Digital intelligent control preparation method of high-solid-waste low-carbon high-durability concrete connecting material

Abstract

The invention relates to a digital intelligent regulation and control preparation method of a high-solid-waste low-carbon high-durability concrete connecting material, and solves the problems that an existing UHPC connecting material is difficult to balance high performance and low carbon, the solid waste resource utilization rate is low, a digital intelligent precise regulation and control means is lacked, and the cost is low. The method comprises the following steps: firstly, integrating data to construct a special database for the ultra-high performance concrete, establishing a raw material-process-performance mapping relation through multi-scale simulation, then optimizing target parameters through machine learning, preprocessing the raw materials, and finally obtaining the ultra-high performance concrete after the target parameters are optimized and the raw materials are preprocessed. And verifying the iterative model by using a high-throughput experiment, determining optimal process parameters, importing the optimal process parameters into a system, and executing stirring, forming and curing. The method has the following beneficial effects that high solid waste recycling and low-carbon emission reduction are achieved through digital intelligent regulation and control, high strength and high durability of materials are synchronously guaranteed, and the core application requirements of an assembly type structure are precisely met.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of building materials, and in particular to a digital control preparation method of a high-solid-waste low-carbon high-durability concrete connecting material. BACKGROUND

[0002] With the rapid development of modern bridge engineering and urban modular construction, the performance of the connecting material for the fabricated structure has dual core requirements: on the one hand, it needs to have high strength and high durability to adapt to the complex service environment such as coastal high-chloride salt; on the other hand, it needs to meet the needs of low carbon and efficient use of resources. Ultra-high performance concrete (UHPC) has a dense microstructure and excellent mechanical properties, and becomes the preferred connecting material, but the existing technology has not formed a mature scheme considering high performance and low carbonization, and it is urgent to break through the bottleneck through innovative technology.

[0003] In the current related technology, the UHPC connecting material is mostly constructed with high cement content as the core cementitious system, and is prepared through traditional empirical proportioning design and conventional process, and industrial solid waste is mostly added as an auxiliary admixture at a low proportion, and a systematic optimization utilization scheme has not been formed.

[0004] The core problem of the existing related technology is that the dual targets of high performance and low carbonization cannot be effectively balanced. On the one hand, high cement content is relied on to ensure strength, which leads to high carbon emission of the cementitious system; on the other hand, there is a lack of precise design and control means, the utilization rate of solid waste is low, and it is difficult to stably guarantee the long-term durability of the material in the complex environment, and the material cannot fully adapt to the actual application requirements of the fabricated structure. SUMMARY

[0005] In order to realize high solid waste resourceization and low carbon emission through digital control, and simultaneously guarantee high strength and high durability of the material, and accurately adapt to the core application requirements of the fabricated structure, the application provides a digital control preparation method of a high-solid-waste low-carbon high-durability concrete connecting material.

[0006] The application provides a digital control preparation method of a high-solid-waste low-carbon high-durability concrete connecting material, which adopts the following technical scheme:

[0007] The digital control preparation method of the high-solid-waste low-carbon high-durability concrete connecting material comprises the following steps:

[0008] Obtaining and integrating the basic physicochemical parameters, historical process parameters and performance data of raw materials, and constructing a special material database for ultra-high performance concrete;

[0009] Based on the special material database for ultra-high performance concrete, the basic mapping relationship of raw materials-process-performance is constructed through multi-scale calculation simulation and the parameter boundary is demarcated, and then the machine learning model is used to nonlinearly fit and optimize the mapping relationship and boundary, and the raw material ratio and key process parameters meeting the preset high solid waste content, preset low carbon emission and preset high durability targets are screened:

[0010] According to the screening results, the raw material pretreatment parameters are determined, the industrial solid waste is activated, the fiber is surface modified, and the aggregate is combined according to the grading;

[0011] Based on the pretreated raw materials, the key process parameters and the preset experimental scheme determined according to the foregoing screening are used to carry out a plurality of parallel high-throughput experiments by using automatic equipment, the process and performance data are collected by using the preset monitoring equipment, and the data are fed back to the special material database for ultra-high performance concrete as new samples to iteratively calibrate the foregoing machine learning model until the deviation between the predicted value and the measured value of the model reaches the preset deviation threshold, so that the optimal process parameters suitable for industrial production are determined.

[0012] The optimal process parameters are imported into the production control system, and mixing, molding and curing are performed according to the preset production process to prepare the high-solid-waste low-carbon high-durability concrete connecting material. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 is a flowchart of a digital intelligent preparation method of a high-solid-waste low-carbon high-durability concrete connecting material according to an embodiment of the present application. DETAILED DESCRIPTION

[0014] The present application will be further described in detail below with reference to the accompanying drawings.

[0015] Reference Figure 1 A digital intelligent preparation method of a high-solid-waste low-carbon high-durability concrete connecting material is disclosed, which comprises:

[0016] In step S100, the basic physicochemical parameters of raw materials, historical process parameters and performance data are obtained and integrated to construct a special material database for ultra-high performance concrete.

[0017] The specific process is described as follows: First, obtain the basic physicochemical parameters of raw materials from the laboratory. These parameters are obtained through precise instrument testing to ensure the accuracy and reliability of the data. For example, the chemical composition of raw materials is determined using an X-ray fluorescence spectrometer (XRF), and the particle size distribution of slag powder is determined by a laser particle size analyzer. Then, integrate the past production process parameters and performance test data of the enterprise. These data reflect the actual performance of materials under different process conditions. For example, process parameters such as stirring time, curing temperature, and vibration frequency are usually stored in the production management system of the enterprise. Through data mining technology, these scattered data can be integrated and cleaned to ensure the consistency and integrity of the data. At the same time, performance data such as compressive strength, tensile strength, and durability are obtained through laboratory tests, such as compressive strength testing by a pressure testing machine and chloride ion permeability testing by electrochemical methods. Finally, integrate these data into a unified database. Through data standardization processing, ensure the consistency of data format from different sources, and facilitate subsequent analysis and use.

[0018] In step S200, based on the special material database for ultra-high performance concrete, a basic mapping relationship between raw materials, process, and performance is constructed through multi-scale calculation simulation, and parameter boundaries are drawn. Then, a machine learning model is used to nonlinearly fit and optimize the mapping relationship and boundaries to screen raw material ratios and key process parameters that meet the preset high solid waste content, preset low carbon emission, and preset high durability targets.

[0019] Wherein, the construction of the basic mapping relationship between raw materials, process, and performance through multi-scale calculation simulation and the drawing of parameter boundaries can refer to steps S210 to S240, or refer to steps S2A0 to S2D0. The nonlinear fitting and optimization of the mapping relationship and boundaries by the machine learning model can refer to steps SA00 to SE00, which will not be repeated here.

[0020] This step integrates the microscopic thermodynamic / kinetic model, mesoscopic particle accumulation model, and macroscopic finite element model to form a cross-scale digital material performance simulation system. Based on the first principle of physical chemistry, the system converts the characteristics of raw materials and process parameters into calculable inputs and simulates the prediction results from hydration products, microstructure to macro mechanics and durability performance. The basic mapping relationship between raw materials, process, and performance and the corresponding multi-scale parameter boundaries not only provide a physically feasible design space, but also embed key domain knowledge into subsequent machine learning models, ensuring that data-driven optimization does not deviate from the basic laws of materials science.

[0021] In step S300, the raw material pretreatment parameters are determined according to the screening results, and the industrial solid waste is activated, the fiber is surface modified, and the aggregate is combined according to the grading.

[0022] wherein the raw material pre-treatment parameters refer to specific processing conditions set to optimize the performance of raw materials, such as grinding fineness, type and concentration of activator, fiber surface modification methods, etc. These parameters are screened through the machine learning model in step S200 to ensure optimal performance in subsequent preparation processes. Activity activation: improving the reactivity of industrial solid waste (such as slag powder, fly ash) through physical or chemical methods, so that it can more effectively participate in the hydration reaction in concrete. Common methods include ultrafine grinding and adding chemical activators. Fiber surface modification: improving the properties of fiber surface through chemical or physical methods, enhancing the bonding performance of fiber and concrete matrix, improving the dispersion and toughening effect of fiber in concrete. Common methods include silane coupling agent treatment. Aggregate gradation combination: optimizing the particle size distribution of aggregate according to the particle packing model to achieve the best packing density and porosity, improving the density and strength of concrete.

[0023] The specific process is as follows: 1. Data input and preliminary screening: input the raw material ratio and key process parameters screened by the machine learning model in step S200 into the pre-treatment parameter analysis system. Based on historical data and experimental results, combined with pre-set performance targets (such as high solid waste content, low carbon emissions, high durability), the input parameters are preliminarily screened. For example, the system recommends slag powder content ≥40%, fly ash content 10%-20%, steel fiber content 1.5%-2.5%, polypropylene fiber content 0.05%-0.15% and other parameters, combined with the actual performance of these parameters in historical data, to preliminarily determine the range of pre-treatment parameters.

[0024] 2. Intelligent analysis and optimization:

[0025] Grinding fineness optimization: use a laser particle size analyzer to measure the particle size distribution of slag powder and fly ash, combined with the pre-set specific surface area target (such as 450 m 2 / kg), optimize the grinding process parameters through intelligent algorithms (such as genetic algorithm). The system automatically adjusts the speed of the ball mill and the grinding time according to the particle size distribution data to ensure that the specific surface area after grinding reaches the target value.

[0026] Chemical activator optimization: test the effect of different concentrations of sodium sulfate on the activity of slag powder through experiments. The system predicts the optimal activator concentration using a machine learning model (such as random forest) based on experimental data and pre-set activity targets (such as 30% early strength improvement). For example, the system predicts that adding 3% sodium sulfate can effectively activate the activity of slag powder and promote its early hydration reaction.

[0027] Fiber surface modification optimization: Steel fibers are surface modified using silane coupling agents. The system adjusts the concentration of silane coupling agents and treatment time to ensure that a uniform modified layer is formed on the surface of the fibers based on preset bonding performance targets (e.g., a 50% increase in the bonding strength of the fibers to the matrix).

[0028] Aggregate grading optimization: The grading of the aggregate is optimized based on the Andreasen & Andersen particle packing model. The system adjusts the ratio of quartz sand and tailings micro-powder based on preset packing density and porosity targets (e.g., porosity ≤ 5%) using intelligent algorithms (e.g., particle swarm optimization algorithm). The system automatically calculates and adjusts the mixing ratio of the aggregate to ensure the best packing density and porosity.

[0029] 3. Parameter verification and adjustment: The preliminary determined pretreatment parameters are verified by experiments. For example, the specific surface area of the ground slag powder is tested to ensure that it reaches 450 m 2 / kg; the hydration reaction experiment of the slag powder with activator is conducted to verify whether the early strength is increased by 30%; the bonding strength test of the surface modified steel fiber is conducted to ensure that the bonding strength to the matrix is increased by 50%; the compaction test and porosity test of the aggregate with optimized grading are conducted to ensure that the porosity is ≤ 5%. Based on the experimental results, the pretreatment parameters are fine-tuned.

[0030] Step S400, based on the pretreated raw materials, the key process parameters determined by the foregoing screening and the preset experimental scheme, a plurality of parallel high-throughput experimental verifications are carried out using automatic equipment, process and performance data are collected through preset monitoring equipment and fed back to the ultra-high performance concrete special material database, which is used as a new sample to iteratively calibrate the foregoing machine learning model until the deviation between the predicted value of the model and the measured value of the experiment reaches the preset deviation threshold, thereby determining the optimal process parameters suitable for industrial production.

[0031] The specific process is as follows:

[0032] 1. High-throughput experimental verification:

[0033] Experimental preparation: Based on the pretreated raw materials in step S300, a plurality of experimental samples are prepared according to the key process parameters determined by the screening in step S200 and the preset experimental scheme. The experimental scheme includes different process parameters such as raw material ratio, mixing time, curing temperature, and vibration frequency.

[0034] Automation equipment application: Use automated mixing equipment, molding equipment, and curing equipment to realize parallel operation of multiple groups of experiments. For example, automated mixing equipment can accurately control the mixing process according to the preset mixing time and sequence; molding equipment can ensure that the molding conditions of each sample are consistent; curing equipment can follow the preset temperature and humidity curve for curing.

[0035] Experiment execution: Under the control of automated equipment, multiple groups of experiments are carried out simultaneously to ensure the accuracy and repeatability of experimental conditions. For example, 10 groups of different proportions of ultra-high performance concrete (UHPC) samples are prepared simultaneously, and the process parameters such as mixing time and curing temperature of each group of samples are adjusted according to the preset scheme.

