deep learning and physical constraints. The method involves constructing a feature input vector for concrete samples, normalizing and performing feature consistency processing on this vector, establishing a deep learning prediction model based on the preprocessed vector, learning the nonlinear mapping relationship between freeze-thaw parameters and performance degradation indices, and introducing a lightweight physical constraint mechanism containing modulus degradation consistency constraints and freeze-thaw damage monotonic evolution constraints into the deep learning model to constrain the physical rationality of the prediction results. The method also associates freeze-thaw damage variables with concrete mechanical property decay parameters and constructs durability evaluation indices through a durability mapping function. This invention fully explores the complex nonlinear laws of freeze-thaw damage while ensuring that the prediction results conform to the physical evolution laws, achieving high-precision prediction of the freeze-thaw damage evolution process and durability indices of concrete.

Description

Technical Field

[0001] This invention relates to the field of structural durability prediction technology in building and civil engineering, specifically to a method and system for predicting freeze-thaw damage and durability of concrete based on deep learning and physical constraints. Background Technology

[0002] Concrete, as the most widely used core building material in the fields of construction engineering and transportation infrastructure, is extensively applied in critical engineering scenarios such as bridges, tunnels, roads, building main structures, and hydraulic structures. In cold and seasonally frozen-thaw regions, concrete structures are subjected to repeated freeze-thaw cycles for extended periods. The expansion and contraction of moisture during freezing and thawing triggers the continuous expansion of internal microcracks and irreversible deterioration of the pore structure, leading to a gradual degradation of mechanical properties such as compressive and flexural strength. This not only significantly reduces the service durability of concrete structures but also poses a serious threat to the safety of engineering structures. Therefore, accurately predicting the evolution of freeze-thaw damage in concrete is a crucial basis for optimizing the design of engineering structures, scientifically assessing service life, and making operation and maintenance decisions, possessing extremely high practical engineering value.

[0003] Currently, numerous studies have been conducted both domestically and internationally on methods for predicting freeze-thaw damage in concrete and rock, and related technologies have been initially applied in engineering fields. However, existing methods still have many technical shortcomings, making it difficult to meet the requirements of high-precision prediction and practical engineering applications. Some deep learning-based freeze-thaw damage prediction methods, such as the rock freeze-thaw damage prediction method proposed in CN119830720A, can automatically extract freeze-thaw damage patterns through deep learning models and improve the model's prediction accuracy and robustness by combining missing value processing and thermal unique encoding. However, these methods are designed specifically for the freeze-thaw characteristics of rock, have a large number of model parameters, are highly dependent on experimental data, and do not fully incorporate the physical evolution mechanism of freeze-thaw damage, resulting in poor physical consistency of the prediction results. Furthermore, they lack sufficient efficiency in comprehensively utilizing experimental data on multiple parameters such as mass loss rate and dynamic elastic modulus.

[0004] Some prediction methods based on traditional mechanics theory, such as the rock elastoplastic damage constitutive model disclosed in CN120628794A, construct yield functions and hardening laws through triaxial compression test data, which can reflect the compaction stage and volume deformation of freeze-thawed rocks and improve prediction accuracy to a certain extent. However, this method is based on empirical formulas and elastoplastic theory, lacks the ability to explore the complex nonlinear degradation law of freeze-thaw damage, is difficult to efficiently utilize multi-parameter, large-scale experimental data, and has a cumbersome calculation process, making it impossible to achieve rapid prediction in engineering scenarios.

[0005] There are also some dedicated freeze-thaw damage prediction models, such as the freeze-thaw damage prediction method for hydraulic concrete proposed in CN119047293B. This method combines equivalent damage age with a fractional-order model and the Grey Wolf optimization algorithm to achieve a coupled analysis of the effects of freeze-thaw temperature and concrete saturation on damage. However, the applicable scenarios of this model are relatively limited, and it does not sufficiently constrain the physical laws of experimental data, which can easily lead to deviations between the prediction results and the actual physical evolution process of freeze-thaw damage.

