A method for proportioning design of low-carbon concrete material

Abstract

This invention discloses a low-carbon concrete mix design method, relating to the field of intelligent mix design technology for low-carbon building materials. The method first constructs a dataset of multiple low-carbon concrete mix proportions; then, it constructs a multi-objective coupled prediction model based on attention mechanism weighted fusion, outputting the comprehensive performance parameters and carbon emission values ​​of concrete corresponding to any mix proportion; next, it constructs a multi-objective optimization objective function with low carbon, high performance, and low cost as the core objectives; it employs an improved sparrow search algorithm that integrates chaotic initialization, nonlinear convergence factors, and adaptive mutation for global optimization to obtain the optimal mix proportion; finally, it performs performance verification and experimental validation on the optimal low-carbon concrete mix proportion. This invention effectively improves prediction accuracy and optimization efficiency, reduces carbon emissions and increases solid waste utilization while ensuring the service performance of concrete, and is suitable for various low-carbon concrete engineering mix design projects.

Description

Technical Field

[0001] This invention relates to the field of intelligent proportioning technology for low-carbon building materials, and in particular to a method for designing the proportioning of low-carbon concrete materials. Background Technology

[0002] Concrete, as a core basic material in construction engineering, has a large carbon emission from the cement clinker calcination process during its production. To reduce concrete carbon emissions, the industry generally adopts a low-carbon improvement scheme that uses fly ash, mineral powder, and other solid waste admixtures to replace part of the cement, combined with recycled aggregates to replace natural aggregates, forming a multi-component composite low-carbon concrete system. However, at present, the design of low-carbon concrete mix proportions still mainly relies on traditional manual trial mixing and experience-based proportioning. This method is highly dependent on the professional experience of technical personnel, resulting in problems such as high frequency of trial mixing, long research and development cycle, serious material loss, and high cost. Moreover, it cannot adapt to complex nonlinear proportioning systems with multiple solid wastes and recycled aggregates, making it difficult to quickly screen out low-carbon proportioning schemes suitable for engineering scenarios, and the level of intelligence and precision is extremely low.

[0003] Current intelligent mix proportioning technologies suffer from two major technical shortcomings, making it difficult to achieve synergistic optimization of multiple indicators in low-carbon concrete. First, conventional machine learning prediction models are mostly single-objective fitting structures, only able to establish a mapping relationship between mix proportioning parameters and single mechanical properties of concrete. They cannot simultaneously couple and fit multiple dimensions of indicators such as mechanical properties, durability, economic cost, and carbon emissions, resulting in low fitting accuracy and poor generalization ability. They also cannot accurately quantify the coupling relationship between carbon reduction gains and performance losses from multiple solid waste admixtures, easily leading to the technical contradiction of "carbon reduction without quality assurance, or quality assurance without carbon reduction." Second, traditional intelligent optimization algorithms (genetic algorithms, particle swarm optimization, conventional sparrow algorithms) have rigid iterative mechanisms, often employing fixed-weight optimization strategies and random population initialization. In multi-constraint, multi-objective low-carbon mix proportioning solutions, they are prone to slow convergence, low iteration efficiency, and getting trapped in local optima, failing to achieve a globally optimal mix proportion solution.

[0004] Most existing publicly available optimization schemes are limited to a one-sided optimization logic of "single carbon reduction under the premise of meeting strength standards," ignoring key durability performance indicators such as concrete impermeability, freeze-thaw resistance, and drying shrinkage deformation. This leads to engineering hazards such as cracking, durability degradation, and shortened service life in the optimized low-carbon concrete. Furthermore, existing technologies lack a scenario-adaptive dynamic weight adjustment mechanism, failing to dynamically adjust the priority of each optimization objective according to engineering carbon reduction needs and performance requirements. This makes it difficult to adapt to different application scenarios such as civil buildings and municipal engineering, severely limiting its versatility and engineering practicality. In summary, there is an urgent need for a low-carbon concrete mix design method that integrates high-precision multi-objective machine learning fitting and adaptive intelligent optimization to overcome the many bottlenecks of existing technologies. Summary of the Invention

[0005] The purpose of this invention is to propose a low-carbon concrete mix design method, which aims to solve the technical pain points of existing low-carbon concrete mix design methods, such as low fitting accuracy, easy local optima in optimization, and inability to achieve coordinated balance among multiple objectives. By constructing a multi-objective coupled prediction model and improving the adaptive optimization algorithm, the invention achieves simultaneous optimal solutions for concrete carbon emissions, mechanical properties, durability, and economic performance, significantly improving the accuracy and efficiency of mix design, significantly reducing carbon emissions from concrete production, and ensuring the reliability of engineering applications.