[0036] 2、Data collection and feedback: 2.1, monitoring equipment deployment: During the experiment, deploy preset monitoring equipment to collect process data and performance data in real time. For example, use a rheometer to monitor the rheological parameters of the slurry, use a hydration heat temperature measurement system to monitor the hydration heat release curve, use a pressure testing machine to measure the compressive strength, and use an electrochemical method to measure the chloride ion permeability resistance. 2.2, data collection: Through automated monitoring equipment, real-time collection of rheological parameters, hydration heat release curve, compressive strength, tensile strength, durability, etc. of each group of samples. These data will be used as experimental measured values for subsequent model calibration. 2.3, data feedback: Feedback the collected process data and performance data to the ultra-high performance concrete special material database. These data are added as new samples for iterative calibration of the machine learning model in step S200.

[0037] 3、Model iterative calibration: 3.1, data integration and preprocessing: integrate the data collected by high-throughput experiments into the material database, and perform data cleaning and standardization processing to ensure data consistency and usability. 3.2, model training and calibration: use the new experimental data to retrain and calibrate the machine learning model. For example, use a random forest model, use the new experimental data as a training set, optimize model parameters, and improve model prediction accuracy. 3.3, comparison of predicted values and measured values: compare the preset deviation threshold of the model predicted value and the experimental measured value to evaluate the accuracy of the model. For example, set the deviation threshold of the compressive strength to ±5MPa, if the deviation between the model predicted value and the experimental measured value exceeds the threshold, the model needs to be further optimized.

[0038] 4. Determine the optimal process parameters: 4.1, parameter optimization: according to the comparison results of model prediction value and experimental measured value, the process parameters are optimized. For example, if the compressive strength predicted by the model is lower than the experimental measured value, it may be necessary to adjust the mixing time or curing temperature and other parameters. 4.2, experimental verification: the optimized process parameters are verified by experiments again to ensure their feasibility and stability in actual production. For example, experiments are conducted on the adjusted mixing time and curing temperature to verify whether they can achieve the expected compressive strength and durability. 4.3, final determination: through multiple experimental verification and model calibration, the optimal process parameters suitable for industrial production are determined.

[0039] Step S500, the optimal process parameters are imported into the production control system, and the mixing, molding and curing are performed according to the preset production process to prepare the high-solid-waste low-carbon high-durability concrete connecting material.

[0040] The process is described as follows: 1, optimal process parameters import: the optimal process parameters determined in step S400 are imported into the production control system (MES). These parameters include mixing time, mixing sequence, vibration frequency, curing temperature, curing humidity, etc.

[0041] 2, production process execution: 2.1, mixing process: according to the optimal process parameters in the MES system, the automatic mixing equipment mixes according to the preset time and sequence. For example, first put the dry materials (such as cement, slag powder, fly ash, etc.) into the mixer and dry mix for 30 seconds, then add water and additives and continue to wet mix for 3 minutes to ensure that the raw materials are fully mixed and uniform. 2.2, molding process: the molding equipment vibrates according to the preset vibration frequency and time. For example, use a high-frequency vibration table, set the vibration frequency to 50Hz, and the vibration time to 30 seconds to ensure that the concrete is compacted and molded, reducing air bubbles and pores. 2.3, curing process: the curing equipment is cured according to the preset temperature and humidity curve. For example, put the molded concrete into the curing room, set the curing temperature to 20°C and the relative humidity to 95%, and the curing time to 28 days. During the curing process, the temperature and humidity are monitored in real time by sensors to ensure the stability and consistency of the curing conditions.

[0042] The basic mapping relationship between raw materials, process and performance is constructed through multi-scale calculation simulation, and the parameter boundary is determined, including:

[0043] Step S210, microscale simulation: using the preset thermodynamic calculation software, based on the principle of Gibbs free energy minimization, by setting different temperature conditions, the stable hydration product phase of the multi-component cementitious system is predicted, in order to avoid calcium hydroxide crystals, which are harmful to the material, and output the type, content and microstructure parameters of the hydration product.

[0044] Thermodynamic calculation software: such as GEMS or FactSage, which are based on the Gibbs free energy minimization principle and can predict the stable phase composition of a multi-component system under different conditions. Gibbs free energy minimization principle: the Gibbs free energy of a system reaches a minimum when it reaches equilibrium. This is the basic principle of thermodynamic calculation, which is used to predict the stable phase of the system. Hydration product phase: refers to the compounds with specific chemical composition and microstructure formed during the hydration process of cement, such as calcium silicate hydrate (C-S-H), calcium aluminate hydrate (C-A-H), etc. Microstructure parameters: describe the microstructure characteristics of hydration products, such as specific surface area, porosity, etc.

[0045] The process is described as follows: 1. Accurate characterization of system parameters and temperature condition setting: X-ray fluorescence spectrometer (XRF) and inductively coupled plasma emission spectrometer (ICP-OES) are used to determine the chemical composition of the cementitious system, including the mass fraction of cement clinker mineral phases (C3S, C2S, C3A, C4AF), glass content and activity index of slag powder, and the total content of SiO2+Al2O3+Fe2O3 in fly ash. The reaction condition setting uses a piecewise function to simulate the actual curing process (such as 20°C curing for 3 days→40°C curing for 4 days→20°C standard curing for 28 days), and by setting different temperature conditions (10°C, 20°C, 40°C) to simulate the temperature dependence of the hydration process. The reaction kinetics parameters are obtained by fitting the hydration exothermic rate curves at different temperatures through the Arrhenius equation (ln(k)=ln(A)-Ea / RT), and the determination error of the pre-exponential factor A and activation energy Ea is controlled within ±5%. 2. Thermodynamic equilibrium calculation and kinetics correction: To connect the thermodynamic equilibrium end state and the kinetic barrier of the actual hydration process, the Gibbs free energy database of CaO-SiO2-Al2O3-Fe2O3-MgO-SO3-H2O multi-component system is constructed in GEMS software. The number of independent variables is calculated by the Gibbs phase rule F=C-P+2, the element conservation is constrained by mass balance equation and charge balance equation, and the Newton-Raphson iteration algorithm is used to solve the nonlinear equation system (convergence criterion is residual <10 -6), and predict the stable phase composition (e.g. C-S-H, C-A-H, CH, AFt) and its equilibrium content. In practical application, the curing age (1d, 3d, 7d, 28d) needs to be combined, and the equilibrium content is converted into a function of time by introducing the Johnson-Mehl-Avrami (JMA) phase transition kinetics equation, in which the crystal nucleus growth index n and the rate constant k are calibrated by isothermal calorimetry. 3. Adverse phase avoidance and parameter verification: Set the CH volume fraction threshold <5%. When the predicted CH content exceeds the limit, automatically trigger the reverse optimization algorithm (based on the gradient descent method) to increase the slag powder content to 55-70% (preferably 55-65%) or the fly ash content to 15-28% (preferably 22-28%), while reducing the cement content, and introducing alkaline activators (Na2SO4, etc.) to adjust the liquid phase pH value to 12.5-13.0, so that the CH generation is reduced to <5%. The final output is the quantitative results of the hydration products at each age (e.g. C-S-H phase content 58-62% at 28 days). Through XRD-Rietveld refinement, mercury intrusion method (MIP), BET method, etc. verification, ensure that the deviation between the simulation value and the measured value is <10%, and the verified parameters are used as the input constraints of the mesoscale simulation.

[0046] Step S220, mesoscale simulation: based on the hydration product data output by the microscale, it is included in the particle system as a superfine powder, combined with the basic physical and chemical parameters of the aggregate and the powder, through the preset particle packing model, the packing density of the system is calculated according to the particle size distribution, the grading design is simulated and optimized in a digital way, the target porosity is virtually calculated, so as to determine the optimal packing scheme, and output the corresponding mesostructure parameters.

[0047] The specific process is as follows: 1, input data and particle packing model construction: the particle size distribution of aggregate and powder (such as quartz sand 0.1-2mm, tailing powder 0.01-0.1mm) is measured by a laser particle size analyzer, combined with the particle density measured by a densimeter, and the volume proportion of hydration product output by microscale (such as C-S-H gel 60%) and pore distribution characteristics. Select the Andreasen & Andersen particle packing model to construct a three-dimensional mesoscopic model of the aggregate-fiber-cementitious system, and the hydration product is included in the particle size grading system as a superfine powder component. 2, grading optimization and mesoscopic parameter calculation: in order to maximize the packing density (target ≥75%) and minimize the porosity (target 10%-15%), an optimization algorithm (such as genetic algorithm) is used to automatically adjust the grading ratio. Based on the three-dimensional mesoscopic model constructed by the discrete element method (DEM), the packing density of each generation scheme is calculated by the mass balance equation, and the micro-pore distribution is converted into the meso-particle gap range. After iterative optimization (such as setting population size 50, iteration 100 times), the mesoscopic structure parameters such as packing density and fiber spacing coefficient of variation (target ≤0.3) under the optimal grading are output. 3, parameter verification and cross-scale transfer: the optimized mesoscopic structure parameters (such as packing density 74-76%, porosity 10%-12%) are output. The porosity is verified by mercury intrusion method (MIP), and the particle gap distribution is analyzed by scanning electron microscope (SEM-BSE), to ensure that the deviation between the simulation value and the measured value is less than 10%. Finally, these parameters are transferred as key inputs to the macro-scale finite element model, realizing the quantitative mapping from mesoscopic structure to macroscopic performance.

[0048] Step S230, macro-scale simulation: on the basis of the optimal packing scheme and porosity parameters at the mesoscopic scale, a finite element model reflecting the mechanical properties of the matrix is established, and the finite element analysis technology is used to simulate the distribution state, orientation characteristics and stress transfer path of the fiber in the matrix, to quantify the reinforcement effect of the fiber on the macroscopic mechanical properties of the material, and then to optimize the fiber type and dosage, and to output the simulation data related to the macroscopic performance.

[0049] The specific process is as follows: 1, multi-scale data input and finite element modeling: the optimal packing scheme and porosity parameters output by the mesoscopic scale simulation are used as the input basis of the finite element model, including packing density (75%), porosity (10%) and aggregate (quartz sand particle size distribution 0.1-2mm, density 2.65g / cm 3 ) and powder (tailing powder particle size distribution 0.01-0.1mm, density 2.80g / cm 3) and the physicochemical parameters of the aggregates. In a finite element analysis platform (such as ABAQUS or ANSYS), a three-dimensional heterogeneous concrete numerical model containing aggregates, powders, and hydration products is established according to the optimal packing structure determined at the mesoscale, and the spatial distribution parameters of each component are set. 2. Numerical simulation of fiber distribution and orientation characteristics: Based on the established finite element model, input the fiber type and dosage parameters (such as steel fiber volume dosage 2%, polypropylene fiber volume dosage 0.1%), and use a random distribution algorithm (such as generating fiber position and direction that conforms to statistical rules through MATLAB) to simulate the distribution of fibers in the matrix, ensuring its randomness and uniformity in three-dimensional space. By setting the orientation probability of different directions (for example, considering the anisotropy caused by the pouring and vibrating process), the distribution state of the fibers under actual construction conditions is simulated, providing a realistic spatial configuration of the fibers for subsequent stress transfer analysis. 3. Stress transfer path simulation and quantitative analysis of enhancement effect: Apply external loads consistent with actual service conditions in the finite element model to simulate the stress transfer path of concrete under loading conditions. The stress distribution of the fiber-matrix system can be visualized using post-processing tools (such as ABAQUS / Viewer), and the inhibition mechanism of fibers on matrix stress redistribution and crack development can be analyzed. By extracting key mechanical indicators (such as compressive strength, tensile strength, and elastic modulus), the performance changes before and after fiber reinforcement are compared, and the reinforcement contribution of fibers is quantified. For example, calculate the percentage increase in compressive strength after fiber incorporation to evaluate the specific impact on macroscopic mechanical properties. 4. Fiber parameter optimization and verification output: Based on the above quantitative results, with the goal of optimizing macroscopic performance, systematically compare the reinforcement effects of different fiber types and dosage combinations to determine the optimal fiber parameters. Output the optimized fiber type, dosage, and corresponding macroscopic performance prediction data (such as compressive strength 250 MPa, tensile strength 30 MPa, and elastic modulus 50 GPa). Further verify the accuracy of the simulation results through standard mechanical experiments (such as cube compression test and splitting tensile test). If there is a significant deviation between the experimental and simulated values, feedback and correct the model parameters, and perform finite element calculations again until the results meet the preset design goals.