[0006] In summary, existing freeze-thaw damage prediction schemes generally suffer from problems such as insufficient characterization of freeze-thaw degradation mechanisms and poor physical consistency of prediction results. They also fail to fully utilize experimental data on multiple parameters, including freeze-thaw environmental parameters, ultrasonic wave velocity, and compressive strength. Furthermore, existing models have significant limitations in terms of nonlinear degradation law discovery, generalization ability, and robustness. Some methods rely excessively on empirical formulas or dedicated models, resulting in a narrow range of applications. They cannot simultaneously meet the needs of high-precision prediction of freeze-thaw damage and convenient application in engineering practice. Therefore, there is an urgent need to propose a new method for predicting freeze-thaw damage of concrete that is more in line with engineering practice. Summary of the Invention

[0007] The purpose of this invention is to propose a method and system for predicting freeze-thaw damage and durability of concrete based on deep learning and physical constraints. This method enables high-precision and physically consistent prediction of the evolution process of freeze-thaw damage and durability indicators of concrete, providing reliable technical support for the design optimization, service life assessment and engineering maintenance decisions of concrete structures.

[0008] According to a first aspect of the present disclosure, a method for predicting freeze-thaw damage and durability of concrete based on deep learning and physical constraints is provided, comprising the following steps: Concrete performance data under different freeze-thaw cycles were obtained through standard concrete freeze-thaw cycle tests. The performance data included at least one of the following: concrete mass loss rate, relative dynamic modulus of elasticity, compressive strength, ultrasonic wave velocity, and freeze-thaw environment parameters. Based on this performance data, a [further details are needed to complete the translation]. Concrete sample feature input vector corresponding to the second freeze-thaw cycle ; Input vector of concrete sample features Normalization and feature consistency processing were performed to construct a unified concrete freeze-thaw prediction dataset; Based on the preprocessed concrete sample feature input vector, a deep learning prediction model for concrete freeze-thaw damage and durability is established. This model learns the nonlinear mapping relationship between concrete freeze-thaw parameters and performance degradation indices. The model expression is as follows: ,in, For deep learning prediction models; These are model parameters; Predict the output vector for the model; and Includes predicted values ​​of concrete freeze-thaw damage under the Nth freeze-thaw cycle. and concrete durability prediction index ; A lightweight physical constraint model is introduced into the deep learning prediction model. The physical constraint model includes modulus degradation consistency constraint and freeze-thaw damage monotonic evolution constraint to impose physical rationality constraints on the prediction results of the deep learning model. By correlating freeze-thaw damage variables with concrete mechanical property degradation parameters, a concrete durability evaluation index is constructed using a durability mapping function, expressed as follows: , in: Indicates the performance index of concrete durability. This represents the set of mechanical performance parameters of concrete. For durability mapping function, This represents the variable indicating freeze-thaw damage to concrete.

[0009] In one embodiment, the process of constructing the modulus degradation consistency constraint is as follows: The degradation relationship of the relative dynamic elastic modulus of concrete is used as a physical constraint: The predicted dynamic elastic modulus is derived from the freeze-thaw damage values ​​predicted by the model: Introducing a modulus degradation residual term: in, This is the initial dynamic elastic modulus; The first output of the deep learning prediction model Predicted dynamic elastic modulus under multiple freeze-thaw cycles; The concrete specimen experienced the first Measured dynamic elastic modulus after one freeze-thaw cycle; The output of the deep learning prediction model Predicting freeze-thaw damage variables under multiple freeze-thaw cycles.

[0010] In one embodiment, a physical constraint loss term is constructed based on the modulus degradation residual term: in, This is the physical consistency loss term constructed based on the dynamic elastic modulus degradation relationship.

[0011] In one embodiment, the process of constructing the monotonic evolution constraint of freeze-thaw damage is as follows: Construct the monotonic evolution constraint relationship of freeze-thaw damage: When the model prediction results do not satisfy the aforementioned constraints, a damage monotonicity violation term is defined: in, The output of the deep learning prediction model Predicting freeze-thaw damage variables under multiple freeze-thaw cycles The first output of the deep learning prediction model +1 Predicting freeze-thaw damage variables under multiple freeze-thaw cycles.

[0012] In one embodiment, a physical constraint loss term is constructed based on the damage monotonicity violation term: in, This is a physical constraint loss term constructed based on the monotonic evolution characteristics of freeze-thaw damage with the number of freeze-thaw cycles.

