Concrete carbon sequestration process parameter optimization method

Abstract

The invention provides a concrete carbon sequestration process parameter optimization method, and belongs to the technical field of concrete carbon sequestration. Chemical component data and real-time parameters are input into a carbonization parameter intelligent regulation and control model to obtain regulation values of carbon dioxide concentration, reaction temperature, reaction pressure and water content, and a carbonization reactor is dynamically regulated and controlled; when the carbonization depth is greater than 20mm and the carbon sequestration amount reaches more than 80% of a theoretical maximum value, completing carbonization treatment and recording a process optimization result, and training an intelligent regulation and control model through a parameter regularization framework based on adaptive weight attenuation so that the intelligent regulation and control model can output an optimal parameter adjustment value according to the chemical component difference of the raw material and the real-time change of the carbonization process; the technical problem that the carbon sequestration efficiency is unstable due to the fact that waste concrete carbon sequestration process parameters cannot be dynamically optimized and adjusted according to the chemical component difference of raw materials and the real-time change of the carbonization process is solved.

Description

Technical Field

[0001] This invention belongs to the field of concrete carbon sequestration technology, and more specifically, relates to a method for optimizing concrete carbon sequestration process parameters. Background Technology

[0002] Carbon sequestration technology for waste concrete is an important carbon reduction technology that involves introducing carbon dioxide gas into waste concrete, causing the carbon dioxide to react with alkaline substances such as calcium hydroxide in the concrete to form calcium carbonate, thereby achieving carbon dioxide sequestration. This technology has broad application prospects in the resource utilization of construction waste and the achievement of carbon neutrality goals, and has already been initially applied in scenarios such as waste concrete recycling and precast component curing. However, in existing technologies, due to the complex sources of waste concrete leading to significant differences in chemical composition, and the continuous changes in parameters such as temperature, humidity, and pressure during carbonization, traditional carbon sequestration processes typically use fixed parameters for carbon dioxide concentration, reaction temperature, reaction pressure, and moisture content. This fails to dynamically adjust the process parameters according to the characteristics of the raw materials and fluctuations in process parameters, resulting in large variations in the carbonization depth and carbon sequestration amount between different batches of raw materials. In other words, existing technologies suffer from the technical problem that the process parameters for waste concrete carbon sequestration cannot be dynamically optimized and adjusted according to the differences in the chemical composition of the raw materials and the real-time changes in the carbonization process, leading to unstable carbon sequestration efficiency. Summary of the Invention

[0003] In view of this, the present invention provides a method for optimizing the carbon sequestration process parameters of concrete, which can solve the technical problem in the prior art that the carbon sequestration process parameters of waste concrete cannot be dynamically optimized and adjusted according to the differences in the chemical composition of raw materials and the real-time changes in the carbonization process, resulting in unstable carbon sequestration efficiency.

[0004] This invention is implemented as follows: This invention provides a method for optimizing concrete carbonation process parameters, comprising the following steps: collecting waste concrete raw material samples and performing chemical composition analysis to obtain the mass percentage content of silica, calcium oxide, aluminum oxide, ferric oxide, potassium oxide, magnesium oxide, sulfur trioxide, and sodium oxide; simultaneously measuring the initial porosity and initial specific surface area of ​​the waste concrete raw material samples; pre-crushing the waste concrete raw material samples and measuring the specific surface area after crushing, calculating the ratio of the crushed specific surface area to the initial specific surface area to obtain the specific surface area increase ratio; placing the crushed waste concrete raw material samples into a carbonization reactor and introducing... The gas was monitored, and real-time reaction temperature, humidity, pressure, and humidity were collected every 30 minutes during the carbonization reaction. Concentration; real-time reaction temperature, real-time reaction humidity, real-time reaction pressure, real-time Concentration and chemical composition detection data are input into the intelligent carbonization parameter control model, and the intelligent carbonization parameter control model outputs... Concentration adjustment value, reaction temperature adjustment value, reaction pressure adjustment value, and moisture content adjustment value; based on the output adjustment values, adjust the real-time values ​​in the carbonization reactor. The concentration, real-time reaction temperature, real-time reaction pressure, and real-time reaction humidity are dynamically adjusted. Waste concrete raw material samples are taken out, and the carbonation depth and carbon fixation amount after carbonation reaction are measured to determine whether the carbonation depth after carbonation reaction is greater than 20 mm and whether the carbon fixation amount after carbonation reaction reaches more than 80% of the theoretical maximum carbon fixation amount. Waste concrete raw material samples are tested by X-ray diffraction analysis to obtain the amount of calcium carbonate generated. The carbonation depth, carbon fixation amount, and calcium carbonate generation amount after carbonation reaction are recorded as process optimization result data.

[0005] In the pre-crushing process, the particle size of the crushed particles is controlled within the range of 5 to 15 mm, and the specific surface area after crushing is measured by a laser particle size analyzer.

[0006] Specifically, the step of placing the crushed waste concrete raw material sample into the carbonization reactor involves setting the initial moisture content to 8 to 12%. The concentration is 15% to 25%, the reaction temperature is controlled at 20 to 40°C, and the reaction pressure is set at 0.1 to 0.3 MPa.

[0007] Among them, the real-time adjustment of the carbonization reactor based on the output adjustment value In the step of dynamically adjusting the concentration, real-time reaction temperature, real-time reaction pressure, and real-time reaction humidity, the actual value after adjustment... Concentration and The deviation of the concentration adjustment value shall not exceed 2%.

[0008] In the dynamic adjustment step, the deviation between the adjusted actual reaction temperature and the adjusted reaction temperature value shall not exceed 3°C, the deviation between the adjusted actual reaction pressure and the adjusted reaction pressure value shall not exceed 0.02 MPa, and the deviation between the adjusted actual moisture content and the adjusted moisture content value shall not exceed 1%.

[0009] Specifically, the step of taking out the waste concrete raw material sample is to take out the waste concrete raw material sample 120 to 180 minutes after the carbonization reaction has been carried out.

[0010] In the step of determining whether the carbonization depth after the carbonization reaction is greater than 20 mm and whether the carbon fixation amount after the carbonization reaction reaches more than 80% of the theoretical maximum carbon fixation amount, if the carbonization depth after the carbonization reaction is not greater than 20 mm or the carbon fixation amount after the carbonization reaction does not reach 80% of the theoretical maximum carbon fixation amount, then the initial... The carbonization reaction was repeated after increasing the concentration by 5% and increasing the reaction pressure by 0.05 MPa.