[0050] Step S240, mapping relationship construction and parameter boundary demarcation: Integrate the simulation results at micro, meso, and macro scales, systematically link the logical relationships between raw material properties, process parameters, and performance at each scale, and combine historical data accumulated in the special material database for ultra-high performance concrete to construct the basic mapping relationship from raw materials to performance; at the same time, according to the parameter variation range covered by each scale simulation, overall plan and demarcate the multi-scale parameter boundary matching the design target.

[0051] The specific process is as follows: 1. Multi-source data fusion and feature engineering: Integrate microscale (hydration product type and content, C-S-H gel specific surface area 400-600 m 2kg), mesoscale (bulk density 70%-80%), and macroscale (fiber content 1.5%-2.5%, compressive strength 200-300 MPa) simulation data, combined with historical process-property data in the UHPC database. Data preprocessing was performed using Z-score standardization and Min-Max normalization, high linear correlation features were removed through Pearson correlation coefficient analysis (|r|>0.8), and principal component analysis (PCA) was used to reduce the feature dimension from 45 to 8 (cumulative variance contribution rate ≥85%), providing high-quality input data for mapping relationship construction. 2. Machine learning modeling and mapping relationship construction: Gradient Boosting Decision Tree (GBDT) algorithm was selected, with raw material properties (chemical composition, particle size distribution, etc. 12-dimensional features) and process parameters (curing system, mixing parameters, etc. 6-dimensional features) as input, and macroscopic properties (compressive strength, elastic modulus, etc.) as output. The model performance was tested on a test set with a determination coefficient R 2 >0.92, and a root mean square error RMSE<5MPa, ensuring that the mapping relationship has high precision and generalization ability. SHAP value analysis was used to analyze the contribution of key parameters, and the fiber content (contribution 18%) and water-binder ratio (contribution 22%) were identified as core control parameters. 3. Multi-scale parameter boundary determination and verification: Based on multi-objective optimization theory, a systematic method was used to determine the key parameter boundaries: First, a multi-objective function was established containing solid waste content, carbon emissions, and durability indicators, and the Pareto optimal solution set was solved by Non-dominated Sorting Genetic Algorithm (NSGA-II). Based on the distribution characteristics of the solution set, K-means clustering algorithm was used to identify high-performance areas, and finally the microscale (C-S-H gel density ≥2.0 g / cm 3 , CH crystal content ≤5%), mesoscale (bulk density ≥72%, porosity ≤12%), and macroscale (compressive strength ≥220 MPa, fiber spacing variation coefficient ≤0.3) were determined. Then, parameter combinations were generated by Latin hypercube sampling, and the trained model was used to predict performance, and the feasible parameter space was determined by Pareto frontier analysis, and the boundary effectiveness was verified by an independent test set, and a closed-loop verification was completed by establishing a deviation warning mechanism.

[0052] The construction of the basic mapping relationship between raw materials, process, and performance and the determination of the parameter boundary through multi-scale calculation simulation include:

[0053] Step S2A0, microscale simulation: based on the basic physicochemical parameters of raw materials in the special material database for ultra-high performance concrete, the hydration reaction process of the cementitious system is simulated by using a hydration reaction kinetics simulation tool to predict the type of hydration product and quantify the density of hydrated calcium silicate gel, control the amount of calcium hydroxide crystals generated to not exceed the preset upper limit to ensure high durability, and output the key parameters of the volume fraction of hydration products and pore distribution.

[0054] The above process can be specifically referred to steps S2A1 to S2A5, which will not be repeated here.

[0055] Step S2B0, mesoscale simulation: based on the volume fraction of hydration products and pore distribution output by the microscale simulation, combined with the aggregate characteristic data under high solid waste content in the special material database for ultra-high performance concrete, the aggregate gradation is optimized and the fiber distribution state is evaluated by using the particle packing model and volume filling model, the inter-particle gap characteristics are quantified, the quantitative conversion relationship between micro-hydration characteristics and meso-structure is established, and the packing density and fiber spacing coefficient of variation are output as meso-structure parameters.

[0056] The above process can be specifically referred to steps S2B1 to S2B4, which will not be repeated here.

[0057] Step S2C0, macro-scale simulation: based on the meso-structure parameters, using a finite element analysis tool, a quantitative mapping relationship between meso-structure, process parameters and macro-performance is established based on the performance mapping equation, thereby relating the macro-mechanical properties, durability performance and low carbon emission indicators under high solid waste content, and outputting the preset threshold of macro-performance and low carbon indicators.

[0058] The above process can be specifically referred to steps S2C1 to S2C4, which will not be repeated here.

[0059] Step S2D0, mapping relationship construction and parameter boundary demarcation: integrate the quantitative conversion relationships and mapping relationships formed by the micro-meso-macro three-level simulation, combine the historical process-performance correlation data in the special material database for ultra-high performance concrete, construct the basic mapping relationship of raw materials-process-performance, and simultaneously demarcate the parameter boundaries of each scale that adapt to high solid waste, low carbon and high durability targets according to the parameter ranges output by each scale simulation.

[0060] The specific process is as follows:

[0061] 1. Data integration and feature engineering: Combine the hydration product volume fraction and pore distribution from step S2A0 output, the bulk density and fiber spacing coefficient of variation from S2B0 output, the macroscopic performance and low-carbon indicators from S2C0 output, and the historical formulations and measured data in the UHPC database to form a dataset covering the characteristics of raw materials (chemical composition, activity index), process parameters (mixing time, curing temperature), and target performance (strength, durability, carbon emissions). Use Z-score standardization and Min-Max normalization processing, analyze and remove high linear redundant features through Pearson correlation coefficient, and use principal component analysis to reduce the dimension to 8 dimensions (cumulative variance contribution rate ≥85%), providing high-quality input for mapping relationship construction.2. Mapping relationship construction: Select gradient boosting decision tree (GBDT) algorithm to construct a nonlinear mapping model, with raw material ratio and process parameters as input, and macroscopic performance (compressive strength ≥220MPa, tensile strength ≥20MPa) and low-carbon indicators (carbon emissions ≤45kgCO2 / m 3 ) as output. Optimize hyperparameters (learning rate 0.1, tree depth 6) through five-fold cross-validation to ensure that the model test set determination coefficient R 2 >0.92, RMSE <5MPa. Use SHAP value analysis to determine that water-binder ratio (contribution 22%), fiber content (contribution 18%), and solid waste activity index (contribution 15%) are the core control parameters, and accurately quantify the influence weight of each factor on performance.3. Multi-scale parameter boundary demarcation: With high solid waste content ≥60%, carbon emissions ≤50kgCO2 / m 3 , and chloride ion permeability ≥1000C as constraints, use NSGA-II algorithm to solve the Pareto optimal solution set, and combine K-means clustering to identify high-performance areas to comprehensively demarcate micro boundaries (C-S-H gel density ≥2.0g / cm 3 , CH crystal content ≤5%, porosity 5%-15%), mesoscopic boundaries (bulk density ≥72%, fiber spacing coefficient of variation ≤0.3, interface thickness ≤50μm), and macro boundaries (compressive strength ≥220MPa, carbon emissions ≤45kgCO2 / m 3 ).4. Verification and solidification output: Generate 100 groups of parameter combinations through Latin hypercube sampling, use the trained GBDT model to predict performance, and combine independent test set (30 groups of data not involved in training) to verify boundary effectiveness, requiring that the deviation between predicted value and actual value is <8%. Finally, output the basic mapping equation, multi-scale parameter boundary table, and model confidence evaluation report (including R 2 , RMSE, MAE, and extrapolation risk prompt), providing benchmark constraints for subsequent machine learning optimization and production control.

[0062] Predicting the type and quantifying the density of the hydration product of calcium silicate gel, controlling the amount of calcium hydroxide crystals generated to not exceed the preset upper limit to ensure high durability, outputting the volume fraction of the hydration product and the key parameters of the pore distribution including:

[0063] Step S2A1, retrieve the active ingredient content, specific surface area, and cementitious material mineral composition data of industrial solid waste from the UHPC database, and establish a multi-component hydration reaction initial parameter matrix based on the differences in reaction activity between industrial solid waste and cementitious materials.

[0064] Among them, the active ingredient content of industrial solid waste refers to the content of components with potential hydration activity in industrial solid waste, such as silicates and aluminates in slag powder. These data are obtained through chemical analysis, such as using an X-ray fluorescence spectrometer (XRF). Specific surface area: refers to the surface area per unit mass of a material, usually expressed in m 2 2 / g. The larger the specific surface area, the higher the reaction activity of the material. The specific surface area is determined by a specific surface area instrument, such as a Blaine specific surface area instrument. Cementitious material mineral composition: refers to the mineral composition of cement, slag powder, fly ash, etc., such as tricalcium silicate (C3S), dicalcium silicate (C2S), and tricalcium aluminate (C3A). These data are obtained through X-ray diffraction (XRD) analysis. Reaction activity difference: refers to the difference in activity of different cementitious materials in the hydration reaction, usually determined by experiment, such as by a hydration heat measuring instrument to measure the hydration heat of different materials. Hydration reaction initial parameter matrix: a matrix containing all the initial conditions of the hydration reaction, including active ingredient content, specific surface area, mineral composition, etc., used to simulate the initial state of the hydration reaction.

[0065] The specific process is described as follows: 1. Data retrieval: extract the active ingredient content, specific surface area, and mineral composition data of industrial solid waste and cementitious materials from the UHPC database, eliminate abnormal data with a coefficient of variation > 15%, and retain at least 30 valid data to ensure statistical significance. 2. Reaction activity analysis: determine the 3, 7, and 28-day hydration exothermic curves using isothermal calorimetry, and obtain the reaction rate constant k and activation energy Ea (slag powder: k = 0.01-0.03 s -1 , Ea = 45-55 kJ / mol; cement: k = 0.08-0.12 s -1, Ea=65-75kJ / mol), to characterize the differences in material reactivity. 3. Constructing the initial parameter matrix: Establish an n x m dimensional parameter matrix, with rows corresponding to cement, slag, fly ash, and other material components, and columns corresponding to active ingredient content, specific surface area, k, Ea, and other parameters, and including material batch, test date, and other metadata to ensure traceability. 4. Parameter correction and verification: Correct and verify the initial parameter matrix through experimental data (such as hydration product content at different ages), adjust the reaction rate constant and activation energy, and ensure its accuracy and reliability. For example, experiments show that the C-S-H content in slag powder at 7-day age is 60%, and the CH content is 5%; the corresponding values of cement are 70% and 10%, respectively. According to these data, the initial parameter matrix is optimized. 5. Output the initial parameter matrix: Output the corrected multi-component hydration reaction initial parameter matrix to provide accurate initial conditions for subsequent hydration reaction simulation, ensuring the scientificity and practicality of the simulation results.

[0066] Step S2A2, based on the hydration reaction initial parameter matrix, a hydration reaction kinetics simulation tool is used to construct a high-solid waste system hydration reaction model, a solid waste activity excitation coefficient is introduced to modify the reaction rate equation, different age hydration paths are simulated, and the types of hydration products including hydrated calcium silicate and calcium hydroxide are calculated and predicted through the reaction process, and the product generation rate curve and the characteristic parameters of the stoichiometric ratio of hydrated calcium silicate and the microstructure characteristics are output.

[0067] wherein the hydration reaction kinetics simulation tool refers to a simulation platform based on thermodynamic equilibrium and reaction kinetics coupling (such as GEMS kinetics module or Python+Cantera), which can realize time-varying hydration process simulation by inputting the initial parameter matrix; the solid waste activity excitation coefficient a is a correction factor considering the improvement of alkaline activator on the reaction activity of solid waste; the generation rate curve describes the evolution law of the content of each hydration product with age; the characteristic parameters include the stoichiometric ratio of C-S-H, specific surface area, porosity and other microstructure indicators.

