A method and system for determining the mix proportion of liquid metal cement composite materials

Abstract

This invention discloses a method and system for determining the mix proportion of liquid metal-cement composite materials, applied in the field of composite material manufacturing technology. The method includes obtaining a set of key parameters by acquiring multiple key parameters and their ranges corresponding to the liquid metal component and the cement matrix; performing multi-objective performance simulation to obtain multiple candidate key parameter combinations; calculating the stress concentration factor of each candidate key parameter combination; determining the candidate key parameter combination with the minimum stress concentration factor as the initial composite material mix proportion scheme; when the minimum stress concentration factor is lower than a preset safety threshold, determining the initial composite material mix proportion scheme as the final composite material mix proportion scheme; when the minimum stress concentration factor is not lower than the preset safety threshold, adjusting the candidate key parameter combinations until the obtained stress concentration factor is lower than the preset safety threshold. This invention effectively improves the accuracy of composite material mix proportions by integrating performance prediction and micromechanical analysis.

Description

Technical Field

[0001] This invention relates to the field of composite material manufacturing technology, and in particular to a method and system for determining the mix proportion of liquid metal cement composite materials. Background Technology

[0002] Liquid metal-cement composites are a novel type of composite material that combines the properties of liquid metals with the advantages of cement as a matrix. Liquid metals typically refer to metals or alloys that remain liquid at room temperature, such as gallium-based liquid metals. These materials possess characteristics such as low viscosity, high thermal and electrical conductivity, and excellent wettability. Cement, as a common building material, has good mechanical properties and durability. Combining liquid metals with cement can create composite materials with unique physicochemical properties, suitable for a variety of engineering and construction applications.

[0003] However, liquid metals and cement differ significantly in their physical and chemical properties, and the macroscopic properties of their composites are the result of complex interactions between their components, microstructure, and interfacial characteristics. Currently, the determination of material formulations in this field still largely relies on the prediction of individual properties or macroscopic average properties, and the composite material proportions are determined based on these performance predictions. While the optimal proportions determined by existing prediction and optimization methods show excellent performance in macroscopic performance prediction, their microstructures exhibit significant mechanical weaknesses, severely impacting the material's durability and long-term service safety. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method and system for determining the mix proportion of liquid metal cement composite materials, thereby resolving the technical issue that existing methods for determining the mix proportion of composite materials cannot obtain accurate mix proportions.

[0005] A first aspect of this invention provides a method for determining the mix proportion of a liquid metal cement composite material, the method comprising:

[0006] Obtain multiple key parameters and their ranges for each of the liquid metal component and the cement matrix to obtain a set of key parameters.

[0007] A multi-scale prediction model is constructed, and multi-objective performance simulation is performed on the parameter combinations in the key parameter set based on the multi-scale prediction model to obtain multiple candidate key parameter combinations.

[0008] Calculate the stress concentration factor for each of the candidate key parameter combinations under standard load;

[0009] The candidate key parameter combination corresponding to the minimum stress concentration factor is determined as the initial composite material mix design; when the minimum stress concentration factor is lower than a preset safety threshold, the initial composite material mix design is determined as the final composite material mix design.

[0010] When the minimum stress concentration factor is not lower than the preset safety threshold, stress concentration areas are identified from the stress cloud map of the corresponding composite material model; based on the spatial distribution characteristics of the stress concentration areas and the geometric morphology of the material phase interface, the micro-stress identification results of the stress concentration areas are derived in reverse; the candidate key parameter combination is adjusted with the micro-stress identification results until the obtained stress concentration factor is lower than the preset safety threshold.

[0011] As one preferred embodiment, the acquisition of multiple key parameters and their ranges corresponding to the liquid metal component and the cement matrix, respectively, yields a set of key parameters, including:

[0012] The knowledge graph of liquid metal and cement is embedded and learned based on graph neural network to identify the coupling path between the liquid metal component and the cement matrix. The coupling path includes at least the stress transfer induced by thermal expansion mismatch.

[0013] Map the coupling path to a structured coupling relationship tensor;

[0014] Based at least on the structured coupling tensor, an artificial intelligence model is used to determine the first set of key parameters corresponding to the liquid metal component and the second set of key parameters corresponding to the cement matrix;

[0015] The first set of key parameters and the second set of key parameters are integrated to obtain the set of key parameters.

[0016] As one preferred embodiment, the construction of a multi-scale prediction model, based on which multi-objective performance simulations are performed on parameter combinations in the key parameter set, yields multiple candidate key parameter combinations, including:

[0017] A first performance prediction model with self-healing efficiency as the prediction target, a second performance prediction model with electrical conductivity as the prediction target, and a third performance prediction model with compressive strength as the prediction target are constructed respectively.

[0018] The first performance prediction model, the second performance prediction model, and the third performance prediction model are coupled through a parameter transfer chain to form a cross-scale performance prediction integrated model.

[0019] Based on the parameter range of each of the key parameters, the key parameters are combined and arranged to obtain multiple combination results;

[0020] The performance prediction of each of the combined results is predicted using the cross-scale performance prediction ensemble model to obtain the performance prediction result corresponding to each of the combined results.

[0021] Based on the selected criteria, multiple candidate key parameter combinations are obtained from the performance prediction results.

[0022] As one preferred embodiment, based on the parameter range of each of the key parameters, the key parameters are combined and arranged to obtain multiple combination results, including:

[0023] Determine the type of parameter range for each of the key parameters, and use the sampling method corresponding to the type to determine multiple sampled values ​​for each of the key parameters;

[0024] Multiple sampled values ​​of each of the key parameters are arranged and combined to obtain multiple combined results.

[0025] As one preferred embodiment, determining multiple sampled values ​​for each key parameter based on the type of parameter range of each key parameter using the sampling method corresponding to the type includes:

[0026] When the parameter range of the key parameter is continuous, the parameter range of the key parameter is divided into multiple equally probable intervals;

[0027] In each of the equal probability intervals, a sample value is determined to obtain multiple sample values ​​of the key parameter;

[0028] When the parameter range of the key parameter is discrete, multiple sample values ​​are randomly selected to obtain multiple sample values ​​of the key parameter.

[0029] As one preferred embodiment, the calculation of the stress concentration factor for each of the candidate key parameter combinations under standard load includes:

[0030] Based on the various candidate key parameter combinations, liquid metal particles of the liquid metal component are embedded in the cement matrix to construct multiple composite material geometric simulation models.

[0031] Uniaxial compression and shear simulations were performed on each of the composite material geometric simulation models to obtain the strain condensation tensor and effective elastic stiffness tensor of each composite material geometric simulation model under standard load.

[0032] Based on the strain concentration tensor and the effective elastic stiffness tensor, the stress concentration factor of each composite material geometric simulation model is obtained.