[0006] To achieve the above objectives, this invention proposes a method for designing the mix proportions of low-carbon concrete, the specific steps of which are as follows: Step S1: Construct a multi-group mix proportion dataset for low-carbon concrete, collect the dosage parameters of cement, fly ash, mineral powder, recycled coarse aggregate, recycled fine aggregate, water, and water-reducing agent, as well as the measured data of concrete compressive strength, impermeability grade, drying shrinkage rate, carbon emissions, and material costs under the corresponding mix proportions, and complete data cleaning, normalization, and outlier removal. Step S2: Construct a multi-objective coupled prediction model based on attention mechanism weighted fusion, establish a nonlinear mapping relationship between mix proportion parameters and multi-dimensional performance and carbon emissions, and output the comprehensive performance parameters and carbon emission values ​​of concrete corresponding to any mix proportion. Step S3: With low carbon, high performance, and low cost as the core objectives, and combined with engineering specification constraints, construct a multi-objective optimization objective function, introduce a dynamic weight adaptive adjustment mechanism, and establish a multi-constraint optimization model for low carbon concrete mix proportions. Step S4: An improved adaptive sparrow search intelligent optimization algorithm is used to iteratively solve the multi-constraint optimization model. The optimization algorithm introduces a chaotic initialization population, a nonlinear convergence factor, and an adaptive mutation perturbation mechanism to quickly converge and obtain the globally optimal low-carbon concrete material mix ratio. Step S5: Perform performance verification and experimental validation on the optimal low-carbon concrete material mix ratio. If the preset engineering indicators and carbon reduction indicators are met, the final mix ratio scheme is output. If not, return to step S3 for iterative optimization to complete the mix ratio closed-loop design.

[0007] Preferably, in step S1, the specific steps are as follows: Step S11: Determine the core independent variables of the low-carbon concrete mix proportion, including cement content C, fly ash content F, mineral powder content K, and recycled coarse aggregate replacement rate R. g , Recycled fine aggregate replacement rate R x The water-cement ratio W / B and the water-reducing agent dosage J are all limited to the range of values ​​allowed by the engineering specifications. Step S12: Expand the sample dataset by combining orthogonal experiments, random sampling experiments, and historical measured data from the project; each sample in the dataset contains all independent and dependent variable parameters, including compressive strength. f cu Permeability grade P, drying shrinkage rate Carbon emissions per unit volume (E), unit material cost (M); Step S13: Use the 3σ criterion to remove outliers from the dataset, and complete the unified quantization process of the data using the extreme value normalization formula; the extreme value normalization formula is as follows: ; in, For normalized data, This is the original data. , These represent the maximum and minimum sample values ​​for the corresponding parameters. This is the offset correction factor; Step S14: Divide the preprocessed dataset into training set, validation set, and test set for subsequent model training and accuracy verification.

[0008] Preferably, in step S2, the attention-based weighted fusion multi-objective coupled prediction model uses a BP neural network as its backbone and introduces a channel attention mechanism to dynamically assign weights to different ratio parameters, as shown in the following formula: ; ; in, As a comprehensive evaluation index for concrete, For the proportion parameter vector, This is the fitting function for mechanical properties. This is the durability performance fitting function. This is the function for fitting economic performance. The function is the carbon emission fitting function. For the dynamic weight coefficients of the attention mechanism, and satisfying... ; The dynamic weight coefficients of the attention mechanism are adaptively updated through the attention mechanism, as shown in the following formula: ; in, For the first The feature importance score of each evaluation indicator For the first The feature importance score of each evaluation indicator For the first Dynamic weighting coefficients for attention mechanisms.

[0009] Preferably, the multi-dimensional performance includes mechanical properties, durability, and economic performance, and the calculation formula is as follows: Mechanical property fitting function The calculation formula is as follows: ; in, , , , It is a non-linear power correction factor. , , , These are the mechanical fitting weighting coefficients. This is the baseline correction constant for mechanical properties; Durability performance fitting function The calculation formula is as follows: ; in, , , These are the durability fitting weighting coefficients. To adjust the index for water-reducing agents, This is a baseline correction constant for durability performance; Economic performance fitting function The calculation formula is as follows: ; in, , , These are the unit prices for cement, fly ash, and mineral powder, respectively. , These represent the cost difference between using recycled coarse and fine aggregates to replace natural aggregates. This is a revised value for the fixed cost of auxiliary materials.