[0068] The specific process is as follows: 1. Model construction and initialization: use the hydration reaction kinetics simulation tool to construct the high-solid waste system hydration reaction model. Input the hydration reaction initial parameter matrix output in step S2A1, including the active ingredient content, specific surface area, mineral composition and reaction kinetics parameters of industrial solid waste and cementitious materials. For example, input the glass content of slag powder as 85%, the specific surface area as 450m 2 / kg, the reaction rate constant k=0.015s -1 , the C3S content of cement is 55%, the specific surface area is 350m 2 / kg, the reaction rate constant k=0.10s -1, to build a reaction system consistent with the actual material properties. 2, Correction of reaction rate equation: Introduce the solid waste activity excitation coefficient α to correct the classical JMA equation: , where r is the reaction rate, k is the rate constant, n is the Avrami index (1.5-2.0), and α is the excitation coefficient. The cumulative heat release of 72h under different Na2SO4 contents (2%, 3%, 4%) is determined by experiment, and the α value is determined by inversion (such as α=1.58 for 3% content is optimal), so that the deviation between the corrected simulated heat release curve and the measured value is <5%. 3, Age simulation and product prediction: Simulate the hydration reaction process at key ages of 1, 3, 7, 28, and 90 days, use the adaptive step Runge-Kutta algorithm to solve the reaction kinetics equation set, and dynamically update the consumption and generation of each component. Through the reaction process calculation, the type and content of hydration product at each age are predicted, and the generation rate curve is generated. For example, the model predicts that at 7 days of age, the generation rate of C-S-H is 0.48mg / h, and the content is 45%, and the generation rate of CH is 0.12mg / h, and the content is 3.5%; at 28 days of age, the generation rate of C-S-H decreases to 0.12mg / h, and the content is 62%, and the content of CH is 4.2%. 4, Output characteristic parameters and model verification: Output the stoichiometric ratio of calcium silicate hydrate (C-S-H) and its microstructure characteristic parameters. Based on the simulation results, the Ca / Si ratio of C-S-H is calculated to be 1.65±0.05, the specific surface area is 520m 2 / kg, and the porosity is 12%. The measured value is corrected by XRD-Rietveld refinement to correct the simulation results, and the product content deviation at each age is required to be <8%; the porosity is verified by mercury intrusion method, and the deviation is <10% to be convergent. If the deviation exceeds the range, return to step S2A1 to adjust the initial parameter matrix, forming an iterative optimization closed loop. Finally, output the JSON format file, including age, product type and content, generation rate curve and C-S-H microstructure parameters, to provide time-varying constraints for step S2B0 mesoscale simulation.

[0069] Step S2A3, relying on the above characteristic parameters, the apparent density of calcium silicate hydrate gel is calculated by density function theory, and the measured data of the same type of gel in the special material database of ultra-high performance concrete are corrected to obtain the density value with preset precision.

[0070] , where the density function theory (DFT) simulates the electronic structure of C-S-H gel based on quantum mechanics to calculate the theoretical density; the measured data correction adjusts the model parameters by comparing the simulation value with the measured value of the same type of gel in the database; the preset precision requires that the relative error of the corrected density value is <3%.

[0071] The specific process can refer to the following: 1. Characteristic parameter integration: collect the characteristic parameters output by step S2A2, including the C-S-H stoichiometric ratio (Ca / Si=1.65±0.05), the specific surface area (520m 2 / kg) and the porosity (12%), to form the input parameter set for DFT calculation. Ensure that the parameter range is physically self-consistent with the S2A2 simulation results (C-S-H content 62%). 2. Application of density function theory: use DFT to calculate the theoretical density of C-S-H gel. Build a C-S-H molecular model (select tobermorite 14Å structure) through MaterialsStudio or VASP software, optimize the geometric structure under GGA-PBE functional, and calculate the total energy and volume. The initial calculation obtains a theoretical density of 2.62g / cm 3 . 3. Real data correction and optimization: compare the measured density (2.40g / cm 3 ) of the same type of C-S-H gel in the UHPC database, and find that the deviation is 8.2%. Adjust the porosity parameter in the DFT model (from 12% to 14.5%) and introduce surface hydroxyl modification, and recalculate the apparent density to be 2.48g / cm 3 , with a deviation of 3.3% from the measured value, which meets the preset accuracy requirement. 4. Output density value: output the corrected C-S-H gel apparent density value as 2.48±0.05g / cm 3 , and simultaneously output the correction parameters (porosity 14.5%, surface hydroxyl degree 1.2OH / nm 2 ) to the UHPC database to provide the benchmark density parameter for the volume filling calculation of step S2B0 mesoscale simulation.

[0072] Step S2A4, with the preset high durability target as the constraint, calibrates the preset upper limit of the calcium hydroxide content through the correlation data between calcium hydroxide content and durability in the special material database for ultra-high performance concrete, and monitors the generation amount in real time during simulation. When it exceeds, adjust the solid waste content or the activity excitation coefficient reversely and update the hydration reaction model synchronously.

[0073] Specific process can refer to the following: 1. Set the durability of CH generation upper limit: based on more than 200 sets of experimental data in UHPC database, using random forest or support vector regression to establish the quantitative correlation model of CH volume fraction with RCPT electric quantity, frost resistance grade, through 5-fold cross validation to determine the critical threshold value is 5% (when CH > 5%, chloride ion electric quantity > 1000C, frost resistance grade < F300), the threshold value is fixed in GEMS / Cantera thermodynamic database and set as the core constraint condition of hydration simulation, at the same time, set 4.5% as the early warning value to reserve the adjustment window. 2. Real-time monitoring of CH generation in simulation process: define CH phase as the key tracking target in GEMS or Cantera platform, use adaptive step (minimum 0.1d) to solve reaction kinetics equation, dynamically extract CH volume fraction evolution curve at 3d, 7d, 28d and 90d age; When the predicted value is ≥4.5%, the system pushes yellow warning to MES interface, ≥5% automatically triggers PID or fuzzy feedback control mechanism and suspends subsequent calculation. 3. Reverse adjustment mechanism: after triggering, start gradient descent or response surface optimization algorithm, under the multi-objective constraints of keeping total solid waste ≥60% and carbon emission ≤45kgCO2 / m 3 , synchronously adjust slag powder content (30%→35%~40%), fly ash ratio (15%→20%~25%), Na2SO4 concentration (3%→4%) and water-binder ratio (0.22 to 0.18~0.20), force secondary hydration reaction to consume CH to ≤4%, control the single adjustment range in 1%~2% to ensure the convergence of algorithm. 4. Synchronous update of hydration reaction model: immediately rewrite the initial concentration matrix of reactants, rate constant k (modified according to Arrhenius equation) and excitation coefficient α (inverted according to isothermal calorimetry value) of kinetics model after parameter adjustment, re-run the simulation until CH content is stable in the target interval, if it still fails to meet the standard for 3 consecutive iterations, trigger the secondary optimization strategy of reducing the adjustment step by 50%. 5. Experimental verification and model calibration closed loop: according to the final proportion, form 100mm cube and φ100×50mm cylinder specimens, carry out RCPT (ASTM C1202, 56d electric quantity), freeze-thaw cycle resistance (ASTM C666, 300 times cycle mass loss rate), XRD-Rietveld refinement (CH phase quantification) and MIP (porosity) tests; The deviation between simulation value and measured value should meet CH content <8%, chloride ion electric quantity <10%, frost resistance grade <1 level, otherwise the error will be transferred to the thermodynamic database to modify the activity coefficient, forming a "simulation-optimization-experiment-feedback" closed loop. 6. Output and solidification: the final output includes optimized proportion, solid waste content-activator concentration mapping table, CH dynamic control logic and updated k, α parameter set, packaged as reusable knowledge module and assigned with version number, stored in the standard interface of UHPC material database, for direct calling by mesoscale particle packing and macroscopic finite element simulation.

[0074] Step S2A5, combine the corrected gel density and the product generation rate curve to calculate the volume fraction of hydration products at different ages. Convert the microstructure evolution caused by hydration into pore distribution parameters through the pore network model. Form a pore characteristic matrix that matches the mesoscale simulation, which serves as the bottom constraint for mesoscale simulation.

[0075] The specific process is as follows: 1. Calculate the volume fraction of hydration products: Combine the corrected calcium silicate hydrate (C-S-H) gel density in step S2A3 and the product generation rate curve in step S2A2 to calculate the volume fraction of hydration products at different ages (such as 3 days, 7 days, and 28 days). Use mass conservation and density formula to convert the mass of each product into volume. For example, assuming that at 28 days of age, the generation rate of C-S-H is 0.5 mg / h and the generation rate of CH is 0.2 mg / h, the volume fraction of C-S-H is calculated to be 60% and the volume fraction of CH is calculated to be 5% through integration. 2. Application of pore network model: Convert the microstructure evolution caused by hydration into pore distribution parameters through the pore network model. The pore network model simulates the filling process of hydration products in the matrix and calculates the porosity and pore size distribution. For example, the model predicts that at 28 days of age, the porosity is 10% and the pore size distribution is mainly concentrated in the range of 0.01-0.1 μm. 3. Form a pore characteristic matrix: Integrate the calculated porosity and pore size distribution parameters into a pore characteristic matrix. This matrix describes the pore characteristics at different ages and provides the bottom constraint for mesoscale simulation. For example, the pore characteristic matrix includes parameters such as porosity, pore size distribution, and pore connectivity, ensuring that the mesoscale simulation accurately reflects the changes in microstructure. 4. Verification and optimization: Verify the accuracy of the pore characteristic matrix through experiments. For example, use mercury intrusion porosimetry (MIP) to test the porosity and pore size distribution of concrete to ensure that the simulation results are consistent with experimental data. Further optimize the pore network model based on experimental results to ensure its prediction accuracy at different ages.

[0076] Quantify the inter-particle gap characteristics, establish the quantitative conversion relationship between micro-hydration characteristics and meso-structure, and output the packing density and fiber spacing coefficient of variation as meso-structure parameters, including:

[0077] Step S2B1, based on the hydration product volume fraction and pore distribution output at the microscale, combine the aggregate characteristics data in the high solid waste content material database for ultra-high performance concrete, and use the discrete element method to build a three-dimensional meso-model of aggregate-fiber-cementitious system, to convert the micro-pore distribution into the initial gap range between aggregate, solid waste particles and fibers at the meso-level.

[0078] The specific process is as follows:

[0079] 1. Base constraints and data integration: Extract aggregate gradation (0.1-2mm quartz sand), fiber parameters (length 10-15mm, diameter 0.2mm, volume content 2%) and microscale output hydration product volume fraction (e.g. 60% C-S-H and 5% CH at 28d) and pore distribution (total porosity 10%, pore size 0.01-0.1pm) from UHPC database, and form a unified multidimensional constraint vector after Z-score standardization.

[0080] 2. Three-dimensional mesoscale model construction: Use discrete element method (DEM) to construct a three-dimensional mesoscale model of aggregate-fiber-cementitious system. DEM simulates the interaction between particles to generate a three-dimensional model reflecting the internal structure of actual concrete. The specific steps are as follows:

[0081] Particle generation: Generate aggregate particles of different sizes and shapes according to the particle size distribution and shape of the aggregate. For example, generate quartz sand particles with a particle size of 0.1-2mm.

[0082] Fiber distribution: Randomly distribute fibers according to their length, diameter and volume content. Ensure uniform distribution of fibers in the model to avoid aggregation. For example, use a random distribution algorithm to evenly distribute fibers in the model.

[0083] Cementitious material filling: Fill cementitious material according to the volume fraction of hydration products and pore distribution. Ensure that the cementitious material fills the pores between aggregate and fiber to form a continuous matrix. For example, fill C-S-H gel to ensure that its volume fraction is 60% and the porosity is 10%.

[0084] 3. Initial gap range determination:

[0085] In the constructed three-dimensional mesoscale model, the initial gap range between aggregate, fiber and cementitious material is determined by simulating their distribution. The specific steps are as follows:

[0086] Gap calculation: Calculate the initial gap between aggregate particles, fibers and aggregate, and fibers. For example, the average gap between aggregate particles is calculated to be 0.05mm, and the average gap between fibers and aggregate is calculated to be 0.03mm.

[0087] Pore distribution conversion: Convert the micro-pore distribution to the initial gap range at the mesoscale level. For example, convert the micro-porosity of 10% to the initial gap range of 0.01-0.1mm at the mesoscale level.

[0088] Step S2B2, based on the above three-dimensional mesoscale model, the gap size data between aggregate particles and solid waste particles and fibers are extracted by three-dimensional structure scanning technology, and the gap filling demand is calculated combined with the microscale hydration product volume fraction, to obtain the average size and distribution uniformity characteristic parameters of the inter-particle gap.

[0089] The specific process is as follows:

[0090] 1. Three-dimensional structure scanning technology application: based on the three-dimensional mesoscopic model constructed in step S2B1, use three-dimensional structure scanning technology (such as X-ray CT) to scan the model, and extract the gap size data between aggregate particles, solid waste particles and fibers. The specific steps are as follows:

[0091] Scan data acquisition: X-ray CT scanning is performed on the actual concrete sample to obtain detailed information about its internal structure. The scanning data includes the distribution of aggregate particles, fibers and cementitious materials, as well as the gap size between them.