[0033] As one preferred embodiment, the process of obtaining the stress concentration factor for each of the composite material geometric simulation models based on the strain concentration tensor and the effective elastic stiffness tensor includes:

[0034] Based on the effective elastic stiffness tensor and macroscopic strain, the macroscopic average stress of each composite material geometric simulation model is obtained;

[0035] Based on the strain concentration tensor, the average stress within the inclusions of each of the composite material geometric simulation models is obtained;

[0036] Based on the average stress within the inclusion, the corresponding equivalent stress within the inclusion is obtained;

[0037] Based on the aforementioned macroscopic average stress, the corresponding macroscopic equivalent stress is obtained;

[0038] The ratio of the equivalent stress within the inclusions to the equivalent stress in the macroscopic region of each composite material geometric simulation model is calculated to obtain the stress concentration factor of each composite material geometric simulation model.

[0039] As one preferred embodiment, the step of reversely deriving the micro-stress identification result of the stress accumulation region based on the spatial distribution characteristics of the stress accumulation region and the geometry of the material phase interface includes:

[0040] If the spatial distribution characteristics of the stress accumulation zone are distributed along the interface between the liquid metal particles and cement, then the cause of accumulation is analyzed, and the interfacial bonding in the cause of accumulation is determined as the first micro-stress identification result.

[0041] If the spatial distribution characteristics of the stress accumulation zone are that it is located between multiple liquid metal particles, then the cause of accumulation is analyzed, and the particle spacing in the cause of accumulation is determined as the second micro-stress identification result.

[0042] If the spatial distribution characteristics of the stress accumulation zone are that it is far away from the liquid metal particles and located in the cement, then the cause of accumulation is analyzed, and the cement strength in the cause of accumulation is determined as the third micro-stress identification result.

[0043] The first micro-stress identification result, the second micro-stress identification result, or the third micro-stress identification result are integrated to deduce the micro-stress identification result of the stress accumulation area in reverse.

[0044] As one preferred embodiment, adjusting the candidate key parameter combination based on the micro-stress identification results includes:

[0045] If the micro-stress identification result includes the first micro-stress identification result, then the surface modification material data in the candidate key parameter combination shall be adjusted.

[0046] If the micro-stress identification result includes the second micro-stress identification result, then the doping data and liquid metal particle size in the candidate key parameter combination are adjusted;

[0047] If the micro-stress identification result includes the third micro-stress identification result, then the water-cement ratio and chemical composition in the candidate key parameter combination are adjusted.

[0048] Another embodiment of the present invention provides a system for determining the proportion of liquid metal cement composite materials, comprising:

[0049] The acquisition module is used to acquire multiple key parameters and their ranges corresponding to the liquid metal component and the cement matrix, and obtain a set of key parameters.

[0050] The prediction module is used to construct a multi-scale prediction model, and perform multi-objective performance simulation on the parameter combinations in the key parameter set based on the multi-scale prediction model to screen and obtain multiple candidate key parameter combinations.

[0051] The calculation module is used to calculate the stress concentration factor of each candidate key parameter combination under standard load;

[0052] An initial module is used to determine the candidate key parameter combination corresponding to the minimum stress concentration factor as an initial composite material mix design; when the minimum stress concentration factor is lower than a preset safety threshold, the initial composite material mix design is determined as the final composite material mix design.

[0053] The adjustment module is used to identify stress concentration areas from the stress cloud map of the corresponding composite material model when the minimum stress concentration factor is not lower than the preset safety threshold; based on the spatial distribution characteristics of the stress concentration areas and the geometric morphology of the material phase interface, reversely deduce the micro-stress identification results of the stress concentration areas; and adjust the candidate key parameter combination with the micro-stress identification results until the obtained stress concentration factor is lower than the preset safety threshold.

[0054] Compared with the prior art, the method and system for determining the mix proportion of liquid metal cement composite materials of the present invention have the following advantages:

[0055] This invention provides a method for determining the mix proportion of liquid metal cement composite materials. By integrating multi-scale modeling, multi-objective performance simulation and stress-driven closed-loop optimization mechanism, it significantly improves the scientificity, accuracy and engineering practicality of composite material mix design.

[0056] This invention first systematically acquires key parameters of the liquid metal component and cement matrix and constructs a structured parameter set, laying the data foundation for subsequent intelligent optimization. Then, it constructs a multi-scale prediction model covering micro-interfaces, meso-structures, and macro-responses, enabling collaborative simulation and efficient screening of multiple target properties such as strength, thermal conductivity, and electrical conductivity, avoiding performance imbalances caused by isolated evaluations. Next, this invention introduces a stress concentration factor as a core criterion and links it with a preset safety threshold, forming a closed-loop optimization path of "initial performance screening—stress verification—mechanism diagnosis—parameter feedback": when the stress concentration factor of the initial mix design does not meet safety requirements, it not only identifies high-stress areas but also, combined with the spatial distribution characteristics of the stress cloud map and the geometric morphology of the phase interface, reversely derives the micro-stress identification results that cause stress accumulation, thereby allowing for targeted adjustment of key parameters rather than blind iteration.

[0057] This invention closely and deeply links macroscopic mechanical response with microstructural evolution. The entire process is highly automated and scalable, and can be seamlessly integrated into a materials digital twin platform. It supports rapid response to customized needs in different engineering scenarios, thereby enabling a determined liquid metal cement composite material mix to significantly reduce local stress peaks while maintaining excellent functional properties (such as high thermal conductivity and electromagnetic shielding), improving structural durability and service safety. This provides reliable technical support for applications such as intelligent building materials, flexible electronic substrates, and extreme environment protection structures, and promotes the intelligentization process in the field of composite materials. Attached Figure Description

[0058] Figure 1 A schematic flowchart of an embodiment of the liquid metal cement composite material mix proportion determination method provided by the present invention;

[0059] Figure 2 A flowchart illustrating the parameter adjustment strategy of one embodiment of the liquid metal cement composite material mix proportion determination method provided by the present invention;

[0060] Figure 3 A system block diagram of one embodiment of the liquid metal cement composite material proportioning determination system provided by the present invention;

[0061] 301. Acquisition Module; 302. Prediction Module; 303. Calculation Module; 304. Initialization Module; 305. Adjustment Module. Detailed Implementation

[0062] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] Please refer to Figure 1 This is a flowchart illustrating one embodiment of the method for determining the mix proportion of liquid metal cement composite material provided by the present invention, including steps S101 to S105, each step as follows:

[0064] S101. Obtain multiple key parameters and their ranges for each of the liquid metal component (hereinafter referred to as liquid metal) and the cement matrix (hereinafter referred to as cement) to obtain a set of key parameters.

[0065] Preferably, in some embodiments, multiple key parameters and their ranges corresponding to the liquid metal component and the cement matrix are obtained to obtain a set of key parameters, including:

[0066] First, a multi-source heterogeneous knowledge base covering liquid metal material systems and cement-based composite materials is constructed. This knowledge base integrates structured and unstructured data from material property databases, academic literature, experimental reports, and engineering cases, covering information such as liquid metal types (e.g., gallium-based and bismuth-based alloys), surface treatment methods, dispersion processes, cement types (e.g., silicate cement and sulfoaluminate cement), admixtures, additives, and curing regimes. Based on this, a liquid metal-cement knowledge graph is constructed, where nodes represent material components, process parameters, or performance indicators, and edges represent semantic relationships such as physical correlation, chemical compatibility, mechanical coupling, or thermal interaction. In the embodiments of this invention, typical failure paths such as "thermal expansion mismatch → interfacial stress → microcrack initiation" are specifically marked.