[0010] Preferably, the carbon emission fitting function adopts an original component superposition correction calculation method to accurately quantify the carbon reduction gain of multi-solid waste admixtures. The calculation formula is as follows: ; in, This refers to the carbon emissions per unit volume of pure cement-based concrete. , These are the carbon reduction correction factors for fly ash and mineral powder, respectively. The carbon emission reduction correction for replacing natural aggregates with recycled aggregates is obtained by linear fitting based on the aggregate replacement rate.

[0011] Preferably, in step S3, the multi-objective optimization objective function focuses on minimizing carbon emissions, achieving optimal performance, and minimizing cost, and the formula is as follows: ; in, To obtain the optimal objective function value, For scene-adaptive weighting coefficients, j =1,2,3,4,5; , , , , These are the normalized carbon emissions, material cost, compressive strength, impermeability grade, and drying shrinkage rate, respectively.

[0012] Preferably, the constraints of the multi-constraint optimization model for low-carbon concrete mix proportions include: compressive strength greater than or equal to the engineering design strength, impermeability grade greater than or equal to the design impermeability grade, drying shrinkage rate less than or equal to the specification limit, the dosage of each raw material within the specification allowable range, and the water-cement ratio within the suitable range for low-carbon concrete.

[0013] Preferably, the scene-adaptive weight coefficient A dynamic adjustment mechanism is adopted, and the adjustment formula is as follows: ; in, As the initial baseline weights, This represents the average carbon emissions of the current iterative population. The system sets a preset target carbon emission threshold. When the actual carbon emissions exceed the target threshold, the carbon emission weight is automatically increased to prioritize the carbon reduction target. Once the carbon emissions meet the target, the weights of mechanical and durability performance are automatically increased to achieve a dynamic balance among multiple objectives.

[0014] Preferably, in step S4, the improved adaptive sparrow search intelligent optimization algorithm specifically includes the following iterative optimization steps: Step S41: Initialize the population using chaotic Tent mapping to generate a uniformly distributed initial population parameter combination, thereby improving population diversity and avoiding convergence bias caused by initial population concentration. The chaotic initialization formula is as follows: ; in, For the first n The chaotic sequence value after +1 iteration update For the first n The chaotic sequence value of the next iteration; The initial population is obtained by iteratively generating a uniform chaotic sequence of [0,1] and mapping it to the range of values ​​of the matching parameter. Step S42: Introduce a nonlinear convergence factor to replace the traditional fixed convergence factor. The formula for calculating the convergence factor is as follows: ; in, The convergence factor is , These are the minimum and maximum values ​​of the convergence factor, respectively. This represents the current iteration number. This represents the maximum number of iterations. Step S43: Introduce an adaptive Gaussian mutation perturbation mechanism. When the optimal value has not been updated for 5 consecutive iterations, the optimal individual is subjected to mutation perturbation. The mutation formula is: ; in, The optimal ratio of individuals after mutation perturbation. The optimal ratio of individuals at present. The coefficient of variation is adaptive and decreases with the number of iterations. For the proportioning parameter dimension; Step S44: Iteratively update the positions of discoverers, followers, and watchdogs, continuously optimize the ratio parameters until the maximum number of iterations or the objective function convergence condition is met, and output the globally optimal ratio scheme.

[0015] Preferably, the specific steps for performance verification and experimental validation in step S5 are as follows: Step S51: Substitute the optimal ratio into the trained multi-objective coupled prediction model to predict various performance and carbon emission indicators, and preliminarily verify whether they meet the preset standards. Step S52: Prepare concrete samples according to the optimal mix ratio, conduct mechanical and durability tests, and simultaneously measure the carbon emissions per unit volume of concrete using the carbon emission accounting formula. Step S53: Compare the measured value with the predicted value. If the error is less than 5% and all indicators meet the standards, the final ratio is determined. If the error exceeds the standard or the indicators do not meet the standards, the sample is added to the dataset, and the process returns to step S3 to retrain the model and iteratively optimize it to achieve closed-loop optimization of the ratio scheme.