[0011] The input layer of the intelligent control model for carbonization parameters contains 12 neurons corresponding to the mass percentages of silicon dioxide, calcium oxide, aluminum oxide, ferric oxide, potassium oxide, magnesium oxide, sulfur trioxide, sodium oxide, real-time reaction temperature, real-time reaction humidity, real-time reaction pressure, and real-time... concentration.

[0012] The hidden layer of the intelligent control model for carbonization parameters is a three-layer fully connected layer. The first hidden layer contains 64 neurons, the second hidden layer contains 32 neurons, and the third hidden layer contains 16 neurons. The activation function of each hidden layer is a modified linear unit function.

[0013] The output layer of the intelligent carbonization parameter control model contains four neurons, each corresponding to... Concentration adjustment value, reaction temperature adjustment value, reaction pressure adjustment value, and moisture content adjustment value.

[0014] The establishment of the training dataset for the intelligent control model of carbonation parameters includes: collecting 300 groups of waste concrete raw material samples with different chemical compositions, and performing different tests on each group of waste concrete raw material samples at different chemical compositions. Carbonization experiments were conducted under different concentrations, reaction temperatures, reaction pressures, and moisture contents.

[0015] In the carbonization experiment, The concentration range is 10 to 30% with an interval of 5%, the reaction temperature range is 15 to 45℃ with an interval of 5℃, the reaction pressure range is 0.05 to 0.35MPa with an interval of 0.05MPa, and the moisture content range is 5% to 15% with an interval of 2%.

[0016] The training dataset establishment includes: recording chemical composition data, process parameter data, carbonization depth data, and carbon fixation amount data for each carbonization experiment, and marking carbonization experiment data with a carbonization depth greater than 20 mm and a carbon fixation amount exceeding 80% of the theoretical maximum carbon fixation amount as high-quality samples, thus constructing a training dataset containing 1500 sets of data.

[0017] The training of the intelligent carbonization parameter control model adopts a parameter regularization framework based on adaptive weight decay. A weight decay term is introduced into the loss function. The weight decay term is composed of the sum of squares of the parameters of each layer and the product of the decay coefficient of the corresponding layer.

[0018] In the training of the intelligent control model for carbonization parameters, the hierarchical importance index is calculated based on the magnitude of the L2 norm of the gradient of each layer and the parameter update magnitude during each iteration. The hierarchical importance index is equal to the product of the quotient of the L2 norm of the current layer gradient divided by the sum of the L2 norms of the gradients of all layers and the quotient of the current layer parameter update magnitude divided by the sum of the parameter update magnitudes of all layers.

[0019] The intelligent control model for carbonization parameters dynamically adjusts the layer decay coefficient of each layer based on the layer importance index, sets the initial learning rate to 0.001, and uses an adaptive moment estimation optimization algorithm to update the parameters. Training is terminated when the validation set loss does not decrease for 20 consecutive rounds.

[0020] This invention solves the technical problem of unstable carbon sequestration efficiency by constructing an intelligent control model for carbonation parameters to dynamically optimize the carbon sequestration process parameters of waste concrete. This intelligent control model takes the content of eight chemical components and four real-time reaction parameters of waste concrete as input. It captures the complex nonlinear mapping relationship between chemical components and optimal process parameters through a three-layer fully connected neural network, outputting parameter adjustment values ​​based on the current raw material characteristics and reaction state. This allows for refined dynamic control of the carbonation process based on raw material differences and process fluctuations. This invention trains the model using a parameter regularization framework based on adaptive weight decay. It identifies network layers with high contributions to carbon sequestration parameter optimization through hierarchical importance indicators and applies differentiated regularization constraints. This ensures accurate prediction of known operating conditions while enhancing the generalization ability to unknown raw material components, avoiding efficiency fluctuations caused by fixed parameters. In summary, this invention solves the technical problem mentioned in the background art where the carbon sequestration process parameters of waste concrete cannot be dynamically optimized and adjusted according to the differences in raw material chemical composition and real-time changes in the carbonation process, leading to unstable carbon sequestration efficiency. Attached Figure Description

[0021] Figure 1 This is a flowchart of the method of the present invention.

[0022] Figure 2 This is a schematic diagram of the carbonization reactor structure.

[0023] Figure 3 This is a schematic diagram of the network structure of the intelligent control model for carbonization parameters.

[0024] Figure 4 This is a curve showing the dynamic adjustment of process parameters during the carbonization reaction.

[0025] Figure 5 This is a comparison of X-ray diffraction patterns before and after the carbonization reaction. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below.

[0027] like Figure 1 The diagram shown is a flowchart of a method for optimizing concrete carbon sequestration process parameters provided by the present invention. This method includes the following steps:

[0028] S01. Collect waste concrete raw material samples and conduct chemical composition analysis to obtain the mass percentage content of silicon dioxide, calcium oxide, aluminum oxide, ferric oxide, potassium oxide, magnesium oxide, sulfur trioxide, and sodium oxide. At the same time, measure the initial porosity and initial specific surface area of ​​the waste concrete raw material samples.

[0029] S02. The waste concrete raw material sample in step S01 is pre-crushed. The particle size of the crushed particles is controlled within the range of 5 to 15 mm. The specific surface area after crushing is measured by a laser particle size analyzer. The ratio of the specific surface area after crushing to the initial specific surface area is calculated to obtain the specific surface area increase ratio.

[0030] S03. Place the crushed waste concrete raw material sample from step S02 into the carbonization reactor, set the initial moisture content to 8 to 12%, introduce carbon dioxide gas with an initial carbon dioxide concentration of 15 to 25%, control the reaction temperature at 20 to 40°C, set the reaction pressure to 0.1 to 0.3 MPa, and collect real-time reaction temperature, real-time reaction humidity, real-time reaction pressure, and real-time carbon dioxide concentration every 30 minutes during the carbonization reaction.

[0031] S04. Input the real-time reaction temperature, real-time reaction humidity, real-time reaction pressure, and real-time carbon dioxide concentration collected in step S03, along with the mass percentages of silicon dioxide, calcium oxide, aluminum oxide, ferric oxide, potassium oxide, magnesium oxide, sulfur trioxide, and sodium oxide obtained in step S01, into the intelligent control model for carbonization parameters. The intelligent control model for carbonization parameters outputs adjustment values ​​for carbon dioxide concentration, reaction temperature, reaction pressure, and moisture content.