[0013] In one embodiment, a construction is made from the prediction error term. Total physical constraint loss term The deep learning prediction model is trained using a joint objective function, wherein the joint objective function is: in, This represents the error between the predicted value and the experimentally measured value. Represents the physical constraint loss term; These are the weighting coefficients.

[0014] According to a second aspect of the present disclosure, a concrete freeze-thaw damage and durability prediction system based on deep learning and physical constraints is provided, comprising: The data acquisition and feature construction module obtains concrete performance data under different freeze-thaw cycles through standard concrete freeze-thaw cycle tests. The performance data includes at least one of the following: concrete mass loss rate, relative dynamic modulus of elasticity, compressive strength, ultrasonic wave velocity, and freeze-thaw environment parameters. Based on this performance data, a feature construction module is built. Concrete sample feature input vector corresponding to the second freeze-thaw cycle ; The data preprocessing module processes the concrete sample feature input vector. Normalization and feature consistency processing were performed to construct a unified concrete freeze-thaw prediction dataset; The deep learning prediction module, based on the preprocessed concrete sample feature input vector, establishes a deep learning prediction model for concrete freeze-thaw damage and durability. It learns the nonlinear mapping relationship between concrete freeze-thaw parameters and performance degradation indices. The model expression is as follows: ,in, For deep learning prediction models; These are model parameters; Predict the output vector for the model; and Includes predicted values ​​of concrete freeze-thaw damage under the Nth freeze-thaw cycle. and concrete durability prediction index ; The physical constraint coupling module introduces a lightweight physical constraint model into the deep learning prediction model. The physical constraint model includes modulus degradation consistency constraint and freeze-thaw damage monotonic evolution constraint, which imposes physical rationality constraints on the prediction results of the deep learning model. The durability evaluation module correlates freeze-thaw damage variables with concrete mechanical property degradation parameters, and constructs a concrete durability evaluation index through a durability mapping function, the expression of which is: ,in: Indicates the performance index of concrete durability. This represents the set of mechanical performance parameters of concrete. For durability mapping function, This represents the variable indicating freeze-thaw damage to concrete.

[0015] According to a third aspect of the present disclosure, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and running on the memory, wherein the processor executes the program to implement the aforementioned method for predicting concrete freeze-thaw damage and durability based on deep learning and physical constraints.

[0016] According to a fourth aspect of the present disclosure, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the aforementioned method for predicting concrete freeze-thaw damage and durability based on deep learning and physical constraints.

[0017] The advantages of the above technical solutions adopted in this invention compared with the prior art are as follows: 1. This invention deeply integrates multi-dimensional experimental data on concrete freeze-thaw cycles with a deep learning model. Relying on the powerful ability of deep learning models to mine complex nonlinear relationships, it fully extracts the intrinsic correlation between freeze-thaw parameters and performance degradation, breaking through the limitations of traditional methods in terms of insufficient data utilization. This enables high-precision prediction of the evolution process of concrete freeze-thaw damage and durability indicators, significantly improving the accuracy of prediction results.

[0018] 2. By introducing a lightweight physical constraint mechanism that includes modulus degradation consistency constraint and freeze-thaw damage monotonic evolution constraint, the prediction results of the deep learning model strictly follow the inherent physical evolution law of concrete freeze-thaw damage, effectively avoiding abnormal prediction results that violate engineering reality, ensuring that the prediction results have both accuracy and physical rationality, and improving the engineering credibility of the prediction conclusions.

[0019] 3. The physical constraint mechanism introduced in this invention is a lightweight architecture that does not require complex calculation processes or massive parameter support. While achieving effective physical constraints, it does not increase the overall computational complexity of the model, making it convenient for rapid prediction and practical application in engineering scenarios, and balancing the practicality and computational efficiency of the prediction method. Attached Figure Description

[0020] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments of this application and their descriptions are used to explain this application and do not constitute an undue limitation of this application.