[0016] Therefore, this invention proposes a low-carbon concrete material mix design method, the beneficial effects of which are as follows: (1) This invention breaks through the limitations of traditional single performance fitting models and constructs an attention mechanism weighted fusion multi-objective coupled prediction model, which realizes accurate coupling fitting of multiple dimensions of concrete mechanics, durability, economy and carbon emissions, and solves the industry pain point that it is difficult to accurately quantify the nonlinear correlation between the mix proportion performance and carbon emissions of multi-solid waste low-carbon concrete.

[0017] (2) This invention improves the adaptive sparrow search algorithm and integrates the triple optimization mechanism of chaotic population initialization, nonlinear convergence factor and adaptive Gaussian mutation. It solves the defects of traditional intelligent optimization algorithm that is slow to converge and prone to local optima. It can quickly solve the optimal ratio of multi-constraint and multi-objective collaborative optimization and adapt to the differentiated needs of different engineering scenarios.

[0018] (3) The present invention constructs a scenario-adaptive dynamic weight multi-objective optimization system, which achieves a dynamic balance between carbon reduction, performance and cost objectives through the weight adjustment formula. It abandons the one-sidedness of traditional fixed weight optimization, which not only ensures the mechanical stability and durability of concrete engineering applications, but also maximizes the carbon reduction effect of solid waste substitution.

[0019] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0020] Figure 1 This is a flowchart of a low-carbon concrete material mix design method according to the present invention; Figure 2 This is a schematic diagram comparing the compressive strength in an embodiment of the present invention; Figure 3 This is a schematic diagram comparing carbon emissions per unit volume in an embodiment of the present invention. Detailed Implementation

[0021] To make the technical solutions, advantages, and objectives of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below. The described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the protection scope of the present invention.

[0022] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0023] Example 1 This embodiment focuses on the mix design of C30 grade low-carbon recycled aggregate concrete, with preset engineering constraints: compressive strength ≥ 30 MPa, impermeability grade ≥ P8, and drying shrinkage ≤ 3 × 10⁻⁶. -4 Carbon emissions per unit volume ≤280kg / m³ 3 Mix proportion parameters range: Cement content C = 220~300 kg / m³ 3 Fly ash content F = 40~80 kg / m³ 3 Mineral powder content K = 30~60 kg / m³ 3 , Recycled coarse aggregate replacement rate R g=0~50%, Recycled fine aggregate replacement rate R x =0~40%, water-cement ratio W / B=0.35~0.45, water-reducing agent dosage J=1.5%~2.5%.

[0024] like Figure 1 As shown, this invention provides a method for designing the mix proportions of low-carbon concrete, and the specific implementation steps are as follows: Step S1: Construct a multi-group mix proportion dataset for low-carbon concrete, collect the dosage parameters of cement, fly ash, mineral powder, recycled coarse aggregate, recycled fine aggregate, water, and water-reducing agent, as well as the measured data of concrete compressive strength, impermeability grade, drying shrinkage rate, carbon emissions, and material costs under the corresponding mix proportions, and complete data cleaning, normalization, and outlier removal. In this embodiment, orthogonal experiments combined with random sampling were used to obtain 300 valid samples. Four outlier samples were removed using the 3σ criterion, leaving 296 valid samples. An extreme value normalization formula was used to set a shift correction coefficient. =0.05, complete the normalization of all parameters, divide the training set into 208 groups, the validation set into 59 groups, and the test set into 29 groups according to the ratio of 7:2:1, eliminate the difference in units and avoid model fitting failure.

[0025] Step S2: Construct a multi-objective coupled prediction model based on attention mechanism weighted fusion, establish a nonlinear mapping relationship between mix proportion parameters and multi-dimensional performance and carbon emissions, and output the comprehensive performance parameters and carbon emission values ​​of concrete corresponding to any mix proportion. In this embodiment, an attention-weighted backpropagation (BP) neural network is constructed, with 7-dimensional parameters in the input layer, 12 neurons in the hidden layer, and 5 evaluation metrics in the output layer. The parameters are set as follows: ~ =1.0, fly ash carbon reduction coefficient =0.70, carbon reduction coefficient of mineral powder =0.75. After the model training converges, the optimal weight coefficients are output: , , =0.20、 The constraint that the sum of the weights is 1 is satisfied.