[0067] Subsequently, a graph neural network (GNN) is used to embed the knowledge graph into the knowledge graph. The neighboring node information is aggregated through a message passing mechanism to generate a low-dimensional dense vector representation of each node. Based on this, potential high-order coupling paths between the liquid metal component and the cement matrix are identified, especially the cross-phase interaction mechanisms that have a decisive impact on the performance of composite materials. These include the interfacial stress transmission path caused by the difference in thermal expansion coefficients under temperature cycling, and the influence of liquid metal fluidity on the pore filling efficiency of cement. In this embodiment of the invention, the identified coupling paths are further formalized into structured coupling relationship tensors. These tensors encode the interdependence strength, direction of action, and physical dimension between different parameter categories in the form of multi-dimensional arrays, serving as prior knowledge constraints for subsequent parameter selection.

[0068] Next, the coupling tensor is input into a pre-trained artificial intelligence model (such as a multi-task learning network based on an attention mechanism). This model is fine-tuned on historical mixing experimental data and can dynamically evaluate the importance and sensitivity of each candidate parameter based on the coupling tensor, thereby outputting a first set of key parameters for the liquid metal component and a second set of key parameters for the cement matrix. While retaining their respective physical meanings, the two parameter sets have embedded cross-phase coupling constraints, avoiding combinatorial explosion or physical infeasibility problems caused by traditional independent parameter enumeration. Finally, the first and second key parameter sets are integrated to obtain the final key parameter set. It should be noted that the integration process may involve logical merging and redundancy handling, which will not be described in detail in this embodiment of the invention.

[0069] To further illustrate the above embodiments, the first set of key parameters includes dosage data, liquid metal particle size, and surface modification material data; the second set of key parameters includes porosity data, chemical composition, water-cement ratio, and curing temperature. In this embodiment, the key parameters of the liquid metal and their parameter ranges are obtained to form the first set of key parameters. Specifically, the dosage data is expressed as the volume fraction of liquid metal in the composite material, and its parameter range is set to 5% to 30%. For example, in a specific formulation, the volume fraction of liquid metal can be 15%. The liquid metal particle size is expressed as the numerical range of its diameter, set to 10 μm to 100 μm. For example, in practical applications, liquid metal particles with a median diameter of 50 μm can be selected. The surface modification material data is defined according to the type of surfactant or coupling agent used, and its parameter range includes three categories: "unmodified," "silane coupling agent modified," and "thiol compound modified." For example, in a preferred embodiment, KH-550 type silane coupling agent is used to surface treat the liquid metal particles. The key parameters of the cement and their parameter ranges are obtained to form the second set of key parameters. Porosity data is expressed as the total porosity of cement, with a parameter range of 8% to 25%. For example, the porosity can be controlled to around 15% by adjusting the preparation process. Chemical composition is expressed as the relative content ratio of C3S to C2S in the cement, with a parameter range of 2.0 to 4.0. For example, this ratio is approximately 2.8 when using ordinary Portland cement. The water-cement ratio is expressed as the ratio of the mass of water to the mass of cement, with a parameter range of 0.35 to 0.50. For example, a water-cement ratio of 0.42 is commonly used in standard mixes. Curing temperature is expressed as the ambient temperature during cement hydration, with a parameter range of 20°C to 80°C. For example, a curing temperature of 60°C can be used under accelerated curing conditions.

[0070] It should be noted that key parameters for liquid metals also include alloy composition, oxide layer thickness, and dynamic viscosity at a specific temperature (e.g., 25°C), as shown in Table 1. Alloy composition directly determines the electrical conductivity, viscosity, oxidation characteristics, and surface tension of the liquid metal, all of which affect the conductivity and self-healing flowability of the composite material. Oxide layer thickness significantly affects the interfacial bond strength and contact resistance between the liquid metal and cement. A moderate oxide layer may help improve interfacial compatibility, but an excessively thick layer will hinder conductivity. Viscosity determines the ability of the liquid metal to flow and spread within cracks and is a core parameter affecting self-healing efficiency.

[0071] Table 1 Key Parameter List

[0072]

[0073] Key parameters for cement also include cement aggregates, admixture types and dosages. Besides the key parameters of liquid metal and cement, there are also process-related parameters, such as mixing process parameters and molding and curing pressure. Mixing process parameters include mixing speed and mixing time, which can affect the uniformity of liquid metal particle dispersion in the cement paste and whether agglomeration occurs, crucial for forming a uniform conductive network and avoiding stress concentration. Molding and curing pressure refers to the pressure applied during material casting. Pressure can reduce macroscopic porosity and defects, increase material density, thereby improving mechanical properties and interfacial integrity.

[0074] S102. Construct a multi-scale prediction model, and perform multi-objective performance simulation on the parameter combinations in the key parameter set based on the multi-scale prediction model to obtain multiple candidate key parameter combinations.

[0075] In a specific embodiment of the present invention, the process of constructing a multi-scale prediction model first establishes dedicated performance prediction sub-models for the three core properties of liquid metal cement composite materials—self-healing efficiency, conductivity, and compressive strength. The first performance prediction model focuses on self-healing efficiency, based on the capillary flow and redox behavior of liquid metal at microcracks, combined with the pore connectivity and crack width distribution of the cement matrix, and uses a combination of phase-field simulation and data-driven regression to predict the functional recovery capability of the material after damage under different mix proportions. The second performance prediction model focuses on conductivity, using percolation theory and stochastic network models to incorporate the spatial distribution, contact probability, and interfacial resistance of liquid metal particles into the calculation, and estimates the overall conductivity through Monte Carlo simulation or graph neural networks. The third performance prediction model focuses on compressive strength, using micromechanical methods to consider the density of cement paste, the characteristics of the aggregate-slurry interface transition zone, and the weakening effect of the liquid metal phase as a soft inclusion on stress transmission, and constructs a strength prediction module based on finite element homogenization or machine learning surrogate models.

[0076] To achieve better model processing, this embodiment of the invention couples the three sub-models through a parameter transfer chain. Preferably, the porosity change after crack healing output by the self-healing model is used as the input correction term for the conductivity model. The reconstruction of the conductive network indirectly affects the local stress redistribution, which in turn feeds back to the damage variable in the strength model. This cross-model information flow is achieved by sharing intermediate state variables and a unified coordinate framework, forming a physically consistent, data-interoperable, cross-scale performance prediction integrated model. Next, based on the parameter ranges of the key parameters in each of the above embodiments, the key parameters are combined and arranged to obtain multiple combined results. Of course, Latin hypercube sampling or Bayesian optimization-guided combination strategies can be used to generate diverse parameter combination results, avoiding computational redundancy caused by full permutation. Specific embodiments will be provided below and will not be described here.