[0092] Image processing: process the images obtained by scanning, and use image segmentation techniques (such as threshold segmentation, region growing, etc.) to distinguish different phases (aggregate, fiber, cementitious material and pore). For example, by threshold segmentation, aggregate particles, fibers and cementitious materials are marked out respectively, and their geometric features are extracted.

[0093] 2. Gap size data extraction: extract the gap size data between aggregate particles, solid waste particles and fibers from the processed images. The specific steps are as follows:

[0094] Gap identification: identify the gaps between aggregate particles, fibers and aggregate, and fibers and fibers. For example, by calculating the distance between the centers of adjacent particles minus the sum of the particle radii, the gap size between them is obtained.

[0095] Data statistics: statistics of the extracted gap size data, calculation of the average size, standard deviation, distribution range and other statistical parameters of the gap. For example, the average gap between aggregate particles is calculated to be 0.05mm, and the standard deviation is 0.01mm; the average gap between fibers and aggregate is 0.03mm, and the standard deviation is 0.005mm.

[0096] 3. Gap filling requirement calculation: combined with the volume fraction of micro-hydration product, calculate the gap filling requirement. The specific steps are as follows:

[0097] Volume fraction conversion: convert the volume fraction of micro-hydration product into the filling volume of cementitious material. For example, assuming that the volume fraction of hydration product is 60%, and the porosity is 10%, then the volume of cementitious material that needs to be filled is 60% of the total pore volume.

[0098] Filling requirement calculation: according to the extracted gap size data and the filling volume of cementitious material, calculate the gap filling requirement. For example, the total gap volume that needs to be filled is calculated to be 0.1cm 3 , combined with the density of cementitious material (such as 2.6g / cm 3), the mass of the cementitious material to be filled is 0.26 g / cm 3 .

[0099] 4. Feature parameter calculation: Calculate the average size of the inter-particle gap and the distribution uniformity feature parameter. The specific steps are as follows:

[0100] Average size calculation: Calculate the average size of all extracted gaps. For example, the average gap between aggregate particles is 0.05 mm, and the average gap between fibers and aggregates is 0.03 mm.

[0101] Distribution uniformity calculation: Calculate the distribution uniformity of the gap size, using standard deviation or coefficient of variation as the measurement index. For example, the standard deviation of the gap between aggregate particles is 0.01 mm, and the coefficient of variation is 20%; the standard deviation of the gap between fibers and aggregates is 0.005 mm, and the coefficient of variation is 16.7%.

[0102] Step S2B3, taking the volume fraction of micro-hydration products and pore distribution as independent variables, and the inter-particle gap feature parameters as intermediate variables, fitting the correlation equation between micro-hydration characteristics and meso-structure parameters through multiple regression analysis, and clarifying the positive correlation between the bulk density and the volume fraction of hydration products, and the negative correlation between the fiber spacing variation coefficient and the distribution uniformity of the inter-particle gap.

[0103] The specific process is as follows: 1. Data preparation: Collect the volume fraction of micro-hydration products, pore distribution data, and inter-particle gap characteristic parameters. These data come from the output results of steps S2B1 and S2B2, including the volume fraction of hydration products, porosity, inter-particle gap size, and inter-fiber and aggregate gap size. 2. Multiple regression analysis: Take the volume fraction of micro-hydration products and pore distribution as independent variables, and take the inter-particle gap characteristic parameters (such as average gap size, distribution uniformity) as intermediate variables to build a multiple regression model. Through statistical analysis software (such as SPSS, R, or Python's scikit-learn library), regression analysis is performed to fit the correlation equation between micro-hydration characteristics and meso-structure parameters. 3. Correlation analysis: In the process of regression analysis, the positive correlation between bulk density and volume fraction of hydration products, and the negative correlation between fiber spacing coefficient of variation and inter-particle gap distribution uniformity are analyzed. The correlation coefficient (such as Pearson correlation coefficient) is calculated to quantify these correlations. For example, the correlation coefficient between bulk density and volume fraction of hydration products is 0.85, indicating a significant positive correlation; the correlation coefficient between fiber spacing coefficient of variation and inter-particle gap distribution uniformity is -0.78, indicating a significant negative correlation. 4. Equation verification and optimization: Use an independent data set to verify the fitted correlation equation, evaluate the predictive ability and accuracy of the equation. According to the verification results, the model is optimized and adjusted, such as adding or deleting independent variables, adjusting model parameters, etc., to improve the fitting degree and predictive ability of the equation. 5. Output correlation equation: Finally, output the correlation equation between micro-hydration characteristics and meso-structure parameters, and clearly define the positive correlation between bulk density and volume fraction of hydration products, and the negative correlation between fiber spacing coefficient of variation and inter-particle gap distribution uniformity.

[0104] Step S2B4, based on the quantitative conversion relationship, the theoretical values of bulk density and fiber spacing coefficient of variation are calculated, and the meso-structure measured data of high solid waste system in the special material database of ultra-high performance concrete are combined for error calibration, and finally the verified bulk density and fiber spacing coefficient of variation are output as meso-structure parameters.

[0105] The specific process is as follows: 1. Theoretical value calculation: Use the correlation equation obtained in step S2B3 to calculate the theoretical values of bulk density and fiber spacing coefficient of variation with the volume fraction of micro-hydration products and pore distribution as input. For example, assume the correlation equation is: bulk density = 0.5 + 0.3 x volume fraction of hydration products - 0.1 x porosity. Fiber spacing coefficient of variation = 0.2 - 0.05 x inter-particle gap distribution uniformity. Substitute the specific numerical values for calculation. 2. Error calibration: Compare the calculated theoretical values with the measured meso-structure data in the special material database of ultra-high performance concrete, and calculate the error. For example, the measured bulk density is 2.4 g / cm 3, the theoretical calculation value is 2.35 g / cm 3 , the error is 0.05 g / cm 3 ; the measured fiber spacing coefficient of variation is 0.15, the theoretical calculation value is 0.16, and the error is 0.01. 3. Model optimization: adjust the parameters of the correlation equation according to the error to optimize the model. For example, by introducing a correction coefficient or adjusting the parameters of the regression model, the error between the theoretical value and the measured value is reduced, and the accuracy of the model is improved. 4. Verification and output: use the optimized model to recalculate and compare with another set of independent measured data for verification. If the error is within an acceptable range, output the verified bulk density and fiber spacing coefficient of variation as mesostructure parameters, and provide accurate input for subsequent macroscopic performance simulation.

[0106] The finite element analysis tool is used to construct the quantitative mapping relationship between mesostructure, process parameters and macroscopic performance, and the preset threshold values of macroscopic performance and low carbon index include:

[0107] Step S2C1, collect the mesostructure parameters output at the mesoscale, combine the process parameters and basic performance data under high solid waste content in the special material database of ultra-high performance concrete, and integrate to form the input parameter set of finite element analysis.

[0108] The specific process is as follows: 1. Data collection: collect mesostructure parameters such as bulk density, fiber spacing coefficient of variation, etc. output at the mesoscale. For example, the bulk density is 2.4 g / cm 3 , and the fiber spacing coefficient of variation is 0.15. 2. Database integration: combine the process parameters and basic performance data under high solid waste content in the special material database of ultra-high performance concrete, and integrate to form the input parameter set of finite element analysis. For example, the process parameters include mixing time, curing temperature, vibration frequency, etc., and the basic performance data include compressive strength, tensile strength, elastic modulus, etc.

[0109] Step S2C2, based on the above input parameter set, the finite element analysis tool is used to build a macroscopic performance correlation model, which converts mesostructure parameters into microstructure input items of the model, process parameters into environmental and operation input items, constructs mechanical performance sub-model, durability sub-model, and embeds the conversion relationship between process parameters and carbon emissions.

[0110] The specific process is as follows: 1. Model building and tool selection: Use finite element analysis tools (such as ABAQUS, ANSYS) to build a macroscopic performance correlation model. Convert mesostructure parameters (such as packing density, fiber spacing coefficient of variation) into model microstructure input items, and convert process parameters (such as mixing time, curing temperature, vibration frequency) into environmental and operational input items. The specific steps are as follows: 1.1 Tool selection: Select ABAQUS as the finite element analysis tool, which has powerful nonlinear analysis capabilities and a rich library of material models. 1.2 Model establishment: Establish a three-dimensional model of concrete materials in ABAQUS, including the distribution of aggregates, fibers, and cementitious materials. According to the mesostructure parameters, set the distribution characteristics of aggregates and fibers, such as packing density and fiber spacing coefficient of variation. 2. Microstructure input item setting: Convert mesostructure parameters into model microstructure input items. For example, set the packing density of aggregates to 2.4 g / cm 3 , and the fiber spacing coefficient of variation to 0.15. Define the distribution characteristics of aggregates and fibers in detail through the geometric module and material attribute module of the model, ensuring that the model can accurately reflect the microstructure of concrete. 3. Environmental and operational input item setting: Convert process parameters into model environmental and operational input items. For example, set the mixing time to 3 minutes, the curing temperature to 20°C, and the vibration frequency to 50 Hz. Define the stress and environmental conditions of concrete under different process conditions through the boundary condition and load module of the model, ensuring that the model can simulate the process impact in actual production. 4. Submodel construction: Construct mechanical performance submodels and durability submodels, and embed the conversion relationship between process parameters and carbon emissions. The specific steps are as follows: 4.1 Mechanical performance submodel: Define the mechanical performance parameters of concrete, such as compressive strength, tensile strength, and elastic modulus. Select appropriate constitutive relations from the material model library, such as the Drucker-Prager model, to simulate the mechanical behavior of concrete under different stress conditions. 4.2 Durability submodel: Define the durability performance parameters of concrete, such as chloride ion permeability and freeze-thaw resistance. Define damage models and diffusion models to simulate the durability behavior of concrete under different environmental conditions. 4.3 Carbon emission conversion relationship: Embed the conversion relationship between process parameters and carbon emissions, calculate the carbon emissions under different process parameters by defining the environmental impact factor of the model. For example, set the relationship between the carbon emissions per cubic meter of concrete and the mixing time, curing temperature, and vibration frequency as follows: carbon emissions = 50 - 10 x mixing time + 20 x curing temperature - 5 x vibration frequency. 5. Model verification and optimization: Use experimental data to verify and optimize the model. For example, verify the predicted compressive strength of the model through standard cube compressive strength testing, and verify the predicted chloride ion permeability of the model through rapid chloride permeability testing (RCPT). According to the verification results, adjust the model parameters to optimize the predictive ability of the model, ensuring that the model can accurately reflect the actual performance of concrete.

[0111] Step S2C3, carry out multi-group variable iterative simulation through the above-mentioned model, adjust mesostructure parameters and process parameters, record corresponding macroscopic performance and low-carbon indicators, and use response surface method to fit quantitative mapping equation of mesostructure parameters-process parameters-macroscopic performance-low-carbon indicators.

[0112] The specific process is as follows: 1, multi-group variable iterative simulation: using finite element model to carry out multi-group variable iterative simulation, adjusting mesostructure parameters (such as bulk density, fiber spacing coefficient of variation) and process parameters (such as stirring time, curing temperature, vibration frequency), recording corresponding macroscopic performance (such as compressive strength, tensile strength, elastic modulus) and low-carbon indicators (such as carbon emissions). For example:

[0113] Simulation 1: bulk density 2.4 g / cm 3 , fiber spacing coefficient of variation 0.15, stirring time 3 minutes, curing temperature 20°C, vibration frequency 50 Hz, compressive strength 220 MPa, carbon emissions 45 kgCO2 / m 3 .

[0114] Simulation 2: bulk density 2.5 g / cm 3 , fiber spacing coefficient of variation 0.10, stirring time 4 minutes, curing temperature 22°C, vibration frequency 55 Hz, compressive strength 230 MPa, carbon emissions 40 kgCO2 / m 3 .

[0115] 2, response surface method fitting: using response surface method to fit quantitative mapping equation between mesostructure parameters, process parameters and macroscopic performance, low-carbon indicators. The specific steps are as follows: 2.1, data arrangement: arranging multi-group simulation data to form data set. 2.2, model establishment: using statistical analysis software (such as Design-Expert, R or Python's scikit-learn library) to establish response surface model. 2.3, equation fitting: fitting to obtain quantitative mapping equation, for example: compressive strength = 200 + 50 × bulk density - 30 × fiber spacing coefficient of variation + 10 × stirring time - 5 × curing temperature + 8 × vibration frequency. Carbon emissions = 50 - 10 × stirring time + 20 × curing temperature - 5 × vibration frequency.