[0026] Substituting the above parameters into the multi-dimensional performance formula (C=245kg / m) 3 F=65kg / m 3 K=45kg / m 3 R g =45%, R x =35%, W / B=0.40, J=2.0%), calculate mechanical properties, durability properties and economic properties: Mechanical property fitting calculation: In this embodiment, the mechanical property baseline correction constant Mechanical fitting weight coefficients , , , Substituting into the mechanical fitting formula yields =33.58MPa, which is basically consistent with the measured compressive strength of 33.6MPa, with a calculation error of 0.06%.

[0027] Durability performance fitting calculation: In this embodiment, the durability performance baseline correction constant is... Durability fitting weight coefficients , , Water-reducing agent compatibility correction index =0.8, substituting into the durability fitting formula, we obtain the comprehensive durability index corresponding to the impermeability grade P10 and the drying shrinkage rate of 2.68×10. -4 , compared with the measured value of 2.7×10 -4 Error 0.7%.

[0028] Economic performance fitting calculation: In this embodiment, =0.45 yuan / kg =0.18 yuan / kg =0.20 yuan / kg, aggregate cost adjustment , =-0.05 yuan / kg, fixed auxiliary material cost =12 yuan, the calculated material cost per unit volume M = 148.6 yuan / m 3 .

[0029] Carbon emission fitting calculation: taking baseline carbon emissions =385kg / m 3 Carbon emission reduction correction of recycled aggregate replacing natural aggregate =0.12, calculated as follows =261.8kg / m 3 Compared with the measured carbon emissions of 262 kg / m³ 3 The error is 0.07%, which meets the preset carbon reduction threshold requirements.

[0030] Substituting the calculated results of mechanical properties, durability, economic performance, and carbon emissions into the multi-objective coupled prediction model, the comprehensive optimal evaluation score Y=0.926 was obtained, which is the optimal score among all the ratio samples in this study.

[0031] Step S3: With low carbon, high performance, and low cost as the core objectives, and combined with engineering specification constraints, construct a multi-objective optimization objective function, introduce a dynamic weight adaptive adjustment mechanism, and establish a multi-constraint optimization model for low carbon concrete mix proportions. In this embodiment, an initial baseline weight is preset. =0.35、 =0.20、 =0.25、 =0.15、 =0.05, preset target carbon emission threshold =280kg / m 3 Average carbon emissions of the population in the early stages of iteration =312kg / m 3 Substitute the values ​​into the dynamic weight adjustment formula to automatically update the weights: =0.39、 =0.20、 =0.23、 =0.13、 =0.05, automatically increasing the carbon emission optimization weight to prioritize carbon reduction; in later iterations, after carbon emissions meet the target, the weight adaptively adjusts back to balance performance and cost. Substituting the updated weights into the multi-objective optimization objective function, the calculation is... =0.083, which is the global minimum convergence value.

[0032] Step S4: An improved adaptive sparrow search intelligent optimization algorithm is used to iteratively solve the multi-constraint optimization model. The optimization algorithm introduces a chaotic initialization population, a nonlinear convergence factor, and an adaptive mutation perturbation mechanism to quickly converge and obtain the globally optimal low-carbon concrete material mix ratio. In this embodiment, the algorithm parameters are set to be fixed: maximum number of iterations T=100, population size 50, and convergence factor extreme value. =0.9、 =0.2, parameter dimension =7. First, a uniform initial population is generated using the chaotic Tent mapping formula to avoid initial clustering problems. During the iteration process, a nonlinear convergence factor formula is used, with large weights for global search in the early stages and small weights for local fine-tuning in the later stages. When the optimal value is reached after 20-25 iterations, an adaptive Gaussian mutation perturbation mechanism is triggered, with the mutation coefficient... Adaptive decreasing algorithm successfully escaped local optima. The algorithm finally converged after 28 iterations, outputting the globally optimal combination of parameters: C = 245 kg / m³. 3 F=65kg / m 3 K=45kg / m 3 R g =45%, R x =35%, W / B=0.40, J=2.0%.

[0033] Step S5: Perform a quantitative analysis on the optimal low-carbon concrete material mix proportion. gIt can be verified and tested. If the preset engineering indicators and carbon reduction indicators are met, the final ratio scheme is output. If not, it returns to step S3 for iterative optimization to complete the closed-loop design of the ratio.