[0077] Finally, the combined results are input into the cross-scale performance prediction ensemble model to obtain the predicted values ​​of the three performance parameters simultaneously. Based on the preset multi-objective screening criteria (such as Pareto non-dominated ranking, weighted comprehensive score or user-specified performance threshold constraints), multiple candidate key parameter combinations that simultaneously meet functional and structural requirements are efficiently screened from a large number of simulation results.

[0078] Specifically, in the above embodiments, the first performance prediction model is constructed using a radial basis function neural network. This network contains a hidden layer, uses a Gaussian function as the activation function, and is trained using historical experimental data to obtain a nonlinear mapping relationship between input parameters and self-repair efficiency. Its core structure includes an input layer, a hidden layer, and an output layer. The input layer receives multi-dimensional parameters related to the self-repair process, such as material composition (e.g., liquid metal volume fraction, cement type), ambient temperature, ambient humidity, and repair agent properties. These parameters undergo data preprocessing to eliminate dimensional differences and improve model training stability. A clustering algorithm is used to determine the center vectors of the hidden layer neurons from the input data, ensuring that the center distribution covers the key areas of the input space. A Gaussian function is used as the activation function in the hidden layer to process the input data. Finally, in the output layer, the outputs of the hidden layer are linearly weighted and summed to obtain the predicted value of the self-repair efficiency. The weight parameters are optimized using least squares or gradient descent to minimize the error between the predicted value and the historical experimental data (true value).

[0079] The second performance prediction model is based on seepage theory and uses an improved Kirkpatrick model, the expression of which is:

[0080]

[0081]

[0082] in, This represents the volume fraction of liquid metal. The seepage threshold, This is the critical index. The resistivity of cement, The resistivity of the composite material, The resistivity of liquid metal.

[0083] Based on the above fitting model, the conductivity of composite materials under any liquid metal volume fraction can be predicted.

[0084] The third performance prediction model employs a support vector regression algorithm, using a radial basis function kernel, and optimizes the penalty parameter C and kernel function parameters through a grid search method. After the performance prediction model is built, based on the parameter ranges of each key parameter, the key parameters are combined and arranged to obtain multiple combined results. For example, the liquid metal volume fraction is taken into 6 levels at 5% intervals within the range of 5%-30%, the liquid metal particle size is taken into 5 levels within the range of 10-100μm, the water-cement ratio is taken into 4 levels within the range of 0.35-0.50, the porosity is taken into 4 levels within the range of 8%-25%, and the aggregate volume fraction is taken into 5 levels within the range of 40%-60%. Through permutation and combination, a total of 6×5×4×4×5=2400 parameter combinations are generated. The performance of these 2400 parameter combinations is then predicted using three pre-trained prediction models. Taking a certain combination as an example, its parameters are: liquid metal volume fraction 20%, particle size 45μm, water-cement ratio 0.42, porosity 15%, aggregate volume fraction 50%, self-healing efficiency prediction model output value 82.5%, electrical conductivity prediction model output value 3.2×10³ S / m, and compressive strength prediction model output value 48.6MPa.

[0085] Furthermore, in the above embodiments, based on the parameter range of each of the key parameters, the key parameters are combined and arranged to obtain multiple combination results, including:

[0086] Based on the type of parameter range of each key parameter, multiple sampled values ​​of each key parameter are determined using the sampling method corresponding to the type.

[0087] Multiple sampled values ​​of each key parameter are arranged and combined to obtain multiple combined results.

[0088] In this embodiment, based on the different data types and value ranges of each key parameter, a targeted sampling method is used to determine multiple sampled values ​​for each parameter: For continuous parameters, the Latin hypercube sampling method is used to ensure effective coverage of the parameter space. Taking the volume fraction of liquid metal as an example, within a continuous range of 5% to 30%, 20 representative values ​​are generated through Latin hypercube sampling, such as [7.2%, 10.5%, 13.8%, 16.4%, 19.1%, 22.6%, 25.9%, 28.3%], etc. These sampled values ​​are uniformly distributed within the parameter range and ensure orthogonality between dimensions. For discrete parameters, a full-factor sampling method is used to ensure complete coverage of all possible values. Taking the surface modification material type as an example, it includes three discrete categories: "unmodified," "silane coupling agent modified," and "thiol compound modified." During sampling, all three categories are directly taken as sampled values. In addition, for parameters with physical constraints, a constrained sampling method is used to ensure the rationality of the sampled values. Taking the combination of water-cement ratio and porosity as an example, when the water-cement ratio is high, such as greater than 0.45, the porosity should also be relatively high, such as greater than 20%. Constraints are introduced to ensure that the sampling combination conforms to the principles of materials science. After independent sampling of each parameter, the sampled values ​​of each parameter are permuted and combined. Specifically, the Cartesian product of the 8 sampled values ​​of liquid metal volume fraction, 6 sampled values ​​of particle size, 3 sampled values ​​of surface modification material, 5 sampled values ​​of water-cement ratio, and 5 sampled values ​​of porosity is performed, generating a total of 8×6×3×5×5=3600 complete parameter combinations. In this embodiment, each combination result is stored in structured data form. For example, the parameter values ​​corresponding to a certain combination result are: liquid metal volume fraction 19.1%, particle size 47μm, surface modification material is silane coupling agent, water-cement ratio 0.41, and porosity 18%. This systematic sampling and combination method not only ensures a full exploration of the parameter space but also controls the total number of combinations through intelligent sampling, providing high-quality input for subsequent performance prediction. The method described in this embodiment can comprehensively cover the design space of key parameters while ensuring computational efficiency, laying a solid foundation for discovering the optimal composite material mix ratio. Compared with traditional uniform or random sampling methods, the combined results generated by this method are more representative and scientific, significantly improving the efficiency and reliability of material design.

[0089] Furthermore, in the above embodiments, based on the type of parameter range of each key parameter, multiple sampled values ​​of each key parameter are determined using a sampling method corresponding to the type, including:

[0090] When the range of a key parameter is continuous, the range of the key parameter is divided into multiple equally probable intervals.

[0091] In each equally probable interval, a sample value is determined to obtain multiple sample values ​​of the key parameter;

[0092] When the range of the key parameter is discrete, multiple sample values ​​are randomly selected to obtain multiple sample values ​​of the key parameter.

[0093] In this embodiment, for key parameters with continuous ranges, a combination of equal probability interval division and sampling is used. Taking liquid metal volume fraction as an example, its parameter range is a continuous interval [5%, 30%], which is divided into 8 equal probability intervals. Assuming the parameter follows a uniform distribution within this range, the probability of each interval is 12.5%. The corresponding interval divisions are: [5%, 8.125%], [8.125%, 11.25%], [11.25%, 14.375%], [14.375%, 17.5%], [17.5%, 20.625%], [20.625%, 23.75%], [23.75%, 26.875%], [26.875%, 30%]. Then, a sample value is randomly determined within each equal probability interval. For example, 6.8% is randomly selected from the first interval [5%, 8.125%], 9.7% is randomly selected from the second interval [8.125%, 11.25%], and so on, ultimately yielding eight sampled values ​​for the liquid metal volume fraction: [6.8%, 9.7%, 12.9%, 16.2%, 19.3%, 22.1%, 25.6%, 28.4%]. Similarly, for the continuous parameter of water-cement ratio, with a range of [0.35, 0.50], after dividing it into five equally probable intervals, five sampled values ​​are obtained: [0.362, 0.391, 0.417, 0.443, 0.478].