[0021] Figure 1 This is a flowchart of a method for predicting freeze-thaw damage and durability of concrete based on deep learning and physical constraints according to the present invention. Figure 2 This is a schematic diagram illustrating the process of concrete freeze-thaw experiment data acquisition, feature construction, and dataset generation. Figure 3 This is a schematic diagram of the network structure of a deep learning prediction model for freeze-thaw damage and durability of concrete, including a feature input layer, a deep feature extraction layer, and a prediction output layer. Figure 4 This is a schematic diagram of the physical constraint model for concrete freeze-thaw and its coupling relationship with the deep learning prediction model, which is used to guide the prediction results to meet the basic physical evolution law of concrete freeze-thaw damage. Figure 5 A schematic diagram showing the comparison between the experimental results and model predictions of concrete freeze-thaw damage after 20 freeze-thaw cycles. Figure 6 A schematic diagram showing the comparison between the experimental results and model predictions of concrete freeze-thaw damage after 20 freeze-thaw cycles. Figure 7 This is a schematic diagram showing a comprehensive comparison between the freeze-thaw damage of concrete under different freeze-thaw cycles and the model prediction results. Detailed Implementation

[0022] The present disclosure will be further described below with reference to the accompanying drawings and embodiments.

[0023] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0024] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0025] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and systems according to various embodiments of this disclosure. It should be noted that each block in a flowchart or block diagram may represent a module, segment, or portion of code, which may include one or more executable instructions for implementing the logical functions specified in the various embodiments. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutively represented blocks may actually be executed substantially in parallel, or they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, may be implemented using a dedicated hardware-based system that performs the specified functions or operations, or using a combination of dedicated hardware and computer instructions.

[0026] Example 1: like Figure 1 As shown, this embodiment provides a method for predicting freeze-thaw damage and durability of concrete based on deep learning and physical constraints, including the following steps: S1. Concrete performance data under different freeze-thaw cycles are obtained through standard concrete freeze-thaw cycle tests. The performance data includes at least one of the following: concrete mass loss rate, relative dynamic modulus of elasticity, compressive strength, ultrasonic wave velocity, and freeze-thaw environment parameters. Based on the performance data, a [further details are needed to complete the process]. Concrete sample feature input vector corresponding to the second freeze-thaw cycle ; like Figure 2 As shown, experimental data of concrete specimens under different freeze-thaw cycles were collected through standard freeze-thaw cycle tests. The experimental data included at least one or more of the following parameters: mass loss rate, relative dynamic modulus of elasticity, compressive strength, ultrasonic wave velocity, and freeze-thaw environment parameters.

[0027] The collected data is processed and encoded to construct a feature input vector that can characterize the evolution of concrete freeze-thaw damage, providing a unified data input format for subsequent deep learning prediction models.

[0028] Specifically, the first The concrete sample corresponding to the second freeze-thaw cycle is represented as follows: in, This is the input vector for the freeze-thaw characteristics of concrete. Indicates the first One freeze-thaw characteristic parameter.

[0029] S2. Input vector of concrete sample features Normalization and feature consistency processing were performed to construct a unified concrete freeze-thaw prediction dataset; Specifically, the concrete freeze-thaw experiment data are normalized, outlier corrections are performed, and missing values ​​are marked to eliminate dimensional differences between different physical quantities, thereby improving the stability of model training and prediction accuracy.

[0030] S3. Based on the preprocessed concrete sample feature input vector, a deep learning prediction model for concrete freeze-thaw damage and durability is established. This model learns the nonlinear mapping relationship between concrete freeze-thaw parameters and performance degradation indices. The model expression is as follows: ,in, For deep learning prediction models; These are model parameters; Predict the output vector for the model; and predict the output vector Includes predicted values ​​of concrete freeze-thaw damage under the Nth freeze-thaw cycle. and concrete durability prediction index ; like Figure 3 As shown, the deep learning prediction model includes an input layer, a feature extraction layer, a nonlinear mapping layer, and an output layer: the input layer is used to receive concrete freeze-thaw feature input vectors; the feature extraction layer uses a multi-layer neural network structure to express the freeze-thaw features in a high dimension; the nonlinear mapping layer is used to learn the complex nonlinear relationship between concrete freeze-thaw parameters and performance degradation indicators; the output layer simultaneously outputs the concrete freeze-thaw damage prediction results and durability prediction results.

[0031] This deep learning prediction model enables the joint prediction of the freeze-thaw damage state and durability of concrete.