[0034] In this embodiment, the sample was prepared according to the optimal ratio output by the algorithm. The measured core indicators were: compressive strength 33.6 MPa, impermeability grade P10, and drying shrinkage rate 2.7 × 10⁻⁶. -4 Carbon emissions per unit volume: 262 kg / m³ 3 The maximum error between all calculated and predicted values ​​and the measured values ​​is 3.2%, which is less than the preset error threshold of 5%, and all meet the engineering constraints and carbon reduction requirements.

[0035] To further and intuitively verify the superiority of the technical solution of this invention, the optimal ratio scheme of this embodiment is compared and verified with the traditional manual trial mixing method and the conventional machine learning + traditional optimization method under the same working conditions and constraints. The comparison results of each technical system are shown in Table 1 below.

[0036] Table 1. Comparison of indicators between the present invention and the prior art

[0037] From Table 1 and Figures 2-3 It can be seen that, under the same C30 engineering constraints and the same raw material system, the technical solution of this invention has significant advantages in all aspects, as detailed in the following comparative analysis: In terms of performance indicators, the optimized mix design of this invention improves the compressive strength by 7.7% and 4.7% compared to traditional trial mix design and conventional intelligent methods, respectively. The impermeability grade is improved from P7 and P8 to P10, the drying shrinkage rate is significantly reduced, and the crack resistance and impermeability durability of concrete are greatly improved. This invention completely solves the engineering defects of existing low-carbon mix design, which are characterized by "low carbon content leading to performance degradation and insufficient durability".

[0038] In terms of low-carbon and economic indicators, the carbon emissions of this invention are reduced by 22.3% compared to traditional concrete and by 12.1% compared to conventional intelligent optimization schemes; the material cost is the lowest, achieving a dual benefit of carbon reduction and cost reduction; at the same time, the comprehensive utilization rate of solid waste is greatly increased to 50.8%, maximizing the utilization of industrial solid waste and recycled aggregates, and the resource utilization effect of solid waste is significantly better than that of existing technologies.

[0039] In terms of modeling and optimization efficiency, this invention relies on an attention-based multi-objective coupling model, with a prediction error of only 3.2%, far lower than the 9.6% error accuracy of conventional models, and significantly improves the accuracy of multi-index nonlinear fitting. At the same time, the improved sparrow search algorithm significantly reduces the number of convergence iterations, improving the convergence efficiency by more than 40% compared to traditional optimization algorithms, eliminating the tedious process of repeated manual matching, and significantly improving the intelligence and efficiency of matching design.

[0040] At the technical mechanism level, existing technologies use fixed-weight optimization logic, which cannot take into account the dynamic balance of multiple indicators. However, this invention uses a dynamic adaptive weight adjustment mechanism to automatically adjust the optimization priority according to the carbon emission threshold. Under the premise of meeting the requirements of concrete mechanics and durability performance specifications, it maximizes the potential for low-carbon and carbon reduction, and solves the technical bottleneck of "single optimization and indicator imbalance" in traditional technologies. Its engineering versatility and practicality are far superior to existing technical solutions.

[0041] It is worth noting that all contents not described in detail in this invention are existing technologies and are well known to those skilled in the art.

[0042] Therefore, this invention provides a low-carbon concrete material mix design method. By constructing a multi-objective coupled prediction model with attention mechanism weighted fusion, it accurately fits multiple dimensions of indicators such as mechanics, durability, economy, and carbon emissions, solving the problem of difficulty in quantifying the nonlinear correlation between the performance and carbon emissions of low-carbon concrete with multiple solid wastes. At the same time, it introduces an improved adaptive sparrow search algorithm, integrating a triple optimization mechanism of chaotic population initialization, nonlinear convergence factor, and adaptive Gaussian mutation, overcoming the shortcomings of traditional algorithms such as slow convergence and easy local optima, and realizing the rapid solution of multi-constraint, multi-objective collaborative optimal mix ratio. Furthermore, it constructs a scenario-adaptive dynamic weighted multi-objective optimization system, which dynamically balances carbon reduction, performance, and cost objectives through weight adjustment, abandoning the one-sidedness of fixed weight optimization, and maximizing the carbon reduction effect of solid waste substitution while ensuring the mechanical stability and durability of concrete.