[0094] For key parameters with discrete parameter ranges, a random selection method is used to determine the sampled values. Taking the surface modification material type as an example, it includes three discrete categories: "unmodified," "silane coupling agent modified," and "thiol compound modified." A random number generator is used to randomly select two sampled values ​​from these three categories. For example, if the first random selection yields "silane coupling agent modified" and the second random selection yields "unmodified," then the set of sampled values ​​for the surface modification material type is {"silane coupling agent modified", "unmodified"}. Similarly, for the discrete parameter of curing temperature, its selectable values ​​are [20℃, 40℃, 60℃, 80℃], and three sampled values ​​are obtained through random selection: [20℃, 60℃, 80℃]. After sampling all parameters, the sampled values ​​of each parameter are combined and verified to ensure that the sampled values ​​cover all regions of the parameter space and that there is no significant sampling bias. Finally, the sampled values ​​of each parameter are stored in matrix form for subsequent parameter combination arrangements. This embodiment distinguishes between continuous and discrete parameters and adopts corresponding sampling strategies, which not only ensures the full exploration of the parameter space but also improves sampling efficiency, providing a reliable parameter basis for subsequent optimization of composite material performance.

[0095] S103. Calculate the stress concentration factor of each candidate key parameter combination under standard load.

[0096] Preferably, the calculation process is as follows:

[0097] Based on various candidate key parameter combinations, liquid metal particles of the liquid metal component are embedded in the cement matrix to construct multiple composite material geometric simulation models. Uniaxial compression and shear simulations are performed on each composite material geometric simulation model to obtain the strain concentration tensor and effective elastic stiffness tensor of each composite material geometric simulation model under standard load. Based on the strain concentration tensor and the effective elastic stiffness tensor, the stress concentration factor of each composite material geometric simulation model is obtained.

[0098] In this embodiment, after obtaining multiple candidate key parameter combinations, a corresponding composite material geometric simulation model is constructed using these combinations. For example, taking a candidate key parameter combination as an example, its parameters are: liquid metal volume fraction 19.1%, particle size 47 μm, surface modification material is silane coupling agent, water-cement ratio 0.41, and porosity 18%. A representative volume element model is constructed using periodic boundary conditions. The calculation process for the dimensions of the composite material geometric simulation model is as follows:

[0099]

[0100] In the formula, This represents the total number of liquid metal particles in the composite material geometric simulation model. The volume of a single liquid metal particle. Let be the side length of the material geometry model. This represents the target volume fraction of the liquid metal.

[0101] When establishing the geometric model (i.e., the geometric simulation model) in Abaqus / CAE, the liquid metal particles were arranged using a random distribution algorithm to ensure that the particle spacing conformed to statistical distribution. The cement adopted an elastoplastic constitutive model, and its parameters were determined through micromechanical model calculations based on a water-cement ratio of 0.41 and a porosity of 18%. The interface between the liquid metal and cement was simulated using a cohesive model, and the interface strength parameters were determined based on the surface modification type: the interfacial shear strength corresponding to silane coupling agent modification was 15 MPa, and the normal bond strength was 12 MPa. Uniaxial compression and pure shear simulations were performed on each composite geometric model. In the uniaxial compression simulation, a 1% strain load was applied along the Z-axis, using a quasi-static analysis step and selecting the geometric nonlinearity option. In the pure shear simulation, a 1% shear strain was applied in the XY plane. After finite element analysis post-processing, the strain condensation tensor and effective elastic stiffness tensor of each model were extracted. The strain condensation tensor was obtained by calculating the ratio of the maximum strain inside the model to the applied strain, using the following formula:

[0102]

[0103] in, This represents the maximum strain component occurring within the model. For the applied macroscopic strain, For strain concentrated tensors.

[0104] The effective elastic stiffness tensor is calculated using the volume-average stress-strain relationship, and its formula is as follows:

[0105]

[0106] in, For the effective elastic stiffness tensor, The volume average stress, The strain is the volume average strain.

[0107] Based on the obtained strain concentration tensor and effective elastic stiffness tensor, the stress concentration factor of each model is calculated. The calculation considers all independent components of the tensor, and the maximum value is taken as the stress concentration factor of that model. For example, taking a certain combination result as an example, the stress concentration factor calculated by the uniaxial compression simulation is 2.15, and the stress concentration factor calculated by the shear simulation is 1.89. The larger value, 2.15, is taken as the final stress concentration factor of this combination. The stress concentration factors of all 15 candidate combinations are calculated using the above method, and the smallest stress concentration factor, 2.15, is finally determined. Therefore, it is determined as the initial composite material mix proportion. This embodiment achieves accurate prediction from microstructure to macroscopic mechanical properties through a systematic numerical simulation method, providing a reliable mechanical basis for composite material mix proportion optimization.

[0108] Preferably, in the above embodiments, the stress concentration factor of each composite material geometric simulation model is obtained based on the strain concentration tensor and the effective elastic stiffness tensor, including:

[0109] Based on the effective elastic stiffness tensor and macroscopic strain, the macroscopic average stress of each composite material geometric simulation model is obtained; based on the strain concentration tensor, the inclusion average stress of each composite material geometric simulation model is obtained; based on the inclusion average stress, the corresponding inclusion equivalent stress is obtained; based on the macroscopic average stress, the corresponding macroscopic equivalent stress is obtained; the ratio of the inclusion equivalent stress to the macroscopic equivalent stress of each composite material geometric simulation model is calculated to obtain the stress concentration factor of each composite material geometric simulation model.

[0110] In this embodiment, after obtaining the strain concentration tensor and effective elastic stiffness tensor based on each composite material geometric model, the average stress of the entire representative volume element is directly calculated using the effective elastic stiffness tensor and the applied macroscopic strain. Then, the average stress inside the inclusion is calculated using constitutive relations. The stress concentration factor is typically defined as the ratio of the local value of the maximum principal stress or equivalent stress (such as von Mises stress) to the average value. Therefore, to obtain a scalar factor, the ratio of equivalent stress is usually calculated. Specifically, the equivalent stress inside the inclusion is calculated based on the average stress inside the inclusion, and the macroscopic equivalent stress is obtained based on the macroscopic average stress. Then, the ratio of the equivalent stress inside the inclusion to the macroscopic equivalent stress is calculated to obtain the stress concentration factor. The stress concentration factor of all composite material geometric models is calculated using the method proposed in this embodiment. The smallest stress concentration factor indicates the most uniform stress distribution inside the composite material under this mix proportion, and therefore it is determined as the initial composite material mix proportion, i.e., step S104 below.