[0032] S05. Based on the carbon dioxide concentration adjustment value, reaction temperature adjustment value, reaction pressure adjustment value, and moisture content adjustment value output in step S04, dynamically adjust the real-time carbon dioxide concentration, real-time reaction temperature, real-time reaction pressure, and real-time reaction humidity in the carbonization reactor. The deviation between the adjusted actual carbon dioxide concentration and the adjusted carbon dioxide concentration value shall not exceed 2%, the deviation between the adjusted actual reaction temperature and the adjusted reaction temperature value shall not exceed 3°C, the deviation between the adjusted actual reaction pressure and the adjusted reaction pressure value shall not exceed 0.02 MPa, and the deviation between the adjusted actual moisture content and the adjusted moisture content value shall not exceed 1%.

[0033] S06. After the carbonation reaction has proceeded for 120 to 180 minutes, take out a sample of the waste concrete raw material, measure the carbonation depth and carbon fixation amount after the carbonation reaction, and determine whether the carbonation depth after the carbonation reaction is greater than 20 mm and whether the carbon fixation amount after the carbonation reaction reaches more than 80% of the theoretical maximum carbon fixation amount. If the carbonation depth after the carbonation reaction is greater than 20 mm and the carbon fixation amount after the carbonation reaction reaches more than 80% of the theoretical maximum carbon fixation amount, proceed to step S07. If the carbonation depth after the carbonation reaction is not greater than 20 mm or the carbon fixation amount after the carbonation reaction does not reach 80% of the theoretical maximum carbon fixation amount, return to step S03 and increase the initial carbon dioxide concentration by 5% and increase the reaction pressure by 0.05 MPa.

[0034] S07. The waste concrete raw material sample taken in step S06 is tested by X-ray diffraction analysis to obtain the amount of calcium carbonate generated. The carbonation depth, the amount of carbon solidified after the carbonation reaction, and the amount of calcium carbonate generated are recorded as process optimization result data.

[0035] The specific structure of the intelligent control model for carbonization parameters is as follows: The input layer contains 12 neurons corresponding to the mass percentage content of silicon dioxide, calcium oxide, aluminum oxide, ferric oxide, potassium oxide, magnesium oxide, sulfur trioxide, sodium oxide, real-time reaction temperature, real-time reaction humidity, real-time reaction pressure, and real-time carbon dioxide concentration, respectively; the hidden layer is a three-layer fully connected layer, with the first hidden layer containing 64 neurons, the second hidden layer containing 32 neurons, and the third hidden layer containing 16 neurons, and the activation function of each hidden layer adopts the modified linear unit function; the output layer contains 4 neurons corresponding to the carbon dioxide concentration adjustment value, reaction temperature adjustment value, reaction pressure adjustment value, and moisture content adjustment value, respectively.

[0036] The steps for establishing the training dataset of the intelligent control model for carbonization parameters specifically include: collecting 300 sets of waste concrete raw material samples with different chemical compositions; conducting carbonization experiments on each set of waste concrete raw material samples under different carbon dioxide concentrations, reaction temperatures, reaction pressures, and moisture contents; the different carbon dioxide concentrations range from 10% to 30% with an interval of 5%; the different reaction temperatures range from 15% to 45℃ with an interval of 5℃; the different reaction pressures range from 0.05% to 0.35MPa with an interval of 0.05MPa; and the different moisture contents range from 5% to 15% with an interval of 2%; recording the chemical composition data, process parameter data, carbonization depth data, and carbon fixation data for each carbonization experiment; and marking carbonization experiment data with a carbonization depth greater than 20 mm and a carbon fixation amount reaching more than 80% of the theoretical maximum carbon fixation amount as high-quality samples; using the chemical composition data and process parameter data of the high-quality samples as input, and the corresponding process parameter data as output labels, constructing a training dataset containing 1500 sets of data.

[0037] The specific steps for training the intelligent carbonization parameter control model include: dividing the training dataset into a training set and a validation set in an 8:2 ratio; training the intelligent carbonization parameter control model using a parameter regularization framework based on adaptive weight decay, introducing a weight decay term into the loss function, wherein the weight decay term is composed of the product of the sum of squares of the parameters of each layer and the corresponding layer decay coefficient; and calculating the layer importance index based on the L2 norm of the gradient of each layer and the parameter update magnitude during each iteration, wherein the layer importance index is expressed as follows: the layer importance index is equal to the current... The product of the quotient of the layer gradient L2 norm divided by the sum of the L2 norms of all layers and the quotient of the current layer parameter update magnitude divided by the sum of the parameter update magnitudes of all layers is used. The layer decay coefficient of each layer is dynamically adjusted according to the layer importance index, so that layers with high layer importance index have smaller layer decay coefficients to retain more parameter information, and layers with low layer importance index have larger layer decay coefficients to enhance regularization constraints. The initial learning rate is set to 0.001, and the adaptive moment estimation optimization algorithm is used for parameter updates. Training is terminated when the validation set loss does not decrease for 20 consecutive rounds.

[0038] The adaptive weight decay-based parameter regularization framework identifies network layers that contribute significantly to the optimization of carbon sequestration process parameters by real-time monitoring of gradient flow characteristics and parameter evolution behavior at each layer. Weaker regularization constraints are applied to these high-contribution layers to fully preserve the complex nonlinear mapping relationships learned from the training dataset, while stronger regularization constraints are applied to low-contribution layers to prevent overfitting to noise and abnormal fluctuations in the training dataset. This adaptive weight decay-based parameter regularization framework enables the intelligent carbonation parameter control model to accurately capture the intrinsic correlation between different chemical components and optimal carbonation parameters when facing waste concrete raw materials with large fluctuations in chemical composition and significant differences in pore structure, avoiding underfitting or overfitting problems caused by fixed regularization strength. Through a layer-differentiated parameter decay strategy, the framework ensures accurate prediction of known operating conditions while enhancing the generalization ability to unknown raw materials. This allows the process parameter adjustment values ​​output by the intelligent carbonation parameter control model to adapt to real-time changes in raw material characteristics, achieving synergistic optimization of carbonation depth and carbon sequestration amount. The introduction of the parameter regularization framework based on adaptive weight decay enables the optimization method of carbon fixation process parameters to break free from the limitations of traditional experience-based parameter tuning. Through data-driven intelligent decision-making, it achieves refined dynamic control of process parameters, effectively solving the problem of unstable carbon fixation efficiency caused by fluctuations in waste concrete composition. At the same time, it avoids the negative impact of excessive carbonization on material strength, thereby improving the adaptability and reliability of the carbon fixation process.