[0043] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for designing the mix proportions of low-carbon concrete, characterized in that, The specific steps are as follows: Step S1: Construct a multi-group mix proportion dataset for low-carbon concrete, collect the dosage parameters of cement, fly ash, mineral powder, recycled coarse aggregate, recycled fine aggregate, water, and water-reducing agent, as well as the measured data of concrete compressive strength, impermeability grade, drying shrinkage rate, carbon emissions, and material costs under the corresponding mix proportions, and complete data cleaning, normalization, and outlier removal. Step S2: Construct a multi-objective coupled prediction model based on attention mechanism weighted fusion, establish a nonlinear mapping relationship between mix proportion parameters and multi-dimensional performance and carbon emissions, and output the comprehensive performance parameters and carbon emission values ​​of concrete corresponding to any mix proportion. Step S3: With low carbon, high performance, and low cost as the core objectives, and combined with engineering specification constraints, construct a multi-objective optimization objective function, introduce a dynamic weight adaptive adjustment mechanism, and establish a multi-constraint optimization model for low carbon concrete mix proportions. Step S4: An improved adaptive sparrow search intelligent optimization algorithm is used to iteratively solve the multi-constraint optimization model. The optimization algorithm introduces a chaotic initialization population, a nonlinear convergence factor, and an adaptive mutation perturbation mechanism to quickly converge and obtain the globally optimal low-carbon concrete material mix ratio. Step S5: Perform performance verification and experimental validation on the optimal low-carbon concrete material mix ratio. If the preset engineering indicators and carbon reduction indicators are met, the final mix ratio scheme is output. If not, return to step S3 for iterative optimization to complete the mix ratio closed-loop design.

2. The low-carbon concrete material mix design method according to claim 1, characterized in that, In step S1, the specific steps are as follows: Step S11: Determine the core independent variables of the low-carbon concrete mix proportion, including cement content C, fly ash content F, mineral powder content K, and recycled coarse aggregate replacement rate R. g , Recycled fine aggregate replacement rate R x The water-cement ratio W / B and the water-reducing agent dosage J are all limited to the range of values ​​allowed by the engineering specifications. Step S12: Expand the sample dataset by combining orthogonal experiments, random sampling experiments, and historical measured data from the project. A single sample in the dataset contains all independent and dependent variable parameters. The dependent variable parameters include compressive strength. f cu Permeability grade P, drying shrinkage rate Carbon emissions per unit volume (E), unit material cost (M); Step S13: Use the 3σ criterion to remove outliers from the dataset, and complete the unified quantization process of the data using the extreme value normalization formula; the extreme value normalization formula is as follows: ； in, For normalized data, This is the original data. , These represent the maximum and minimum sample values ​​for the corresponding parameters. This is the offset correction factor; Step S14: Divide the preprocessed dataset into training set, validation set, and test set for subsequent model training and accuracy verification.

3. The low-carbon concrete material mix design method according to claim 2, characterized in that, In step S2, the multi-objective coupled prediction model with attention mechanism weighted fusion uses a BP neural network as its basic framework and introduces a channel attention mechanism to dynamically assign weights to parameters with different ratios, as shown in the following formula: ； ； in, As a comprehensive evaluation index for concrete, For the proportion parameter vector, This is the fitting function for mechanical properties. This is the durability performance fitting function. This is the function for fitting economic performance. The function is the carbon emission fitting function. For the dynamic weight coefficients of the attention mechanism, and satisfying... ; The dynamic weight coefficients of the attention mechanism are adaptively updated through the attention mechanism, as shown in the following formula: ； in, For the first The feature importance score of each evaluation indicator For the first The feature importance score of each evaluation indicator For the first Dynamic weighting coefficients for attention mechanisms.

4. The low-carbon concrete material mix design method according to claim 3, characterized in that, Multi-dimensional performance includes mechanical properties, durability, and economic performance, and the calculation formula is as follows: Mechanical property fitting function The calculation formula is as follows: ； in, , , , It is a non-linear power correction factor. , , , These are the mechanical fitting weighting coefficients. This is the baseline correction constant for mechanical properties; Durability performance fitting function The calculation formula is as follows: ； in, , , These are the durability fitting weighting coefficients. To adjust the index for water-reducing agents, This is a baseline correction constant for durability performance; Economic performance fitting function The calculation formula is as follows: ； in, , , These are the unit prices for cement, fly ash, and mineral powder, respectively. , These represent the cost difference between using recycled coarse and fine aggregates to replace natural aggregates. This is a revised value for the fixed cost of auxiliary materials.