[0111] S104. The candidate key parameter combination corresponding to the minimum stress concentration factor is determined as the initial composite material mix design; when the minimum stress concentration factor is lower than the preset safety threshold, the initial composite material mix design is determined as the final composite material mix design.

[0112] In this embodiment, the relationship between the minimum stress concentration factor and the preset threshold is determined. If it is less than the threshold, the initial composite material ratio is determined as the final composite material ratio. If it is greater than or equal to the threshold, the following method step S105 is executed.

[0113] S105. When the minimum stress concentration factor is not lower than the preset safety threshold, identify the stress concentration area from the stress cloud map of the corresponding composite material model; based on the spatial distribution characteristics of the stress concentration area and the geometric morphology of the material phase interface, deduce the micro-stress identification result of the stress concentration area in reverse; adjust the candidate key parameter combination with the micro-stress identification result until the obtained stress concentration factor is lower than the preset safety threshold.

[0114] In the above embodiments, the stress cloud map of the composite material corresponding to the minimum stress concentration factor is obtained, and the area with the highest stress value in the stress cloud map is taken as the stress concentration area, i.e., the area to be analyzed. For example, the preset threshold of the stress concentration factor is set to 2.0. The basis for determining this threshold is that when the stress concentration factor is lower than 2.0, the fatigue life of the composite material under cyclic loading can meet the requirements of engineering applications, while when it exceeds 2.0, the material is prone to microcracks in the stress concentration area, which significantly affects service durability. The minimum stress concentration factor among all candidate composite material proportions is calculated to be 2.15. Since 2.15 > 2.0, the result is that the requirement is not met, and the next step of stress cloud map analysis is required. The von Mises stress cloud map of the composite material obtained from the initial composite material proportion under uniaxial compressive loading is extracted. Through the Abaqus post-processing module, the cloud map display range is set to 0-400MPa, and Jet chromatography (blue indicates low stress, red indicates high stress) is used for visualization. Analysis of the stress cloud map revealed a distinct red high-stress zone in the liquid metal particle aggregation area, with a maximum stress value of 378.6 MPa, located within a narrow matrix channel formed by three liquid metal particles. A cubic region with a side length of 50 μm was defined as the analysis area, centered on the maximum stress point. This region contained three liquid metal particles, a minimum matrix channel width of 8.2 μm, an average von Mises stress of 285.4 MPa, and a stress gradient of 12.3 MPa / μm. Since the minimum stress concentration factor of 2.15 was greater than the preset threshold of 2.0, the system automatically entered the parameter adjustment and optimization process, using the geometric features and stress distribution data of the analysis area as input for subsequent stress aggregation cause analysis (i.e., microscopic stress identification results) and parameter adjustment. This embodiment, through setting a clear quantification threshold and a systematic region identification method, achieved automatic decision-making from stress concentration factor judgment to specific problem area location, providing a precise direction for subsequent ratio optimization.

[0115] Preferably, in the above embodiments, if the spatial distribution characteristics of the stress accumulation zone are distributed along the interface between the liquid metal particles and the cement, then the cause of accumulation is analyzed, and the interfacial bonding in the cause of accumulation is determined as the first micro-stress identification result.

[0116] If the spatial distribution characteristics of the stress accumulation zone are that it is located between multiple liquid metal particles, for example, in the narrowest region between multiple liquid metal particles, then the cause of accumulation is analyzed, and the smaller particle spacing in the cause of accumulation is identified as the second micro-stress identification result.

[0117] If the spatial distribution characteristics of the stress accumulation zone are that it is far away from the liquid metal particles and located in the cement, then the cause of accumulation is analyzed, and the insufficient cement strength in the cause of accumulation is identified as the third micro-stress identification result.

[0118] In this embodiment, when the location of the area to be analyzed is detected along the interface between the liquid metal particles and cement, the microscopic cause / identification result of stress accumulation is determined to be insufficient interfacial bonding. Specifically, in the stress cloud map analysis, if the maximum stress point is found to be located at the interface between the liquid metal particles and cement, and the high-stress area is distributed in a continuous ring shape, covering approximately 60% of the particle circumference, and a significant stress peak is observed at the interface by extracting the stress distribution curve along the interface path, the interfacial shear stress reaches 25.8 MPa, while the internal stress of the matrix is ​​only 185.3 MPa. This stress concentration mode indicates that due to the modulus mismatch between the liquid metal (elastic modulus 16.5 GPa) and cement (elastic modulus 28.4 GPa), the interface bears excessive shear stress during load transfer, and the interfacial strength provided by the current silane coupling agent modification is insufficient to effectively coordinate deformation.

[0119] When the area to be analyzed is located in a narrow region between multiple liquid metal particles, the microscopic cause of stress accumulation is determined to be excessively small particle spacing. For example, a high-stress region was identified as being located within a triangular area formed by three liquid metal particles. The minimum width of the matrix channel in this region was only 6.5 μm, while the statistical average particle spacing of this combination was 18.3 μm. Finite element analysis showed a significant "bridging effect" within this narrow channel, with highly dense flow lines, resulting in a local stress of 356.8 MPa, which is 2.8 times the average stress. Spatial distribution analysis of the particles further confirmed the presence of local agglomeration in this region, with the ratio of particle center-to-diameter distance being only 1.15, far below the design requirement of 1.5.

[0120] When the area to be analyzed is far from the interface and located inside the cement, the microscopic cause of stress accumulation / identification result is determined to be insufficient cement strength. For example, when the high-stress area is completely located inside the cement, the nearest liquid metal particle interface is 35 μm away. The water-cement ratio of this combination is 0.48, the porosity is 23%, and material performance tests show that the cement compressive strength is only 32.5 MPa. Stress analysis shows that there are extensive stress concentration areas inside the matrix, with the area of ​​stress >250 MPa accounting for 18%, and the stress distribution is not directly related to the particle location. Furthermore, it was observed that there are a large number of micropores and unhydrated particles in this area, resulting in a decrease in the load-bearing capacity of the matrix, making it prone to failure even under normal stress levels. If the area to be analyzed is located inside the liquid metal particles, it indicates that there is a high-stress area inside the particles, and the type of liquid metal is changed. This embodiment achieves rapid and accurate diagnosis of defect types by establishing the correspondence between spatial location characteristics and stress accumulation causes, providing a reliable basis for subsequent targeted parameter adjustments.

[0121] Furthermore, in the above embodiments, adjusting the candidate key parameter combinations based on the micro-stress identification results includes:

[0122] If the stress accumulation is caused by interfacial adhesion, the surface modification material data in the candidate key parameters should be adjusted; if the stress accumulation is caused by small particle spacing, the doping amount data and liquid metal particle size in the candidate key parameters should be adjusted; if the stress accumulation is caused by insufficient composite material strength, the water-cement ratio and chemical composition in the candidate key parameters should be adjusted.