[0039] The traditional concrete carbon sequestration process refers to introducing carbon into waste concrete... Gas, making With waste concrete Alkaline substances undergo carbonization reactions to produce Thus achieving The technical process of carbon sequestration. Traditional concrete carbon sequestration processes typically employ fixed... The carbonation process used parameters such as concentration, reaction temperature, reaction pressure, and moisture content, but failed to dynamically adjust them according to the differences in the chemical composition of the waste concrete and the changes in parameters during the carbonation process, resulting in limited carbon fixation efficiency and carbonation depth.

[0040] The pre-crushing treatment refers to mechanically crushing waste concrete raw material samples to increase their specific surface area and improve their efficiency. The increased contact area with alkaline substances promotes the carbonization reaction. The pre-crushing process controls the particle size after crushing, ensuring carbonization efficiency while avoiding blockage of gas diffusion channels caused by excessively fine particles.

[0041] The carbonization depth after the carbonization reaction refers to The depth to which carbonization occurs, penetrating from the surface of the waste concrete raw material sample into its interior. This carbonization depth reflects the extent of the carbonization reaction; a greater carbonization depth indicates a higher degree of participation of alkaline substances in the reaction, and consequently, a greater amount of carbon solidified.

[0042] The carbon sequestration content after carbonation refers to the amount of carbon sealed in the waste concrete raw material sample during the carbonation process. The ratio of the mass of the waste concrete raw material sample to the mass of the waste concrete raw material sample. The carbon sequestration amount after the carbonation reaction is a core indicator for evaluating the effectiveness of the carbon sequestration process, and is jointly influenced by the chemical composition of the waste concrete raw material sample, especially the mass percentage content of calcium oxide, and the carbonation process parameters.

[0043] The theoretical maximum carbon sequestration capacity refers to the theoretically achievable amount of carbon that can be sealed, calculated using stoichiometry based on the percentage of calcium oxide and other alkaline oxides in waste concrete raw material samples. The maximum ratio of the mass of the waste concrete raw material sample to its mass. The theoretical maximum carbon sequestration capacity provides a benchmark for evaluating actual carbon sequestration efficiency.

[0044] The initial moisture content refers to the ratio of the mass of water in the waste concrete raw material sample to the total mass of the waste concrete raw material sample. The initial moisture content plays a dual role in the carbonation reaction; a suitable initial moisture content can promote… dissolution and The diffusion of the gas does not hinder the transport of gas in the pores. Both excessively high and excessively low initial moisture content will reduce carbonization efficiency.

[0045] The X-ray diffraction analyzer refers to an analytical device that utilizes the diffraction phenomenon of X-rays in crystals to determine the crystal structure and phase composition of a substance by measuring the position and intensity of diffraction peaks. The X-ray diffraction analyzer is used for the quantitative detection of substances generated in the carbonization reaction. The content provides accurate data support for evaluating the carbon sequestration effect.

[0046] The amount of calcium carbonate generated refers to the amount generated during the carbonation reaction of the waste concrete raw material sample. The ratio of the mass of the waste concrete raw material sample to its mass. The amount of calcium carbonate generated directly reflects the degree of completion of the carbonation reaction and... The actual sealing effect.

[0047] The specific surface area increase ratio refers to the ratio of the specific surface area of ​​the waste concrete raw material sample after pre-crushing treatment to the initial specific surface area. The specific surface area increase ratio reflects the effect of pre-crushing treatment on increasing the surface area of ​​the waste concrete raw material sample; a larger specific surface area increase ratio indicates... The larger the contact area with alkaline substances, the better it is to increase the carbonization reaction rate.

[0048] The specific implementation methods of the above steps are described in detail below.

[0049] The specific implementation of step S01 is as follows: First, the surface of the waste concrete raw material sample is cleaned to remove attached impurities and contaminants. Then, the waste concrete raw material sample is ground into powder using a grinder. The ground powder is sieved through a 200-mesh sieve to ensure particle uniformity. The sieved powder sample is subjected to quantitative chemical composition analysis using X-ray fluorescence spectrometry. By measuring the fluorescence intensity of characteristic wavelengths, eight chemical component data are calculated, including the mass percentage content of silicon dioxide, calcium oxide, aluminum oxide, ferric oxide, potassium oxide, magnesium oxide, sulfur trioxide, and sodium oxide. At the same time, the initial porosity of the waste concrete raw material sample is determined by mercury intrusion porosimetry, and the initial specific surface area is determined by nitrogen adsorption. The quantitative chemical composition analysis provides raw material characteristic input data for the subsequent intelligent control model of carbonization parameters, and the initial specific surface area determination provides a benchmark reference value for evaluating the pre-crushing treatment effect.

[0050] The specific implementation of step S02 is as follows: The waste concrete raw material sample from step S01 is fed into a jaw crusher for primary crushing. After crushing, particles with a diameter in the range of 5 to 15 mm are screened out by a vibrating screener. The specific surface area of ​​the screened particles after crushing is measured using a laser particle size analyzer. The laser particle size analyzer measures the particle size distribution and calculates the specific surface area based on the principle of laser scattering. Then, the ratio of the specific surface area after crushing to the initial specific surface area measured in step S01 is calculated to obtain the specific surface area increase ratio. The specific surface area increase ratio is used to evaluate the effect of pre-crushing treatment on... The optimization effect of the diffusion path, with a specific surface area increase ratio typically between 2 and 5, indicates that the pre-crushing treatment has achieved the expected results. The particle size is controlled within the range of 5 to 15 mm, which ensures sufficient specific surface area while avoiding the problem of gas diffusion channel blockage caused by excessively fine particles.