5. The low-carbon concrete material mix design method according to claim 4, characterized in that, The carbon emission fitting function employs an original component superposition correction calculation method to accurately quantify the carbon reduction gain of multi-solid waste admixtures. The calculation formula is as follows: ； in, This refers to the carbon emissions per unit volume of pure cement-based concrete. , These are the carbon reduction correction factors for fly ash and mineral powder, respectively. The carbon emission reduction correction for replacing natural aggregates with recycled aggregates is obtained by linear fitting based on the aggregate replacement rate.

6. The low-carbon concrete material mix design method according to claim 5, characterized in that, In step S3, the multi-objective optimization objective function focuses on minimizing carbon emissions, achieving optimal performance, and minimizing cost, and the formula is as follows: ； in, To obtain the optimal objective function value, For scene-adaptive weighting coefficients, j =1,2,3,4,5; , , , , These are the normalized carbon emissions, material cost, compressive strength, impermeability grade, and drying shrinkage rate, respectively.

7. The low-carbon concrete material mix design method according to claim 6, characterized in that, The constraints of the multi-constraint optimization model for low-carbon concrete mix proportions include: compressive strength greater than or equal to the engineering design strength, impermeability grade greater than or equal to the design impermeability grade, drying shrinkage rate less than or equal to the specification limit, the dosage of each raw material within the specification allowable range, and the water-cement ratio within the suitable range for low-carbon concrete.

8. The low-carbon concrete material mix design method according to claim 7, characterized in that, Scene Adaptive Weight Coefficient A dynamic adjustment mechanism is adopted, and the adjustment formula is as follows: ； in, As the initial baseline weights, This represents the average carbon emissions of the current iterative population. The system sets a preset target carbon emission threshold. When the actual carbon emissions exceed the target threshold, the carbon emission weight is automatically increased to prioritize the carbon reduction target. Once the carbon emissions meet the target, the weights of mechanical and durability performance are automatically increased to achieve a dynamic balance among multiple objectives.

9. The low-carbon concrete material mix design method according to claim 8, characterized in that, In step S4, the improved adaptive sparrow search intelligent optimization algorithm specifically includes the following iterative optimization steps: Step S41: Initialize the population using chaotic Tent mapping to generate a uniformly distributed initial population parameter combination, thereby improving population diversity and avoiding convergence bias caused by initial population concentration. The chaotic initialization formula is as follows: ； in, For the first n The chaotic sequence value after +1 iteration update For the first n The chaotic sequence values ​​of the next iteration; The initial population is obtained by iteratively generating a uniform chaotic sequence of [0,1] and mapping it to the range of values ​​of the matching parameter. Step S42: Introduce a nonlinear convergence factor to replace the traditional fixed convergence factor. The formula for calculating the convergence factor is as follows: ； in, The convergence factor is , These are the minimum and maximum values ​​of the convergence factor, respectively. This represents the current iteration number. This represents the maximum number of iterations. Step S43: Introduce an adaptive Gaussian mutation perturbation mechanism. When the optimal value has not been updated for 5 consecutive iterations, the optimal individual is subjected to mutation perturbation. The mutation formula is: ； in, The optimal ratio of individuals after mutation perturbation. The optimal ratio of individuals at present. The coefficient of variation is adaptive and decreases with the number of iterations. For the proportioning parameter dimension; Step S44: Iteratively update the positions of discoverers, followers, and watchdogs, continuously optimize the ratio parameters until the maximum number of iterations or the objective function convergence condition is met, and output the globally optimal ratio scheme.

10. The low-carbon concrete material mix design method according to claim 9, characterized in that, In step S5, the specific steps for performance verification and experimental validation are as follows: Step S51: Substitute the optimal ratio into the trained multi-objective coupled prediction model to predict various performance and carbon emission indicators, and preliminarily verify whether they meet the preset standards. Step S52: Prepare concrete samples according to the optimal mix ratio, conduct mechanical and durability tests, and simultaneously measure the carbon emissions per unit volume of concrete using the carbon emission accounting formula. Step S53: Compare the measured value with the predicted value. If the error is less than 5% and all indicators meet the standards, the final ratio is determined. If the error exceeds the standard or the indicators do not meet the standards, the sample is added to the dataset, and the process returns to step S3 to retrain the model and iteratively optimize it to achieve closed-loop optimization of the ratio scheme.

Images