[0123] In this embodiment, as Figure 2 As shown, Figure 2 A strategy map for adjusting candidate key parameters was used. When the cause of stress accumulation was determined to be insufficient interfacial adhesion, the focus was on optimizing the surface modification material data. To address the existing interfacial issues, the surface modification material could be adjusted; for example, changing it from "silane coupling agent modification" to "thiol compound modification," while keeping other parameters unchanged to obtain the final composite material formulation. Analysis of the composite material formed by the final formulation showed that the interfacial shear strength increased from 15 MPa to 22 MPa. Furthermore, finite element analysis verified that after adjustment, the maximum stress value at the same location decreased from 378.6 MPa to 285.3 MPa, and the stress concentration factor improved from 2.15 to 1.72, significantly lower than the preset threshold of 2.0.

[0124] When the stress aggregation was determined to be caused by excessively small particle spacing, the doping data and liquid metal particle size were simultaneously optimized. To address the particle agglomeration problem, a dual adjustment was implemented: the liquid metal volume fraction was reduced from 22% to 18%, while the particle size was reduced from 45 μm to 35 μm. After adjusting the final composite material's proportions, the uniformity of particle spatial distribution was significantly improved, and the minimum matrix channel width increased from 6.5 μm to 12.8 μm. Simulation calculations showed that the stress level in the original high-stress region decreased from 356.8 MPa to 241.6 MPa, the stress concentration factor was optimized from 2.34 to 1.65, and the material conductivity remained at a satisfactory level of 2.8 × 10³ S / m.

[0125] When the stress concentration was determined to be caused by insufficient cement strength, adjustments were made to the water-cement ratio and chemical composition. To address the matrix strength issue, the water-cement ratio was reduced from 0.48 to 0.40, while the C3S content in the chemical composition was increased from 52% to 58%, resulting in the final composite material mix. After the adjustment, the compressive strength of the cement increased from 32.5 MPa to 46.8 MPa, and the porosity decreased from 23% to 16%. Stress analysis showed that the area of ​​high-stress regions within the matrix decreased from 18% to 6%, the maximum stress value decreased from 295.4 MPa to 216.8 MPa, and the stress concentration factor improved from 2.28 to 1.58. Through these targeted parameter adjustments, all optimized mix combinations met the technical requirement of a stress concentration factor less than 2.0. After comprehensive performance evaluation, the final composite material mix was determined, with a stress concentration factor of 1.72, the lowest among all schemes, while maintaining a good balance of overall performance. This achieved efficient and precise optimization of the composite material mix, improving the success rate and reliability of material design.

[0126] Another embodiment of the present invention provides a system for determining the proportion of liquid metal cement composite materials, such as... Figure 3 As shown, it illustrates a block diagram of a liquid metal cement composite material proportioning determination system 300 provided in an embodiment of this application, including:

[0127] The acquisition module 301 is used to acquire multiple key parameters and their ranges corresponding to the liquid metal component and the cement matrix, respectively, to obtain a set of key parameters;

[0128] Prediction module 302 is used to construct a multi-scale prediction model, and perform multi-objective performance simulation on the parameter combinations in the key parameter set based on the multi-scale prediction model to screen and obtain multiple candidate key parameter combinations.

[0129] Calculation module 303 is used to calculate the stress concentration factor of each candidate key parameter combination under standard load;

[0130] The initial module 304 is used to determine the candidate key parameter combination corresponding to the minimum stress concentration factor as the initial composite material mix design; when the minimum stress concentration factor is lower than a preset safety threshold, the initial composite material mix design is determined as the final composite material mix design.

[0131] The adjustment module 305 is used to identify stress concentration areas from the stress cloud map of the corresponding composite material model when the minimum stress concentration factor is not lower than the preset safety threshold; based on the spatial distribution characteristics of the stress concentration area and the geometric morphology of the material phase interface, reversely deduce the micro-stress identification result of the stress concentration area; and adjust the candidate key parameter combination with the micro-stress identification result until the obtained stress concentration factor is lower than the preset safety threshold.

[0132] The method and system for determining the mix proportion of liquid metal cement composite materials provided in this invention have the following beneficial effects:

[0133] (1) This invention provides a method for determining the mix proportion of liquid metal cement composite materials. By integrating multi-scale modeling, multi-objective performance simulation and stress-driven closed-loop optimization mechanism, the scientificity, accuracy and engineering practicality of composite material mix design are significantly improved.

[0134] (2) In this embodiment of the invention, the key parameters of the liquid metal component and the cement matrix are first systematically acquired and a structured parameter set is constructed to lay the data foundation for subsequent intelligent optimization. Then, a multi-scale prediction model covering micro-interface, meso-structure and macro-response is constructed to achieve collaborative simulation and efficient screening of multiple target performances such as strength, thermal conductivity and electrical conductivity, avoiding performance imbalance caused by isolated evaluation. Next, this embodiment of the invention introduces stress concentration factor as the core criterion and links it with the preset safety threshold to form a closed-loop optimization path of "performance initial screening - stress verification - mechanism diagnosis - parameter feedback": when the stress concentration factor of the initial mix design does not meet the safety requirements, not only is the high stress area identified, but also the spatial distribution characteristics of the stress cloud map and the geometric morphology of the phase interface are combined to reversely deduce the micro-stress identification results that cause stress accumulation, so as to adjust the key parameters in a targeted manner, rather than blindly iterating.

[0135] (3) The embodiments of the present invention closely and deeply link macroscopic mechanical response with microstructure evolution. The entire process is highly automated and scalable, and can be seamlessly integrated into the material digital twin platform. It supports rapid response to customized needs in different engineering scenarios, thereby enabling the determined liquid metal cement composite material ratio to significantly reduce local stress peaks while maintaining excellent functional characteristics (such as high thermal conductivity and electromagnetic shielding), improving structural durability and service safety. It provides reliable technical support for application scenarios such as intelligent building materials, flexible electronic substrates and extreme environment protection structures, and promotes the intelligent process in the field of composite materials.

[0136] The technical features of the above embodiments can be combined in any way. For the sake of brevity, 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.

[0137] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for determining the mix proportion of a liquid metal cement composite material, characterized in that, include: Obtain multiple key parameters and their ranges for each of the liquid metal component and the cement matrix to obtain a set of key parameters. A multi-scale prediction model is constructed, and multi-objective performance simulation is performed on the parameter combinations in the key parameter set based on the multi-scale prediction model to obtain multiple candidate key parameter combinations. Calculate the stress concentration factor for each of the candidate key parameter combinations under standard load; The candidate key parameter combination corresponding to the minimum stress concentration factor is determined as the initial composite material mix design; when the minimum stress concentration factor is lower than a preset safety threshold, the initial composite material mix design is determined as the final composite material mix design. When the minimum stress concentration factor is not lower than the preset safety threshold, stress concentration areas are identified from the stress cloud map of the corresponding composite material model; Based on the spatial distribution characteristics of the stress accumulation zone and the geometric morphology of the material phase interface, the micro-stress identification result of the stress accumulation zone is derived in reverse. The candidate key parameter combination is adjusted based on the micro-stress identification results until the obtained stress concentration factor is lower than the preset safety threshold.