[0051] The specific implementation of step S03 is as follows: The crushed waste concrete raw material sample from step S02 is evenly spread on a porous tray of the carbonization reactor. The initial moisture content of the waste concrete raw material sample is adjusted to within the range of 8% to 12% using a spray humidification system. The initial moisture content is determined by drying and weighing. Then, the carbonization reactor is sealed and a mixed gas with an initial carbon dioxide concentration of 15% to 25% is introduced, with the remainder being air. The reaction temperature is controlled and maintained at 20% to 40℃ by a heating jacket on the outer wall of the carbonization reactor, and the reaction pressure is controlled at 0.1% to 0.3MPa by a gas pressure regulating valve. Temperature sensors, humidity sensors, and pressure sensors are arranged inside the carbonization reactor. A concentration sensor automatically collects real-time reaction temperature, real-time reaction humidity, real-time reaction pressure, and real-time carbon dioxide concentration every 30 minutes and transmits them to the data acquisition system. The parameter acquisition frequency of 30 minutes ensures real-time monitoring of the carbonization reaction process while avoiding data redundancy caused by excessively high acquisition frequency. The initial process parameters are set based on preliminary matching of the chemical composition characteristics of the waste concrete raw material sample.

[0052] The specific implementation of step S04 is as follows: The real-time reaction temperature, real-time reaction humidity, real-time reaction pressure, and real-time carbon dioxide concentration collected in step S03, along with the 12 parameters obtained in step S01 (silicon dioxide, calcium oxide, aluminum oxide, ferric oxide, potassium oxide, magnesium oxide, sulfur trioxide, and sodium oxide), are combined to form an input vector. This input vector, after normalization, is input into the intelligent carbonization parameter control model. The normalization process uses a minimum-maximum normalization method to map each parameter to the range of 0 to 1. The intelligent carbonization parameter control model performs forward propagation calculations sequentially through the input layer, three hidden layers, and the output layer, ultimately outputting adjustment values ​​for carbon dioxide concentration, reaction temperature, reaction pressure, and moisture content. Based on a nonlinear mapping relationship learned from historical high-quality samples, the intelligent carbonization parameter control model achieves intelligent decision-making from raw material characteristics and real-time operating conditions to optimal process parameters. The adjustment values ​​output by the model are incremental values ​​relative to the current process parameters, rather than absolute values.

[0053] The specific implementation of step S05 is as follows: Based on the carbon dioxide concentration adjustment value, reaction temperature adjustment value, reaction pressure adjustment value, and moisture content adjustment value output in step S04, the gas mixing ratio, heating jacket power, pressure regulating valve opening, and spray humidification system flow rate are adjusted through the automatic control system of the carbonization reactor to achieve dynamic adjustment of the real-time carbon dioxide concentration, real-time reaction temperature, real-time reaction pressure, and real-time reaction humidity in the carbonization reactor. The automatic control system adopts a proportional-integral-derivative control algorithm to achieve rapid response and precise adjustment. After the adjustment is completed, the actual carbon dioxide concentration, actual reaction temperature, actual reaction pressure, and actual moisture content are monitored in real time by sensors, and the deviation between the actual parameters and the target adjustment value is calculated. The adjustment is considered complete when the deviation between the actual carbon dioxide concentration and the carbon dioxide concentration adjustment value does not exceed 2%, the deviation between the actual reaction temperature and the reaction temperature adjustment value does not exceed 3°C, the deviation between the actual reaction pressure and the reaction pressure adjustment value does not exceed 0.02 MPa, and the deviation between the actual moisture content and the moisture content adjustment value does not exceed 1%. The precise deviation control ensures that the carbonization reaction is carried out under optimal process conditions.

[0054] The specific implementation of step S06 is as follows: After the carbonization reaction has proceeded for 120 to 180 minutes, the reaction is paused and the carbonization reactor is opened. Five locations are randomly selected from the waste concrete raw material sample. The carbonization depth after the reaction is determined at each sampling location using the phenolphthalein solution spraying method. Phenolphthalein solution appears purple-red in an alkaline environment but remains colorless in the carbonized area. The carbonization depth after the reaction is obtained by measuring the depth of the colorless area. The average value of the measurements at the five locations is taken as the final carbonization depth after the reaction. Simultaneously, the carbon fixation amount after the reaction is determined using the gravimetric method. The gravimetric method measures the mass change of the waste concrete raw material sample before and after the carbonization reaction, combined with... The amount of carbon fixed after carbonization is calculated by molecular weight. Then, it is determined whether the carbonization depth after carbonization is greater than 20 mm and whether the amount of carbon fixed after carbonization reaches more than 80% of the theoretical maximum amount of carbon fixed. The 20 mm carbonization depth threshold and the 80% carbon fixed amount threshold are determined based on the minimum requirements for carbon fixation effect in industrial applications. If the conditions are met, the process is optimized in step S07. If the conditions are not met, the process returns to step S03 and the initial carbon dioxide concentration is increased by 5% and the reaction pressure is increased by 0.05 MPa to enhance the driving force of carbonization reaction.

[0055] The specific implementation of step S07 is as follows: the waste concrete raw material sample taken in step S06 is ground into powder and then placed into the sample cell of an X-ray diffraction analyzer. The X-ray diffraction analyzer emits X-rays to irradiate the sample and records the diffraction pattern. The sample is then analyzed... The intensity of characteristic diffraction peaks and the amount of calcium carbonate formed are used to quantitatively calculate the calcium carbonate formation based on a standard curve. This calcium carbonate formation reflects... and The degree of completion of the reaction is determined, and then the carbonization depth, the amount of carbon fixed after the carbonization reaction, and the amount of calcium carbonate generated are recorded and stored as process optimization result data. The process optimization result data is used to evaluate the actual effect of the carbon fixation process and to provide feedback data for the continuous optimization of the intelligent control model of subsequent carbonization parameters.

[0056] It should be noted that one of the key technical ideas of this invention is the establishment of an intelligent control mechanism based on real-time sensing of chemical components and dynamic matching of process parameters. This mechanism achieves a nonlinear mapping from the chemical composition of waste concrete raw material samples to the optimal carbonation process parameters through an intelligent carbonation parameter control model. This overcomes the technical bottleneck of traditional fixed-parameter processes being unable to adapt to fluctuations in raw material composition. The intelligent control mechanism collects temperature, humidity, pressure, and other parameters in real time during the carbonation reaction process. By combining concentration data with information on the chemical composition of raw materials, process parameters can be continuously optimized during the carbonization reaction. This effectively solves the adverse effects of differences in CaO content and pore structure caused by the diverse sources of waste concrete on carbon fixation efficiency, enabling the carbon fixation process to adapt to different raw material characteristics and achieve stable high carbon fixation efficiency.