[0116] 3, equation verification: using independent data set to verify the fitted mapping equation, ensuring its prediction ability and accuracy. For example, using a group of simulation data not involved in fitting to verify the deviation between predicted value and actual value of the equation, ensuring that the deviation is within an acceptable range.

[0117] Step S2C4, with high solid waste, low carbon, and high durability as constraints, reference the high performance sample interval in the database, substitute into the mapping equation to determine the preset threshold of macro performance and low carbon index.

[0118] 1. Constraint setting: high solid waste, low carbon, and high durability as constraints, reference the high performance sample interval in the special material database of ultra-high performance concrete. 2. Mapping equation back calculation: substitute the mapping equation fitted in step S2C3 to determine the preset threshold of macro performance and low carbon index that meets the constraint conditions.

[0119] Nonlinear fitting and optimization of the mapping relationship and boundary by machine learning model includes:

[0120] Step SA00, collect the raw material-process-performance basic mapping relationship data output by multi-scale simulation, combine the historical formula, process, and corresponding performance data in the special material database of ultra-high performance concrete to form a training data set containing raw material ratio, process parameters, scale characteristic parameters, and target performance.

[0121] The specific process is as follows: 1. Data integration range: extract the “raw material-process-performance” basic mapping relationship from the multi-scale simulation results, including micro-hydration product volume fraction, mesoscopic packing density, fiber spacing variation coefficient, and macro compressive strength, chloride ion permeability, and carbon emission; simultaneously call the historical formula, mixing system, and measured performance data in the special material database of ultra-high performance concrete to form an original data pool covering multiple solid waste contents, multiple ages, and multiple environmental conditions.

[0122] 2. Feature engineering and cleaning:

[0123] Conduct consistency check on the original data pool: eliminate samples missing key performance items; remove outliers using the 3σ criterion; Z-score standardization for variables with large dimensional differences such as raw material chemical composition, specific surface area, process time, and temperature; use random forest to evaluate feature importance, retain variables with importance >0.01, and finally obtain a training data set containing raw material ratio, process parameters, scale characteristic parameters, and target performance (strength + durability + carbon emission), with a sample size of about 1.2×10 4 pieces, which can be directly used for subsequent machine learning modeling.

[0124] Step SB00, based on the nonlinear correlation characteristics of the training data set, use gradient boosting tree algorithm to build a machine learning model, with raw material ratio and process parameters as input, macro performance and low carbon index as output, and parameter boundaries defined by multi-scale simulation as input constraints.

[0125] Specifically, algorithm selection and model construction: based on the complex nonlinear characteristics of the training data set, the gradient boosting decision tree algorithm is selected to construct the machine learning model. The input features of the model are clear: 1, raw material ratio characteristics (such as the mass fraction of cement, slag powder, fly ash, and fiber, and aggregate gradation parameters); 2, key process parameter characteristics (such as water-binder ratio, mixing time, and curing system number).

[0126] The output target of the model is clear: 1, macro performance indicators (such as 28-day compressive strength and chloride ion diffusion coefficient); 2, low-carbon indicators (such as carbon emission equivalent per cubic meter of material).

[0127] Before model training, the physical boundaries of each parameter (such as solid waste content ≥60% and water-binder ratio ≤0.22) defined by multi-scale simulation in step S200 are coded as hard constraints and directly applied to the boundary check in the training data screening and model prediction stages. Through grid search and cross-validation, the optimal hyperparameter set of the model is determined as: learning rate 0.05, maximum tree depth 8, and sub-sample ratio 0.8. After testing set verification, the prediction determination coefficient (R 2 ) of the model on the core performance indicators is greater than 0.90.

[0128] Step SC00, divide the above training data set into training set and validation set according to the preset proportion, input the above constructed machine learning model for iterative training, monitor the prediction error in real time through the validation set, optimize the model hyperparameters using grid search method, and until the error meets the preset accuracy requirement.

[0129] The specific process is as follows: 1. Data set division: divide the training data set into training set and validation set according to the preset proportion. Usually, 60% of the data is used as the training set, 20% of the data is used as the validation set, and the remaining 20% of the data is used as the test set. This division method can ensure that the model has enough data for learning features, and has an independent validation set for hyperparameter adjustment and prevention of overfitting. 2. Model training: input the training set into the constructed machine learning model for iterative training. During the training process, the model continuously adjusts the parameters through optimization algorithms (such as gradient descent) to minimize the prediction error. For example, when training a gradient boosting tree (GBDT) model, the model will gradually improve the prediction accuracy through iterative optimization. 3. Validation set monitoring: during the training process, the validation set is used to monitor the prediction error of the model in real time. This helps to discover whether the model is overfitting or underfitting in a timely manner. For example, if the error of the validation set starts to increase while the error of the training set is still decreasing, it may indicate that the model is starting to overfit. 4. Grid search method to optimize hyperparameters: use the grid search method to search for the best combination of hyperparameters in the pre-defined hyperparameter space. For example, for a GBDT model, the learning rate, tree depth, subsampling ratio, and other hyperparameters can be adjusted. By evaluating the performance of different hyperparameter combinations on the validation set, select the combination that minimizes the prediction error as the optimal hyperparameters. 5. Model optimization and validation: retrain the model using the optimized hyperparameters and perform the final evaluation on the test set to ensure the generalization ability and prediction accuracy of the model. For example, by calculating the mean square error (MSE), mean absolute error (MAE), and other indicators on the test set, the performance of the model is verified. If the model's performance on the test set meets the preset accuracy requirements, the model training is considered complete and can be used for practical applications.

[0130] Step SD00, based on the trained machine learning model, import high solid waste content, low carbon emission, high durability core target value, under the constraint of parameter boundary, use genetic algorithm to drive the model to carry out multi-objective optimization, output a group of candidate schemes of raw material ratio and key process parameters.

[0131] The specific process is as follows: 1. Target value setting and boundary constraint: import high solid waste content, low carbon emission, high durability as core target value into the trained machine learning model. For example, set the solid waste content target to 60%, the carbon emission target to 50 kgCO2 / m 3, the compressive strength target is 220 MPa, and the chloride ion permeability target is 1000C. At the same time, the parameter boundaries drawn by multi-scale simulation are used as input constraints to ensure that the model does not exceed these boundaries during optimization. 2. Genetic algorithm initialization: Initialize the genetic algorithm (GA), set the population size, crossover probability, mutation probability, etc. For example, the population size is set to 100, the crossover probability is 0.8, and the mutation probability is 0.01. Each individual in the population represents a combination of raw material ratios and process parameters. 3. Definition of fitness function: Define the fitness function to evaluate the pros and cons of each individual. The fitness function combines macroscopic performance and low-carbon indicators, for example: fitness=w1 x compressive strength+w2 x chloride ion permeability+w3 x carbon emissions. Where w1, w2, w3 are weight coefficients, set according to actual needs. 4. Genetic algorithm iterative optimization: Use genetic algorithm to drive the model to carry out multi-objective optimization. Through selection, crossover, mutation and other operations, constantly optimize the population and improve the fitness value. For example, in each iteration, the individual with the highest fitness is selected for crossover and mutation to generate a new population. Repeat the iteration until the termination condition is met (such as the maximum number of iterations or the convergence of fitness). 5. Output of candidate schemes: Output a preset number of raw material ratios and key process parameter candidate schemes. For example, output 1 set of candidate schemes, each of which includes raw material ratios (such as slag powder, fly ash, cement, aggregate, and fiber volume fraction) and process parameters (such as mixing time, curing temperature, and vibration frequency). These schemes meet the high solid waste, low carbon, and high durability targets while having high fitness values.

[0132] Step SE00, for the above output candidate schemes, filter the parameter combinations whose performance prediction value and target value deviation is within the preset deviation threshold, verify with historical measured data in the ultra-high performance concrete special material database, retain the schemes within the preset error range, and determine the comprehensive optimal parameter combination.

[0133] The specific process is as follows: 1. Screening candidate schemes: For the candidate schemes output in step SD00, the deviation of the performance prediction value of each scheme from the target value is calculated. For example, for a compressive strength target value of 220 MPa, if the prediction value of a certain scheme is 210 MPa, the deviation is -10 MPa. Set the deviation threshold, such as ±5%, and select the schemes with deviations within this range. 2. Historical data verification: Combine the historical measured data in the ultra-high performance concrete special material database to verify the selected schemes. Compare the actual performance under similar mix proportions and process conditions in the historical data to verify the reliability of the prediction value. 3. Retain effective schemes: retain schemes with errors within the preset range to form a preliminary optimal parameter combination list. For example, if the compressive strength, carbon emissions and other key indicators of a certain scheme are within the threshold, the scheme is retained. 4. Comprehensive evaluation to determine the optimal combination: comprehensive evaluation of the retained schemes, considering multi-objective balance, to determine the final optimal parameter combination. For example, select the scheme with the lowest carbon emissions while meeting all performance requirements as the optimal solution.

[0134] The digital adjustment method of the high-solid-waste low-carbon high-durability concrete connecting material further comprises:

[0135] Step S600, during the process of stirring, molding and curing, sensors arranged in advance at key nodes of stirring, molding and curing are used to collect key data and upload them to the production control system synchronously.

[0136] Specifically, sensors are arranged at key production nodes such as stirring, molding and curing to collect real-time process data (such as temperature, humidity, pressure) and material performance data (such as early strength) and upload the data to the production control system synchronously. For example, temperature and humidity sensors are installed in the curing room to monitor and upload curing environment data in real time.

[0137] Step S700, compare the uploaded real-time data with the preset digital twin model dynamically, if the parameter deviation exceeds the preset range, automatically adjust the corresponding process parameters, and issue a warning when the deviation exceeds the preset threshold to realize self-adaptive regulation and control of the production process.

[0138] The specific process can refer to steps S710 to S750, which will not be repeated here.

[0139] Step S800, after curing, the macro mechanical properties, durability and low-carbon indicators of the finished product are detected, and the detection data are compared with the preset target value for compliance determination.

[0140] Specifically, after curing, the finished product is tested for macro-mechanical properties, durability, and low-carbon indicators. The test items include compressive strength, resistance to chloride ion penetration, carbon emissions, etc. The test data is compared with the preset target value to determine compliance. For example, if the target value for compressive strength is 220 MPa and the test value is 215 MPa, the deviation is within the allowed range (±5 MPa), and it is determined to pass.

[0141] Step S900, if the determination fails, immediately identify the isolated unqualified products, trace the causes combined with production data and handle them specifically, record the whole process data and feedback to the database; if the determination passes, give the outgoing connector a unique digital identity card, associate the raw material batch, proportioning, process and test data, and form a traceability file.

[0142] 1. Unqualified product processing:

[0143] Immediate identification and isolation: if the finished product test result does not pass the check, the system immediately identifies the unqualified product and isolates it. For example, if the compressive strength test value is lower than the target value, the system automatically marks the batch of products as unqualified and moves it to the isolation area.

[0144] Traceability and processing: trace the causes combined with production data, such as deviation of certain process parameters or quality problems of raw materials. After targeted processing, record the whole process data and feedback to the database for optimization of subsequent production.

[0145] 2. Qualified product traceability file:

[0146] Unique digital identity card: if the check passes, give the outgoing connector a unique digital identity card, associate the raw material batch, proportioning, process and test data.

[0147] Form a traceability file: integrate the whole process data from raw materials to finished products to form a complete traceability file, ensure product quality traceability, and provide consistent verification guarantee for users.

[0148] It needs to be further pointed out that the model system constructed by the present application is an intelligent system with continuous evolution capability. From the new data obtained from high-throughput experimental verification (step S400), to the process data recorded by the production adaptive control (step S700), to the long-term performance data fed back by the connection service monitoring (as described in the specification subsequent steps), they are all fed back to the super high performance concrete special material database. These new data samples from the real world are periodically used to calibrate the parameters of the multi-scale physical and chemical simulation model, and incrementally train the machine learning prediction and optimization model. Through this "design-production-service" full-chain data closed-loop feedback and model iterative optimization mechanism, the system can continuously adapt to real changes such as raw material batch fluctuations and process equipment aging, making the optimization results of material ratio and process parameters more and more accurate and robust, so as to realize the continuous self-improvement and performance improvement of the preparation method.