2. The method for determining the mix proportion of liquid metal cement composite material as described in claim 1, characterized in that, The process involves obtaining multiple key parameters and their ranges corresponding to both the liquid metal component and the cement matrix, resulting in a set of key parameters, including: The knowledge graph of liquid metal and cement is embedded and learned based on graph neural network to identify the coupling path between the liquid metal component and the cement matrix. The coupling path includes at least the stress transfer induced by thermal expansion mismatch. Map the coupling path to a structured coupling relationship tensor; Based at least on the structured coupling tensor, an artificial intelligence model is used to determine the first set of key parameters corresponding to the liquid metal component and the second set of key parameters corresponding to the cement matrix; The first set of key parameters and the second set of key parameters are integrated to obtain the set of key parameters.

3. The method for determining the mix proportion of liquid metal cement composite material as described in claim 1, characterized in that, The process involves constructing a multi-scale prediction model, performing multi-objective performance simulations on parameter combinations within the key parameter set based on the multi-scale prediction model, and filtering out multiple candidate key parameter combinations, including: A first performance prediction model with self-healing efficiency as the prediction target, a second performance prediction model with electrical conductivity as the prediction target, and a third performance prediction model with compressive strength as the prediction target are constructed respectively. The first performance prediction model, the second performance prediction model, and the third performance prediction model are coupled through a parameter transfer chain to form a cross-scale performance prediction integrated model. Based on the parameter range of each of the key parameters, the key parameters are combined and arranged to obtain multiple combination results; The performance prediction of each of the combined results is predicted using the cross-scale performance prediction ensemble model to obtain the performance prediction result corresponding to each of the combined results. Based on the selected criteria, multiple candidate key parameter combinations are obtained from the performance prediction results.

4. The method for determining the mix proportion of liquid metal cement composite material as described in claim 3, characterized in that, Based on the parameter range of each of the key parameters, the key parameters are combined and arranged to obtain multiple combination results, including: Determine the type of parameter range for each of the key parameters, and use the sampling method corresponding to the type to determine multiple sampled values ​​for each of the key parameters; Multiple sampled values ​​of each of the key parameters are arranged and combined to obtain multiple combined results.

5. The method for determining the mix proportion of liquid metal cement composite material as described in claim 4, characterized in that, The process of determining the parameter range type of each of the key parameters and using the sampling method corresponding to the type to determine multiple sampled values ​​of each of the key parameters includes: When the parameter range of the key parameter is continuous, the parameter range of the key parameter is divided into multiple equally probable intervals; In each of the equal probability intervals, a sample value is determined to obtain multiple sample values ​​of the key parameter; When the parameter range of the key parameter is discrete, multiple sample values ​​are randomly selected to obtain multiple sample values ​​of the key parameter.

6. The method for determining the mix proportion of liquid metal cement composite material as described in claim 1, characterized in that, The calculation of the stress concentration factor for each of the candidate key parameter combinations under standard loads includes: Based on the various candidate key parameter combinations, liquid metal particles of the liquid metal component are embedded in the cement matrix to construct multiple composite material geometric simulation models. Uniaxial compression and shear simulations were performed on each of the composite material geometric simulation models to obtain the strain condensation tensor and effective elastic stiffness tensor of each composite material geometric simulation model under standard load. Based on the strain concentration tensor and the effective elastic stiffness tensor, the stress concentration factor of each composite material geometric simulation model is obtained.

7. The method for determining the mix proportion of liquid metal cement composite material as described in claim 6, characterized in that, The stress concentration factor of each composite material geometric simulation model is obtained based on the strain concentration tensor and the effective elastic stiffness tensor, including: Based on the effective elastic stiffness tensor and macroscopic strain, the macroscopic average stress of each composite material geometric simulation model is obtained; Based on the strain concentration tensor, the average stress within the inclusions of each of the composite material geometric simulation models is obtained; Based on the average stress within the inclusion, the corresponding equivalent stress within the inclusion is obtained; Based on the aforementioned macroscopic average stress, the corresponding macroscopic equivalent stress is obtained; The ratio of the equivalent stress within the inclusions to the equivalent stress in the macroscopic region of each composite material geometric simulation model is calculated to obtain the stress concentration factor of each composite material geometric simulation model.

8. The method for determining the mix proportion of liquid metal cement composite material as described in claim 1, characterized in that, The method of reversely deriving the micro-stress identification result of the stress accumulation region based on the spatial distribution characteristics of the stress accumulation region and the geometry of the material phase interface includes: If the spatial distribution characteristics of the stress accumulation zone are distributed along the interface between the liquid metal particles and cement, then the cause of accumulation is analyzed, and the interfacial bonding in the cause of accumulation is determined as the first micro-stress identification result. If the spatial distribution characteristics of the stress accumulation zone are that it is located between multiple liquid metal particles, then the cause of accumulation is analyzed, and the particle spacing in the cause of accumulation is determined as the second micro-stress identification result. If the spatial distribution characteristics of the stress accumulation zone are that it is far away from the liquid metal particles and located in the cement, then the cause of accumulation is analyzed, and the cement strength in the cause of accumulation is determined as the third micro-stress identification result. The first micro-stress identification result, the second micro-stress identification result, or the third micro-stress identification result are integrated to deduce the micro-stress identification result of the stress accumulation area in reverse.

9. The method for determining the mix proportion of liquid metal cement composite material as described in claim 8, characterized in that, The adjustment of the candidate key parameter combination based on the micro-stress identification results includes: If the micro-stress identification result includes the first micro-stress identification result, then the surface modification material data in the candidate key parameter combination shall be adjusted. If the micro-stress identification result includes the second micro-stress identification result, then the doping data and liquid metal particle size in the candidate key parameter combination are adjusted; If the micro-stress identification result includes the third micro-stress identification result, then the water-cement ratio and chemical composition in the candidate key parameter combination are adjusted.

10. A system for determining the proportion of liquid metal cement composite materials, characterized in that, include: The acquisition module is used to acquire multiple key parameters and their ranges corresponding to the liquid metal component and the cement matrix, and obtain a set of key parameters. The prediction module is used to construct a multi-scale prediction model, and perform multi-objective performance simulation on the parameter combinations in the key parameter set based on the multi-scale prediction model to screen and obtain multiple candidate key parameter combinations. The calculation module is used to calculate the stress concentration factor of each candidate key parameter combination under standard load; An initial module is used to determine the candidate key parameter combination corresponding to the minimum stress concentration factor as an initial composite material mix design; when the minimum stress concentration factor is lower than a preset safety threshold, the initial composite material mix design is determined as the final composite material mix design. The adjustment module is used to identify stress concentration areas from the stress cloud map of the corresponding composite material model when the minimum stress concentration factor is not lower than the preset safety threshold. Based on the spatial distribution characteristics of the stress accumulation zone and the geometric morphology of the material phase interface, the micro-stress identification result of the stress accumulation zone is derived in reverse. The candidate key parameter combination is adjusted based on the micro-stress identification results until the obtained stress concentration factor is lower than the preset safety threshold.

Images