[0032] S4. A lightweight physical constraint model is introduced into the deep learning prediction model. The physical constraint model includes modulus degradation consistency constraint and freeze-thaw damage monotonic evolution constraint to impose physical rationality constraints on the prediction results of the deep learning model. like Figure 4 As shown, the physical constraint model is used to limit the trend of the prediction results with the number of freeze-thaw cycles, so that the prediction results conform to the objective law of the gradual deterioration of concrete under repeated freeze-thaw cycles, thereby avoiding prediction results that do not conform to the actual engineering situation.

[0033] S5. Correlate freeze-thaw damage variables with concrete mechanical property degradation parameters, and construct a concrete durability evaluation index through a durability mapping function, the expression of which is: , in: Indicates the performance index of concrete durability. This represents the set of mechanical performance parameters of concrete. For durability mapping function, This represents the variable indicating freeze-thaw damage to concrete.

[0034] like Figure 5-7 As shown, the predicted concrete freeze-thaw damage index and durability results output by the deep learning prediction model are compared and analyzed with the actual experimental test results under the corresponding number of freeze-thaw cycles. The comparison results show that the prediction results of the method of the present invention have a high consistency with the actual experimental results at different freeze-thaw cycle stages, and can accurately reflect the development trend of concrete freeze-thaw damage and the decay law of durability performance, thus verifying the effectiveness and reliability of the method of the present invention in predicting concrete freeze-thaw damage.

[0035] Preferably, it further includes: training and optimizing the deep learning prediction model using a joint loss function that includes a prediction error term and a physical constraint term.

[0036] Once the model is trained, inputting the concrete freeze-thaw test parameters into the model will output the predicted results of concrete freeze-thaw damage and durability under the corresponding working conditions, providing a basis for the durability assessment of concrete structures and engineering decisions.

[0037] Example 2: This embodiment provides a concrete freeze-thaw damage and durability prediction system based on deep learning and physical constraints, including: The data acquisition and feature construction module obtains concrete performance data under different freeze-thaw cycles through standard concrete freeze-thaw cycle tests. The performance data includes at least one of the following: concrete mass loss rate, relative dynamic modulus of elasticity, compressive strength, ultrasonic wave velocity, and freeze-thaw environment parameters. Based on this performance data, a feature construction module is built. Concrete sample feature input vector corresponding to the second freeze-thaw cycle ; The data preprocessing module processes the concrete sample feature input vector. Normalization and feature consistency processing were performed to construct a unified concrete freeze-thaw prediction dataset; The deep learning prediction module, based on the preprocessed concrete sample feature input vector, establishes a deep learning prediction model for concrete freeze-thaw damage and durability. It learns the nonlinear mapping relationship between concrete freeze-thaw parameters and performance degradation indices. The model expression is as follows: ,in, For deep learning prediction models; These are model parameters; Predict the output vector for the model; and Includes predicted values ​​of concrete freeze-thaw damage under the Nth freeze-thaw cycle. and concrete durability prediction index ; The physical constraint coupling module introduces a lightweight physical constraint model into the deep learning prediction model. The physical constraint model includes modulus degradation consistency constraint and freeze-thaw damage monotonic evolution constraint, which imposes physical rationality constraints on the prediction results of the deep learning model. The durability evaluation module correlates freeze-thaw damage variables with concrete mechanical property degradation parameters, and constructs a concrete durability evaluation index through a durability mapping function, the expression of which is: ,in: Indicates the performance index of concrete durability. This represents the set of mechanical performance parameters of concrete. For durability mapping function, This represents the variable indicating freeze-thaw damage to concrete.

[0038] The above modules can be deployed on the same device or distributed devices; the division of modules is only a functional logic description and does not limit the specific physical boundaries or implementation order.