[0057] The second key technical idea of ​​this invention is to train an intelligent carbonization parameter control model using a parameter regularization framework based on adaptive weight decay. This framework achieves a dynamic balance between the model's generalization ability and fitting ability by dynamically adjusting the regularization strength of each layer of the neural network. Compared with the traditional training method with fixed regularization strength, this framework can identify key layers in the network that contribute significantly to the optimization of carbon fixation process parameters and apply weaker constraints to them to fully learn the complex correlation between raw material characteristics and process parameters. At the same time, it applies stronger constraints to layers with low contribution to suppress overfitting. This differentiated regularization strategy enables the intelligent carbonization parameter control model to output reasonable process parameter adjustment values ​​when facing new raw material components that are not present in the training dataset, significantly improving the adaptability and robustness of the carbon fixation process to changes in raw materials.

[0058] The third key technical idea of ​​this invention is to design a closed-loop control strategy with dual threshold judgment and iterative optimization for carbonization depth and carbon fixation amount. In step S06, it is determined whether the carbonization depth and carbon fixation amount after the carbonization reaction simultaneously meet the process requirements. If not, the initial carbon dioxide concentration and reaction pressure are automatically adjusted and the carbonization reaction is repeated. Compared with the traditional one-time fixed parameter carbonization method, the closed-loop control strategy can dynamically correct the process parameters according to the actual carbonization effect, avoiding the loss of material strength caused by substandard carbon fixation effect or excessive carbonization due to improper initial parameter setting. The iterative optimization mechanism ensures that the carbon fixation process reaches the optimal balance point between carbonization depth and carbon fixation amount.

[0059] The synergistic effect of the above three key technical approaches is reflected in the following aspects: the intelligent control mechanism of real-time chemical component sensing and dynamic matching of process parameters provides accurate parameter decision-making basis for carbon fixation process; the parameter regularization framework based on adaptive weight decay ensures the stability and reliability of intelligent control mechanism in the face of raw material fluctuations; and the closed-loop control strategy of dual threshold judgment and iterative optimization of carbonization depth and carbon fixation amount verifies and corrects the intelligent control output at the actual execution level. Together, the three constitute a complete closed-loop system from raw material characteristic identification to intelligent parameter decision-making and effect verification feedback. Compared with the traditional carbon fixation process that relies on experience-based parameter adjustment, the synergistic mechanism realizes the full-process intelligence and adaptability of carbon fixation process, fundamentally solving the process optimization problem caused by the large differences in waste concrete components and the complex carbonization reaction mechanism. This enables the carbon fixation process to achieve both high carbon fixation amount and uniformity of carbonization depth, significantly improving the industrial application value of waste concrete carbon fixation technology.

[0060] It should be noted that this invention also solves the following technical problem: Traditional neural network training methods with fixed regularization strength are prone to underfitting or overfitting when faced with training data containing large fluctuations in the chemical composition and significant differences in pore structure of waste concrete, resulting in insufficient generalization ability of the model for unknown raw materials. This invention introduces a parameter regularization framework based on adaptive weight decay. In each iteration, the importance index of each layer is calculated based on the gradient L2 norm and parameter update magnitude, and the decay coefficient of each layer is dynamically adjusted. A weaker regularization constraint is applied to network layers with high contribution to fully preserve the complex nonlinear mapping relationship they have learned, while a stronger regularization constraint is applied to network layers with low contribution to prevent overfitting to noise and abnormal fluctuations in the training data. This enables the model to accurately predict known working conditions and effectively cope with real-time changes in raw material properties, improving the adaptability and reliability of the intelligent control model in practical applications.

[0061] Specifically, the principle of this invention is as follows: The solution to the problem of unstable carbon fixation efficiency lies in establishing an intelligent mapping mechanism from the chemical composition of raw materials to optimal process parameters. The content ratio of eight chemical components, including silica and calcium oxide, in waste concrete directly affects the reactivity and pore structure characteristics of alkaline substances. Meanwhile, four process parameters—carbon dioxide concentration, reaction temperature, reaction pressure, and moisture content—jointly determine the diffusion rate of carbon dioxide and the carbonization reaction process. The optimal process parameter combinations vary significantly under different chemical component ratios. This invention collects a large amount of carbonization experimental data covering different combinations of chemical components and process parameters, and selects high-quality samples with a carbonization depth greater than 20 mm and a carbon fixation amount exceeding 80% of the theoretical maximum to construct a training dataset. This enables the intelligent control model to learn the inherent correlation between chemical components and optimal parameters. In practical applications, the model continuously outputs parameter adjustment values ​​and dynamically controls the carbonization reactor based on the current chemical composition detection results of the raw materials and real-time parameters collected every 30 minutes during the carbonization process. This ensures that the process parameters always adapt to changes in raw material characteristics and reaction state, guaranteeing stable carbon fixation effects for different batches of waste concrete.

[0062] The following provides a specific embodiment 1 of the present invention, and the specific implementation of each step in this embodiment 1 is described in detail below.

[0063] The specific implementation methods of steps S01 and S03 are the same as those described above, and will not be repeated in detail here.

[0064] The specific implementation of step S02 involves pre-crushing the waste concrete raw material sample from step S01. The particle size of the crushed particles is controlled within the range of 5 to 15 millimeters. The specific surface area after crushing is measured using a laser particle size analyzer, and the ratio of the specific surface area after crushing to the initial specific surface area is calculated to obtain the specific surface area increase ratio. The formula for calculating the specific surface area increase ratio is as follows:

[0065] ;

[0066] In the formula, The ratio of the increase in specific surface area is dimensionless. Specific surface area after crushing, in units of , This is the initial specific surface area, in units of , For reference specific surface area, the default value is 1. The parameter acquisition method is as follows: and All measurements were obtained using a laser particle size analyzer. The measurement was obtained in step S01. The specific surface area increase ratio was measured after the crushing treatment in step S02. This ratio reflects the effect of pre-crushing treatment on increasing the surface area of ​​the waste concrete raw material sample; a larger specific surface area increase ratio indicates... The larger the contact area with alkaline substances, the better it is to increase the carbonization reaction rate.