[0149] The uploaded real-time data is dynamically compared with the preset digital twin model. If the parameter deviation exceeds the preset range, the corresponding process parameters are automatically fine-tuned, including:

[0150] Step S710, extract the material temperature and viscosity in the mixing stage, the vibration frequency and forming pressure in the forming stage, and the environment temperature and humidity and steam flow in the curing stage from the uploaded real-time data, and compare them with the corresponding standard parameter range in the digital twin model to identify single disturbance or multiple disturbance concurrent scenarios.

[0151] The digital twin model is a dynamic mirror of the physical production line in the information space, and its core consists of three parts: 1. "State-performance" mapping sub-model, which is a fast prediction model of product performance from the final optimal process parameters determined in step S400 through experimental verification; 2. Device response and control sub-model, which describes the response characteristics of key devices such as mixers, forming machines, and curing kilns to control instructions; 3. Multi-objective control decision sub-model, which embeds adjustment strategies based on historical data and expert rules (such as PID control logic, priority rules). The model synchronizes data with production line sensors and control systems in real time through industrial protocols such as OPCUA. When performing dynamic comparison, the model not only judges whether the parameters are out of limits, but also simulates the deviation of the final product performance if no adjustment or different adjustment strategies are adopted under the current disturbance, thereby generating optimal control instructions that can simultaneously consider quality stability and lowest energy consumption.

[0152] 1. Real-time data extraction: Extract key process parameters from uploaded real-time data, including: 1.1, mixing stage: material temperature and viscosity; 1.2, forming stage: vibration frequency and forming pressure; 1.3, curing stage: environment temperature and humidity and steam flow.

[0153] 2. Data comparison: Compare the extracted real-time data with the corresponding standard parameter ranges in each stage of the digital twin model to identify single disturbance or multi-disturbance concurrent scenarios: 2.1. Single disturbance identification: For example, if only the material temperature in the real-time data exceeds the standard range (standard range: 20-25°C, real-time temperature: 26°C), it is identified as a single disturbance. 2.2. Multi-disturbance concurrent identification: For example, if the material temperature in the real-time data is 26°C, and the molding pressure exceeds the standard range (standard range: 50-60MPa, real-time pressure: 65MPa), it is identified as a multi-disturbance concurrent scenario.

[0154] 3. Deviation identification: Determine whether the real-time data exceeds the allowed range by setting a deviation threshold. For example: 3.1. Material temperature: standard range 20-25°C, deviation threshold ±2°C. If the real-time temperature is 26°C, it exceeds the deviation threshold. 3.2. Molding pressure: standard range 50-60MPa, deviation threshold ±5MPa. If the real-time pressure is 65MPa, it exceeds the deviation threshold.

[0155] 4. Scenario identification: According to the comparison results, identify the disturbance type in the current production process to provide a basis for subsequent hierarchical control decisions. For example, if both the material temperature and the molding pressure exceed the standard range, the system identifies it as a multi-disturbance concurrent scenario and needs to be evaluated and handled comprehensively.

[0156] Step S720, execute hierarchical control decisions based on the identification results: For single disturbance, directly call historical process-performance correlation data and model sensitivity analysis results to lock core process parameters; for multi-disturbance concurrency, use the analytic hierarchy process to construct a multi-scale priority evaluation model, use the microstructure parameters, mesostructure parameters and macro performance parameters output by the aforementioned multi-scale calculation simulation as performance impact criteria, combine historical production data in the ultra-high performance concrete special material database to quantify the influence of each disturbance on the multi-scale performance criteria, calculate the comprehensive influence weight of each disturbance parameter and determine the adjustment priority order.

[0157] 1. Single disturbance handling:

[0158] Call historical data: Directly call historical process-performance correlation data to find which process parameter adjustments have effectively solved the problem under similar single disturbance conditions.

[0159] Model sensitivity analysis: Refer to the model sensitivity analysis results to determine the core process parameters that have the most impact on performance. For example, if the material temperature exceeds the standard range, historical data shows that adjusting the mixing time or cooling system parameters can effectively control the temperature, and the model shows that temperature has a significant impact on early strength, so the mixing time and cooling system parameters are locked as the core adjustment objects.

[0160] 2. Multi-disturbance concurrent processing:

[0161] Building evaluation model: Adopting AHP to build a multi-scale priority evaluation model. Microstructure parameters (such as hydration product density), mesostructure parameters (such as aggregate distribution uniformity), and macro-performance parameters (such as compressive strength) are used as performance impact criteria.

[0162] Quantifying the degree of influence: Based on historical production data in the ultra-high performance concrete special material database, the influence of each disturbance (such as material temperature anomaly, molding pressure anomaly) on the above performance criteria is quantified.

[0163] Calculate the weight and order: Calculate the comprehensive influence weight of each disturbance parameter through the AHP model to determine the adjustment priority order. For example, the calculation shows that the influence weight of material temperature anomaly on macro compressive strength is 0.4, the influence weight of molding pressure anomaly is 0.3, and the influence weight of environmental humidity anomaly is 0.3. According to the weight order, the priority is to handle material temperature anomaly, followed by molding pressure anomaly, and finally environmental humidity anomaly.

[0164] Step S730, according to the output result of hierarchical regulation decision, make parameter adjustment scheme: based on the priority order sequence, allocate adjustment action, adjust high priority parameters first, and then adjust secondary parameters.

[0165] The process is described as follows:

[0166] 1. According to the adjustment priority order determined in step S720, develop a detailed parameter adjustment scheme. For example, if the priority order is material temperature, molding pressure, and environmental humidity, the adjustment scheme is as follows:

[0167] First step: Adjust the material temperature. Adjust the cooling system or heating device of the mixer to control the temperature within the standard range (such as 20-25°C).

[0168] Second step: Adjust the molding pressure. Adjust the pressure control system of the molding equipment to adjust the pressure to the standard range (such as 50-60 MPa).

[0169] Third step: Adjust the environmental humidity. Adjust the humidification equipment of the curing room to control the humidity within the standard range (such as 90-95%).

[0170] 2. Sequence allocation of adjustment action: sequence allocation of adjustment action according to the priority order to ensure that high priority parameters are adjusted first and secondary parameters are adjusted subsequently. For example:

[0171] Priority 1: Material temperature adjustment, performed by the mixer operator.

[0172] Priority 2: Molding pressure adjustment, performed by the molding equipment operator.

[0173] Priority 3: Environmental humidity adjustment, performed by maintenance room operators.

[0174] 3. Specific implementation of adjustment actions:

[0175] Material temperature adjustment: If the real-time temperature is 26°C, which exceeds the standard range of 20-25°C, gradually reduce the temperature to the standard range by increasing the cooling system power or reducing the heating device power.

[0176] Molding pressure adjustment: If the real-time pressure is 65MPa, which exceeds the standard range of 50-60MPa, gradually reduce the pressure to the standard range by reducing the output of the pressure controller.

[0177] Environmental humidity adjustment: If the real-time humidity is 85%, which is lower than the standard range of 90-95%, gradually increase the humidity to the standard range by increasing the operating power of the humidification equipment.

[0178] 4. Record the adjustment plan: Record the developed parameter adjustment plan in the production control system, including the adjusted parameters, adjustment sequence, adjustment target value, etc. For example, the recorded adjustment plan is: "Adjust the material temperature to 20-25°C, adjust the molding pressure to 50-60MPa, and adjust the environmental humidity to 90-95%."

[0179] Step S740, calculate the fine-tuning amplitude of each core process parameter through the parameter response equation built-in the digital twin model, combined with the deviation degree and priority order; convert the calculated fine-tuning amplitude into device control instructions and issue them to the corresponding process equipment for execution.

[0180] Fine-tuning amplitude calculation: Calculate the specific fine-tuning amplitude of each core process parameter through the parameter response equation built-in the digital twin model, combined with the deviation degree and priority order. For example, if the real-time value of the material temperature is 26°C, which exceeds the standard range (20-25°C), the model calculates that the cooling system needs to increase its power by 15% to reduce the temperature; if the molding pressure is 65MPa, which exceeds the standard range (50-60MPa), the model calculates that the output of the pressure controller needs to be reduced by 20% to reduce the pressure.

[0181] Device control instruction issuing and execution: the fine-tuning amplitude calculated is converted into specific device control instructions and issued to the corresponding process equipment for execution. For example, the system sends instructions to the cooling system of the mixer to increase the cooling water flow or increase the cooling fan speed; sends instructions to the pressure control system of the molding equipment to adjust the output pressure of the hydraulic pump; sends instructions to the humidifying equipment of the curing chamber to increase the operating power or spray frequency of the humidifier. After receiving the instructions, the equipment performs the adjustment action, and the system collects the adjusted process data in real time and feeds back to the production control system for comparison again to ensure that the parameters return to the preset range. If the deviation is still not eliminated, the system repeats the adjustment process until the parameters are stable within the standard range.

[0182] Step S750, the system collects the adjusted process data in real time and compares it with the model again. If the deviation returns to the preset range, the process data is recorded. If it still exceeds the range, the adjustment process is repeated.

[0183] After the conformity of the digital intelligent adjustment preparation method of the high-solid-waste low-carbon high-durability concrete connecting material is determined to pass, it further comprises:

[0184] Step 1, for the connecting pieces determined to be qualified, before leaving the factory, the sensing elements are precisely embedded in combination with the distribution of the key stress positions of the structure, to ensure close adhesion and cooperative deformation with the concrete matrix. Specifically, before leaving the factory, the fiber grating or piezoelectric ceramic sensing elements are precisely embedded in the key stress positions of the structure (such as the maximum bending moment section and the shear force concentration area), the positioning mold is used to ensure that the elements are cooperatively bound with the steel reinforcement and synchronously poured with the concrete, and after vibration, the sensing elements are closely adhered to the matrix and cooperatively deformed, avoiding air pockets and interface slip.

[0185] Step 2, calibrate and debug the monitoring accuracy of the sensing elements, associate and bind them with the unique digital identity card of the connecting piece, and input them into the special material database for ultra-high performance concrete, to form a production-detection-sensing integrated traceability file. Specifically: after calibrating and debugging the monitoring accuracy of the sensing elements to ±2με / ±0.1°C, associate and bind them with the unique digital identity card of the connecting piece through RFID or two-dimensional code, and input them into the UHPC database together with the production batch, process parameters, and performance detection data, to form a production-detection-sensing integrated traceability file.

[0186] Step 3, establish a real-time monitoring system for the service state of the connecting piece through network access. Specifically, establish a real-time monitoring system for the service state of the connecting piece through NB-IoT or 5G network access, to realize encrypted data transmission and remote access.

[0187] Step 4, during the service period of the connecting piece, the service state data of stress, strain, temperature and vibration frequency are collected by the monitoring system at a preset frequency and uploaded to the intelligent operation and maintenance platform in real time. Specifically, the stress, strain, temperature and vibration frequency data are automatically collected at a preset frequency (such as 1 time / h) during the service period and uploaded to the intelligent operation and maintenance platform in real time.

[0188] Step 5, the intelligent operation and maintenance platform combines the integrated traceability file to analyze the collected data in real time, judges the health state of the connecting piece and realizes abnormal classification early warning. Specifically, the intelligent operation and maintenance platform combines the integrated traceability file and the preset health threshold to analyze the collected stress, strain, temperature and vibration data in real time, judges the health state of the connecting piece through edge computing and triggers green-yellow-red three-level abnormal early warning.

[0189] Step 6, generate a targeted preventive maintenance scheme based on the analysis result. Specifically, a targeted preventive maintenance scheme is automatically generated based on the warning level, the maintenance cycle, the reinforcement measures and the spare parts list are specified, and the disposal is ensured to be completed before the abnormality worsens.

[0190] Step 7, return the service data to the ultra-high performance concrete special material database, iteratively calibrate the multi-scale calculation simulation model and the machine learning model, and continuously optimize the raw material ratio and the key process parameters. Specifically, the service data is returned to the UHPC database after desensitization, the micro / meso / macro multi-scale calculation simulation model and the machine learning model are iteratively calibrated by using the transfer learning strategy, and the raw material ratio and the key process parameters are continuously optimized.

[0191] The embodiments of the specific implementation are the preferred embodiments of the present application, and do not limit the protection scope of the present application, so that: any equivalent changes made according to the structure, shape, principle of the present application should be covered within the protection scope of the present application.