[0039] Example 3: An electronic device is provided for running the aforementioned "A method for predicting concrete freeze-thaw damage and durability based on deep learning and physical constraints". The electronic device includes: a processor, a memory, and optional communication interfaces / display devices / input devices, etc.; the memory stores a computer program that can run on the processor, and when the processor executes the program, it implements steps S1 to S5 of the method described in Embodiment 1, specifically including but not limited to: S1. Concrete performance data under different freeze-thaw cycles are obtained through standard concrete freeze-thaw cycle tests. The performance data includes at least one of the following: concrete mass loss rate, relative dynamic modulus of elasticity, compressive strength, ultrasonic wave velocity, and freeze-thaw environment parameters. Based on the performance data, a [further details are needed to complete the process]. Concrete sample feature input vector corresponding to the second freeze-thaw cycle ; S2. Input vector of concrete sample features Normalization and feature consistency processing were performed to construct a unified concrete freeze-thaw prediction dataset; S3. Based on the preprocessed concrete sample feature input vector, a deep learning prediction model for concrete freeze-thaw damage and durability is established. This model learns the nonlinear mapping relationship between concrete freeze-thaw parameters and performance degradation indices. The model expression is as follows: ,in, For deep learning prediction models; These are model parameters; Predict the output vector for the model; and Includes predicted values ​​of concrete freeze-thaw damage under the Nth freeze-thaw cycle. and concrete durability prediction index ; S4. A lightweight physical constraint model is introduced into the deep learning prediction model. The physical constraint model includes modulus degradation consistency constraint and freeze-thaw damage monotonic evolution constraint to impose physical rationality constraints on the prediction results of the deep learning model. S5. Correlate freeze-thaw damage variables with concrete mechanical property degradation parameters, and construct a concrete durability evaluation index through a durability mapping function, the expression of which is: , in: Indicates the performance index of concrete durability. This represents the set of mechanical performance parameters of concrete. For durability mapping function, This represents the variable indicating freeze-thaw damage to concrete.

[0040] The electronic device hardware can be one of a server, personal computer, workstation, industrial controller, edge computing device, or mobile terminal; the processor can be a general-purpose CPU, GPU, NPU, FPGA, or a combination thereof; the memory can be RAM, ROM, flash memory, or disk array. The device can interact with local / remote data storage (acquiring observation data and outputting inversion results) through a communication interface. The above hardware configuration does not constitute a limitation of the present invention.

[0041] Example 4: A computer-readable storage medium storing a computer program, which, when run on a processor of an electronic device, causes the program to execute the method steps S1 to S5 described in Embodiment 1; the storage medium may be a disk, optical disk, flash memory, solid-state drive, read-only memory, random access memory, or any combination of the above media.

[0042] Those skilled in the art will understand that the modules or steps described above can be implemented using general-purpose computer devices. Optionally, they can be implemented using computer-executable program code, which can then be stored in a storage device for execution by a computer device. Alternatively, they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. This disclosure is not limited to any particular combination of hardware and software.

[0043] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0044] While the specific embodiments of this disclosure have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of this disclosure. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of this disclosure are still within the scope of protection of this disclosure.

Claims

1. A method for predicting freeze-thaw damage and durability of concrete based on deep learning and physical constraints, characterized in that, Includes the following steps: Concrete performance data under different freeze-thaw cycles were obtained through standard concrete freeze-thaw cycle tests. The performance data included at least one of the following: concrete mass loss rate, relative dynamic modulus of elasticity, compressive strength, ultrasonic wave velocity, and freeze-thaw environment parameters. Based on this performance data, a [further details are needed to complete the translation]. Concrete sample feature input vector corresponding to the second freeze-thaw cycle ; Input vector of concrete sample features Normalization and feature consistency processing were performed to construct a unified concrete freeze-thaw prediction dataset; Based on the preprocessed concrete sample feature input vector, a deep learning prediction model for concrete freeze-thaw damage and durability is established. This model learns the nonlinear mapping relationship between concrete freeze-thaw parameters and performance degradation indices. The model expression is as follows: ,in, For deep learning prediction models; These are model parameters; Predict the output vector for the model; and Including the N Predicted value of freeze-thaw damage to concrete under one freeze-thaw cycle and concrete durability prediction index ; A lightweight physical constraint model is introduced into the deep learning prediction model. The physical constraint model includes modulus degradation consistency constraint and freeze-thaw damage monotonic evolution constraint to impose physical rationality constraints on the prediction results of the deep learning model. By correlating freeze-thaw damage variables with concrete mechanical property degradation parameters, a concrete durability evaluation index is constructed using a durability mapping function, expressed as follows: , in: Indicates the performance index of concrete durability. This represents the set of mechanical performance parameters of concrete. For durability mapping function, This represents the variable indicating freeze-thaw damage to concrete.