[0067] The specific implementation of step S04 involves inputting the real-time reaction temperature, real-time reaction humidity, real-time reaction pressure, and real-time carbon dioxide concentration collected in step S03, along with the mass percentages of silicon dioxide, calcium oxide, aluminum oxide, ferric oxide, potassium oxide, magnesium oxide, sulfur trioxide, and sodium oxide obtained in step S01, into the intelligent control model for carbonization parameters. This intelligent control model outputs adjusted values ​​for carbon dioxide concentration, reaction temperature, reaction pressure, and moisture content. Each parameter in the input vector needs to be normalized, and the normalization formula is as follows:

[0068] ;

[0069] ;

[0070] ;

[0071] ;

[0072] In the formula, The normalized real-time reaction temperature is dimensionless. The real-time reaction temperature is expressed in °C. This is the lower limit of the reaction temperature, which is 15℃ by default. This is the upper limit of the reaction temperature, which is 45℃ by default; The normalized real-time humidity is dimensionless. To reflect humidity in real time, it is expressed as a percentage by mass. This is the lower limit of moisture content, defaulted to 5%. This represents the upper limit of moisture content, with a default value of 15%. The normalized real-time reaction pressure is dimensionless. To provide real-time pressure feedback, the unit is MPa. This is the lower limit of the reaction pressure, with a default value of 0.05 MPa. This is the upper limit of the reaction pressure, with a default value of 0.35 MPa. This represents the normalized real-time carbon dioxide concentration, dimensionless. Real-time carbon dioxide concentration, expressed as a volume percentage. This is the lower limit for carbon dioxide concentration, with a default value of 10%. This represents the upper limit of carbon dioxide concentration, with a default value of 30%. The input vector of the intelligent control model for carbonization parameters is expressed as follows:

[0073] ;

[0074] In the formula, For the input vector, This refers to the percentage content of silicon dioxide by mass. This represents the percentage content of calcium oxide by mass. This refers to the percentage content of aluminum oxide by mass. This refers to the percentage content of ferric oxide by mass. This represents the percentage content of potassium oxide by mass. This refers to the percentage content of magnesium oxide by mass. This refers to the percentage content of sulfur trioxide by mass. The values ​​represent the mass percentage content of sodium oxide; all the contents of the above chemical components are dimensionless mass percentage values. The output vector of the intelligent control model for carbonization parameters is expressed as follows:

[0075] ;

[0076] In the formula, For the output vector, The carbon dioxide concentration adjustment value is expressed as a volume percentage. This is the reaction temperature adjustment value, in °C. This is the reaction pressure adjustment value, in MPa. This is the moisture content adjustment value, expressed as a percentage by mass.

[0077] The specific implementation method of step S05 is the same as described above, and will not be repeated in detail here.

[0078] The specific implementation of step S06 is as follows: After the carbonation reaction has proceeded for 120 to 180 minutes, a sample of the waste concrete raw material is taken out, and the carbonation depth and carbon sequestration amount after the carbonation reaction are measured. It is determined whether the carbonation depth after the carbonation reaction is greater than 20 mm and whether the carbon sequestration amount after the carbonation reaction reaches more than 80% of the theoretical maximum carbon sequestration amount. The formula for calculating the theoretical maximum carbon sequestration amount is as follows:

[0079] ;

[0080] In the formula, The theoretical maximum carbon sequestration capacity is dimensionless. for Molar mass, numerical value is 44 g / mol. for The molar mass is 56 g / mol. for Molar mass, with a value of 40 g / mol. This represents the percentage content of calcium oxide by mass. This represents the percentage content of magnesium oxide by mass. The parameter is obtained as follows: and All were obtained through chemical component detection in step S01. The carbon fixation content after the carbonization reaction... The calculation formula is expressed as follows:

[0081] ;

[0082] In the formula, The carbon content after the carbonization reaction is dimensionless. The mass of the waste concrete raw material sample after carbonation reaction is expressed in kg. The mass of the waste concrete raw material sample before carbonation is given, in kg. For reference weight, the default value is 1 kg. The parameter can be obtained as follows: and All measurements were obtained by weighing using an electronic balance.

[0083] The specific implementation method of step S07 is the same as described above, and will not be repeated in detail here.

[0084] The formula for calculating the hierarchical importance index during the training process of the intelligent control model for carbonization parameters is as follows:

[0085] ;

[0086] In the formula, For the first The hierarchy importance index is dimensionless. For the current level index, For the first The L2 norm of the layer gradient vector, For the first Layer gradient vector, This is the gradient reference value, which defaults to 1. This represents the total number of network layers, with a default value of 3. For the first The L2 norm of the layer parameter update vector. For the first Layer parameter update vector, Update the reference value for the parameter; the default value is 1. This is a hierarchical traversal index, with values ​​ranging from 1 to... , For the first The L2 norm of the layer gradient vector, For the first The L2 norm of the layer parameter update vector. The parameter is obtained as follows: It is calculated using the backpropagation algorithm during each iteration. The hierarchy importance index is calculated based on the gradient descent algorithm. It comprehensively considers the gradient flow intensity and parameter variation amplitude, and is used to identify network hierarchies that contribute significantly to the optimization of carbon fixation process parameters.

[0087] In the aforementioned parameter regularization framework based on adaptive weight decay, the loss function is expressed as follows:

[0088] ;

[0089] In the formula, The total loss function is dimensionless. For mean square error loss, This is a reference value for loss, with a default value of 1. For the first Layer attenuation coefficient, dimensionless. For the first The parameter vector of the layer, For the first The squared L2 norm of the layer parameter vector, This is the baseline value for parameter normalization; the default value is 1. This represents the total number of network layers, which defaults to 3. The squared norm of the parameter vectors is normalized, as detailed below:

[0090] ;

[0091] In the formula, For the first Total number of layer parameters For the first Layer The current training values ​​of each parameter. For the first Layer Initial values ​​for each parameter. The parameter is the traversal index, with a value range from 1 to... The parameter acquisition method is as follows: Recorded during model initialization. The hierarchical decay coefficient is updated in real time during training. The calculation formula is expressed as follows:

[0092] ;

[0093] In the formula, The basic attenuation coefficient is dimensionless and has an empirical value of 0.01. For the first The hierarchy importance index is dimensionless. The parameter is obtained as follows: The calculation formula is: ,in This represents the batch sample size; an empirical value is 32. For the first The model's predicted output vector for each sample. For the first The true label vector of each sample This is the sample traversal index, with values ​​ranging from 1 to... The loss function applies regularization constraints to the parameters of each layer through a weight decay term. The layer decay coefficient is adaptively adjusted according to the layer importance index, so that layers with high contribution receive smaller decay coefficients to retain more parameter information, while layers with low contribution receive larger decay coefficients to enhance the regularization constraint. This avoids underfitting or overfitting problems caused by fixed regularization strength, and improves the generalization ability and prediction accuracy of the intelligent carbonation parameter control model when facing waste concrete raw materials with large fluctuations in chemical composition.