Claims

1. A digital and intelligent control preparation method of high solid waste low carbon high durability concrete connecting material, characterized in that, The application relates to a method for preparing high-solid-waste low-carbon high-durability concrete connecting materials. The method comprises the following steps: acquiring and integrating basic physicochemical parameters of raw materials, historical process parameters and performance data to construct a special material database for ultra-high performance concrete; based on the special material database for ultra-high performance concrete, a basic mapping relationship among raw materials, process and performance is constructed through multi-scale calculation simulation and parameter boundaries are demarcated, and then a machine learning model is used to nonlinearly fit and optimize the mapping relationship and the boundaries, so that raw material ratios and key process parameters meeting preset high solid waste content, preset low carbon emission and preset high durability targets are screened out; material pretreatment parameters are determined according to the screening results, industrial solid wastes are activated, fibers are surface modified, and aggregates are combined according to gradation; based on the pretreated raw materials, high-throughput experiments are carried out in parallel by using automatic equipment according to the key process parameters and a preset experiment scheme, process and performance data are collected by using a preset monitoring device, and the data are fed back to the special material database for ultra-high performance concrete, so that the machine learning model is iteratively calibrated with the new samples until the deviation between the predicted value and the measured value of the model reaches a preset deviation threshold, and thus the optimal process parameters suitable for industrial production are determined; 2. The digital control preparation method of a high-solid-waste low-carbon high-durability concrete connecting material according to claim 1, characterized in that, the optimal process parameters are introduced into a production control system, and mixing, molding and curing are carried out according to a preset production process, so that the high-solid-waste low-carbon high-durability concrete connecting materials are prepared. The method comprises the following steps: microscale simulation: a preset thermodynamic calculation software is used to predict stable hydration product phases of a multi-component cementing system by setting different temperature conditions based on the principle of Gibbs free energy minimization, so as to avoid calcium hydroxide crystals, and type, content and microstructure parameters of the hydration products are outputted; mesoscale simulation: based on the hydration product data outputted by the microscale simulation, the hydration products are taken as ultra-fine powders, and combined with the basic physical and chemical parameters of aggregates and powders, a particle packing model is used to calculate the packing density of the system according to the particle size distribution, so that the gradation design is simulated and optimized in a digital way, the target porosity is virtually calculated, the optimal packing scheme is determined, and corresponding mesoscale structure parameters are outputted; macro-scale simulation: based on the optimal packing scheme and the porosity parameters of the mesoscale simulation, a finite element model reflecting the mechanical properties of the matrix is established, finite element analysis technology is used to simulate the distribution state, orientation characteristics and stress transmission path of the fibers in the matrix, the enhancement effect of the fibers on the macroscopic mechanical properties of the material is quantified, the type and content of the fibers are optimized, and simulation data related to the macroscopic performance are outputted; mapping relationship construction and parameter boundary demarcation: the simulation results of the microscale, mesoscale and macro-scale are integrated, the logical correlation among the material characteristics, process parameters and performance at different scales is systematically connected, historical data accumulated in the special material database for ultra-high performance concrete are combined, and a basic mapping relationship from the raw materials to the performance is constructed; meanwhile, the multi-scale parameter boundaries matched with the design targets are demarcated and designed according to the parameter variation ranges covered by the simulation at different scales.

3. The digital control preparation method of a high-solid-waste low-carbon high-durability concrete connecting material according to claim 1, characterized in that, The basic mapping relationship between raw materials, process and performance is constructed through multi-scale calculation simulation, and the parameter boundary is drawn, including: Microscale simulation: Based on the basic physicochemical parameters of raw materials in the special material database of ultra-high performance concrete, the hydration reaction process of cementitious system is simulated by using the hydration reaction kinetics simulation tool, the type of hydration product is predicted, and the density of hydrated calcium silicate gel is quantified. The amount of calcium hydroxide crystal generated is controlled not to exceed the preset upper limit to ensure high durability, and the key parameters of hydration product volume fraction and pore distribution are output; Mesoscale simulation: Based on the microscale output of hydration product volume fraction and pore distribution, combined with the aggregate characteristic data under high solid waste content in the special material database of ultra-high performance concrete, the aggregate gradation is optimized and the fiber distribution state is evaluated through the particle packing model and volume filling model, the inter-particle gap characteristics are quantified, the quantitative conversion relationship between micro-hydration characteristics and meso-structure is established, and the packing density and fiber spacing coefficient of variation are output as meso-structure parameters; Macro-scale simulation: Based on the meso-structure parameters, using finite element analysis tools, based on the performance mapping equation, the quantitative mapping relationship between meso-structure, process parameters and macro-performance is constructed, so as to correlate the macro-mechanical performance, durability performance and low carbon emission index under high solid waste content, and output the preset threshold of macro-performance and low carbon index; Mapping relationship construction and parameter boundary demarcation: The quantitative conversion relationship and mapping relationship formed by micro-meso-macro three-level simulation are integrated, combined with the historical process-performance correlation data in the special material database of ultra-high performance concrete, the basic mapping relationship between raw materials, process and performance is constructed, and the parameter boundary of each scale suitable for high solid waste, low carbon and high durability target is comprehensively demarcated according to the parameter range output by each scale simulation.

4. The digital control preparation method of a high-solid-waste low-carbon high-durability concrete connecting material according to claim 3, characterized in that, Predicting the type of hydration product and quantifying the density of hydrated calcium silicate gel, controlling the amount of calcium hydroxide crystal generated not to exceed the preset upper limit to ensure high durability, outputting the key parameters of hydration product volume fraction and pore distribution, including: The active ingredient content, specific surface area and mineral composition data of industrial solid waste are retrieved from the special material database of ultra-high performance concrete, and the initial parameter matrix of multi-component hydration reaction is established combined with the reaction activity difference between industrial solid waste and cementitious materials; Based on the initial parameter matrix of hydration reaction, the hydration reaction model of high solid waste system is constructed by using the hydration reaction kinetics simulation tool, the reaction rate equation is modified by introducing the activity excitation coefficient of solid waste, the hydration path at different ages is simulated, the type of hydration product containing hydrated calcium silicate and calcium hydroxide is predicted by reaction progress calculation, and the generation rate curve of each product and the characteristic parameters of hydrated calcium silicate stoichiometry and microstructure characteristics are output; Based on the above characteristic parameters, the apparent density of hydrated calcium silicate gel is calculated by using the density function theory, and the density value with preset precision is obtained by combining with the measured data of the same type of gel in the special material database of ultra-high performance concrete. With the preset high durability target as a constraint, the generation amount of calcium hydroxide is calibrated to a preset upper limit through the correlation data between calcium hydroxide content and durability in the special material database for ultra-high performance concrete, the generation amount is monitored in real time in the simulation, and when the generation amount exceeds the upper limit, the solid waste content or the activity excitation coefficient is adjusted reversely and the hydration reaction model is updated synchronically; Combined with the corrected gel density and the generation rate curve of each product, the volume proportion of hydration products at different ages is calculated, the microstructure evolution caused by hydration reaction is converted into pore distribution parameters through the pore network model, and a pore characteristic matrix adapted to mesoscopic simulation is formed as the bottom constraint of mesoscopic simulation.

5. The digital control preparation method of a high-solid-waste low-carbon high-durability concrete connecting material according to claim 3, characterized in that, Quantify the inter-particle gap characteristics, establish a quantitative conversion relationship between micro-hydration characteristics and meso-structure, and output the packing density and fiber spacing variation coefficient as meso-structure parameters including: Based on the volume proportion of hydration products and pore distribution output at the micro-scale, combined with the aggregate characteristic data under high solid waste content in the special material database for ultra-high performance concrete, a three-dimensional meso-model of aggregate-fiber-cementing system is constructed by using the discrete element method, and the micro-pore distribution is converted into the initial gap range between aggregate, solid waste particles and fibers at the meso-level; Based on the above three-dimensional meso-model, the gap size data between aggregate particles, solid waste particles and fibers are extracted by three-dimensional structure scanning technology, the inter-particle gap filling requirement is calculated combined with the micro-hydration product volume proportion, and the average size and distribution uniformity characteristic parameters of the inter-particle gap are obtained; Taking the micro-hydration product volume proportion and pore distribution as independent variables, and the inter-particle gap characteristic parameters as intermediate variables, the correlation equation between micro-hydration characteristics and meso-structure parameters is fitted through multiple regression analysis, and the positive correlation between packing density and hydration product volume proportion, and the negative correlation between fiber spacing variation coefficient and inter-particle gap distribution uniformity are determined. Based on the above quantitative conversion relationship, the theoretical values of packing density and fiber spacing variation coefficient are calculated, and error calibration is performed combined with the meso-structure measured data of high solid waste system in the special material database for ultra-high performance concrete, and finally the verified packing density and fiber spacing variation coefficient are output as the meso-structure parameters.

6. The digital control preparation method of a high-solid-waste low-carbon high-durability concrete connecting material according to claim 2 or 3, characterized in that, Nonlinear fitting and optimization of the mapping relationship and boundary by machine learning model including: Collect the raw material-process-performance basic mapping relationship data output by multi-scale simulation, combine the historical formula, process and corresponding performance data in the special material database for ultra-high performance concrete, and form a training data set containing raw material ratio, process parameters, scale characteristic parameters and target performance; Based on the nonlinear correlation characteristics of the training data set, a machine learning model is constructed by using gradient boosting tree algorithm, taking raw material ratio and process parameters as input, and macro-performance and low-carbon index as output, and the parameter boundary defined by multi-scale simulation as input constraint; The above training data set is divided into training set and validation set according to a preset proportion, and is input into the machine learning model constructed above for iterative training, the prediction error is monitored in real time through the validation set, the model hyperparameters are optimized by grid search method, and the error is calibrated until it meets the preset accuracy requirement; Based on the trained machine learning model, import the core target values of high solid waste content, low carbon emission and high durability, and use genetic algorithm to drive the model for multi-objective optimization under parameter boundary constraints, output a set of candidate schemes of raw material ratio and key process parameters; For the above output candidate schemes, filter the parameter combinations whose performance prediction value deviation from the target value is within the preset deviation threshold, and verify them with historical measured data in the ultra-high performance concrete special material database, retain the schemes with error within the preset range, and determine the comprehensive optimal parameter combination.

7. The method of claim 1, wherein the method is characterized by: It also includes the following steps: During the execution of stirring, molding and curing, sensors pre-arranged at key nodes of stirring, molding and curing are used to collect key data and upload them to the production control system synchronously; The uploaded real-time data are dynamically compared with the preset digital twin model, if the parameter deviation exceeds the preset range, the corresponding process parameters are automatically fine-tuned, and a warning is issued immediately when the deviation exceeds the preset threshold, realizing self-adaptive regulation and control of the production process; After curing, the finished product is tested for macro-mechanical properties, durability and low-carbon indicators, and the test data are compared with the preset target values for compliance determination; If the determination fails, the unqualified product is immediately identified and isolated, the production data are traced to find the cause and processed accordingly, and the whole process data are recorded and fed back to the database; if the determination passes, a unique digital identity card is given to the finished connector, the raw material batch, ratio, process and test data are associated, and a traceability file is formed.

8. The digital control preparation method of a high-solid-waste low-carbon high-durability concrete connecting material according to claim 7, characterized in that, The uploaded real-time data are dynamically compared with the preset digital twin model, if the parameter deviation exceeds the preset range, the corresponding process parameters are automatically fine-tuned, and a warning is issued immediately when the deviation exceeds the preset threshold, realizing self-adaptive regulation and control of the production process; From the uploaded real-time data, the material temperature and viscosity in the stirring stage, the vibration frequency and molding pressure in the molding stage, and the environmental temperature and humidity and steam flow in the curing stage are extracted, compared with the standard parameter range of each stage in the digital twin model, and single disturbance or multiple disturbance concurrent scenarios are identified; Based on the identification results, hierarchical regulation and control decisions are made: for single disturbance, the historical process-performance correlation data and model sensitivity analysis results are directly called to lock the core process parameters; for multiple disturbance concurrency, a multi-scale priority evaluation model is constructed using the analytic hierarchy process, the microstructure parameters, mesostructure parameters and macro-performance parameters output by the aforementioned multi-scale calculation simulation are used as performance influence criteria, and the historical production data in the ultra-high performance concrete special material database are used to quantify the influence of each disturbance on the multi-scale performance criteria, calculate the comprehensive influence weight of each disturbance parameter and determine the adjustment priority order; According to the output results of hierarchical regulation and control decisions, parameter adjustment schemes are developed: based on the priority order sequence, adjustment actions are sequentially assigned, high-priority parameters are adjusted first, and then secondary parameters are adjusted; Through the parameter response equation built in the digital twin model, the fine-tuning amplitude of each core process parameter is calculated based on the deviation degree and priority order; the calculated fine-tuning amplitude is converted into equipment control instructions and sent to the corresponding process equipment for execution.

Images