2. The method for predicting concrete freeze-thaw damage and durability based on deep learning and physical constraints according to claim 1, characterized in that, The process of constructing the modulus degradation consistency constraint is as follows: The degradation relationship of the relative dynamic elastic modulus of concrete is used as a physical constraint: The predicted dynamic elastic modulus is derived from the freeze-thaw damage values ​​predicted by the model: Introducing a modulus degradation residual term: in, This is the initial dynamic elastic modulus; The first output of the deep learning prediction model Predicted dynamic elastic modulus under multiple freeze-thaw cycles; The concrete specimen experienced the first Measured dynamic elastic modulus after one freeze-thaw cycle; The output of the deep learning prediction model Predicting freeze-thaw damage variables under multiple freeze-thaw cycles.

3. The method for predicting concrete freeze-thaw damage and durability based on deep learning and physical constraints according to claim 2, characterized in that, Construct a physical constraint loss term based on the aforementioned modulus degradation residual term: in, This is the physical consistency loss term constructed based on the dynamic elastic modulus degradation relationship.

4. The method for predicting concrete freeze-thaw damage and durability based on deep learning and physical constraints according to claim 1, characterized in that, The process of constructing the monotonic evolution constraint of freeze-thaw damage is as follows: Construct the monotonic evolution constraint relationship of freeze-thaw damage: When the model prediction results do not satisfy the aforementioned constraints, a damage monotonicity violation term is defined: in, The output of the deep learning prediction model Predicting freeze-thaw damage variables under multiple freeze-thaw cycles The first output of the deep learning prediction model +1 Predicting freeze-thaw damage variables under multiple freeze-thaw cycles.

5. The method for predicting concrete freeze-thaw damage and durability based on deep learning and physical constraints according to claim 4, characterized in that, Construct a physical constraint loss term based on the aforementioned damage monotonicity violation term: in, This is a physical constraint loss term constructed based on the monotonic evolution characteristics of freeze-thaw damage with the number of freeze-thaw cycles.

6. The method for predicting concrete freeze-thaw damage and durability based on deep learning and physical constraints according to claim 3 or 5, characterized in that, Constructing from the prediction error term Total physical constraint loss term The deep learning prediction model is trained using a joint objective function, wherein the joint objective function is: in, This represents the error between the predicted value and the experimentally measured value. Represents the physical constraint loss term; These are the weighting coefficients.

7. A concrete freeze-thaw damage and durability prediction system based on deep learning and physical constraints, characterized in that, include: The data acquisition and feature construction module obtains concrete performance data under different freeze-thaw cycles through standard concrete freeze-thaw cycle tests. The performance data includes at least one of the following: concrete mass loss rate, relative dynamic modulus of elasticity, compressive strength, ultrasonic wave velocity, and freeze-thaw environment parameters. Based on this performance data, a feature construction module is built. Concrete sample feature input vector corresponding to the second freeze-thaw cycle ; The data preprocessing module processes the concrete sample feature input vector. Normalization and feature consistency processing were performed to construct a unified concrete freeze-thaw prediction dataset; The deep learning prediction module, based on the preprocessed concrete sample feature input vector, establishes a deep learning prediction model for concrete freeze-thaw damage and durability. It learns the nonlinear mapping relationship between concrete freeze-thaw parameters and performance degradation indices. The model expression is as follows: ,in, For deep learning prediction models; These are model parameters; Predict the output vector for the model; and Includes predicted values ​​of concrete freeze-thaw damage under the Nth freeze-thaw cycle. and concrete durability prediction index ; The physical constraint coupling module introduces a lightweight physical constraint model into the deep learning prediction model. The physical constraint model includes modulus degradation consistency constraint and freeze-thaw damage monotonic evolution constraint, which imposes physical rationality constraints on the prediction results of the deep learning model. The durability evaluation module correlates freeze-thaw damage variables with concrete mechanical property degradation parameters, and constructs a concrete durability evaluation index through a durability mapping function, the expression of which is: ,in: Indicates the performance index of concrete durability. This represents the set of mechanical performance parameters of concrete. For durability mapping function, This represents the variable indicating freeze-thaw damage to concrete.

8. An electronic device, characterized in that, The invention includes a memory, a processor, and a computer program stored in the memory and running thereon. When the processor executes the program, it implements the method for predicting freeze-thaw damage and durability of concrete based on deep learning and physical constraints as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the method for predicting concrete freeze-thaw damage and durability based on deep learning and physical constraints as described in any one of claims 1-6.

Images