[0094] To better understand and implement this invention, the following is a specific application scenario of this invention, Example 2:

[0095] A sample of waste concrete raw materials from a building demolition site, comprising three types: waste concrete blocks, waste aerated concrete blocks, and waste concrete mortar, was used by the technical team to perform carbon sequestration treatment on the waste concrete raw material sample using the concrete carbon sequestration process parameter optimization method described in this invention. The specific implementation process is as follows.

[0096] The technical team first conducted chemical composition analysis on the collected waste concrete raw material samples. The waste concrete blocks were ground into powder and sieved through a 200-mesh sieve. Then, X-ray fluorescence spectrometry was used for quantitative analysis of chemical composition. The test results are shown in Table 1.

[0097] Table 1. Results of Chemical Composition Analysis of Waste Concrete Blocks

[0098]

[0099] Simultaneously, the initial porosity of the waste concrete block was determined to be 18.6% using mercury intrusion porosimetry, and the initial specific surface area was determined to be 0.45% using nitrogen adsorption. Based on the calcium oxide mass percentage content of 21.08%, the theoretical maximum carbon fixation capacity is calculated to be 20.22%.

[0100] The technical team fed waste concrete blocks into a jaw crusher for pre-crushing. After crushing, particles with a diameter ranging from 5 to 15 millimeters were screened out using a vibrating screen. The specific surface area after crushing was measured to be 1.83 using a laser particle size analyzer. The calculated specific surface area increase ratio was 4.07. This ratio indicates that the pre-crushing treatment significantly increased the specific surface area of ​​the waste concrete blocks, which is beneficial for subsequent processing. Diffusion provides ample contact area.

[0101] The technical team evenly spread the crushed waste concrete blocks onto the porous tray of the carbonization reactor, adjusted the initial moisture content to 10% using a spray humidification system, then sealed the carbonization reactor and introduced a mixed gas with an initial carbon dioxide concentration of 20%. The reaction temperature was controlled at 30℃, and the reaction pressure was set at 0.2 MPa. Figure 2 The diagram shows the structure of a carbonization reactor, including temperature sensors, humidity sensors, pressure sensors, and [other components]. The concentration sensor automatically collects data every 30 minutes. The initial real-time reaction temperature is 30℃, the real-time reaction humidity is 10%, the real-time reaction pressure is 0.2MPa, and the real-time carbon dioxide concentration is 20%.

[0102] The technical team combined the collected real-time reaction temperature (30℃), real-time reaction humidity (10%), real-time reaction pressure (0.2MPa), and real-time carbon dioxide concentration (20%) with eight mass percentage data from the chemical component detection results to form a 12-dimensional input vector. After normalization, this vector was input into the intelligent control model for carbonization parameters, such as... Figure 3The diagram shows the network structure of the intelligent control model for carbonation parameters. This model calculates and outputs the following adjustment values ​​through forward propagation: carbon dioxide concentration adjustment value is 22%, reaction temperature adjustment value is 32℃, reaction pressure adjustment value is 0.22MPa, and moisture content adjustment value is 9.5%. These output adjustment values ​​indicate that the current process parameters need to be appropriately increased. Concentration and reaction pressure were adjusted to enhance the driving force of the carbonization reaction, while the water content was slightly reduced to improve gas diffusion conditions.

[0103] Based on the adjustment values ​​output by the intelligent control model of carbonization parameters, the technical team adjusted the gas mixing ratio through the automatic control system of the carbonization reactor to increase the real-time carbon dioxide concentration from 20% to 22%, adjusted the heating jacket power to increase the real-time reaction temperature from 30℃ to 32℃, adjusted the pressure regulating valve opening to increase the real-time reaction pressure from 0.2MPa to 0.22MPa, and adjusted the flow rate of the spray humidification system to decrease the real-time reaction humidity from 10% to 9.5%. The automatic control system, using a proportional-integral-derivative control algorithm, completed the parameter adjustment within 5 minutes. After adjustment, the actual carbon dioxide concentration was measured to be 21.8%, the actual reaction temperature to be 31.7℃, the actual reaction pressure to be 0.219MPa, and the actual moisture content to be 9.6%. The deviations of each actual parameter from the target adjustment value all met the control requirements.

[0104] like Figure 4 The figure shows the dynamic adjustment curve of process parameters during the carbonization reaction. During the carbonization reaction, the intelligent control model of carbonization parameters outputs new adjustment values ​​every 30 minutes based on the real-time data collected. The automatic control system continuously optimizes the process parameters to adapt to the dynamic changes of the carbonization reaction. After 150 minutes of carbonization reaction, the technical team paused the reaction and opened the carbonization reactor. Five locations were randomly selected from the waste concrete block for sampling. The carbonization depth after the carbonization reaction was determined by the phenolphthalein solution spraying method. The measured values ​​at the five locations were 23.5 mm, 24.2 mm, 22.8 mm, 23.9 mm, and 24.6 mm, respectively. The average carbonization depth after the carbonization reaction was calculated to be 23.8 mm. The carbon fixation amount after the carbonization reaction was determined by the gravimetric method to be 17.4%. The carbon fixation amount after the carbonization reaction reached 86% of the theoretical maximum carbon fixation amount of 20.22%, which meets the process requirements of a carbonization depth greater than 20 mm and a carbon fixation amount reaching more than 80% of the theoretical maximum carbon fixation amount.

[0105] The technical team ground the extracted waste concrete blocks into powder and placed them into the sample cell of an X-ray diffraction analyzer for testing. The X-ray diffraction pattern showed... The characteristic diffraction peaks were significantly enhanced. Quantitative calculation using a standard curve showed that the calcium carbonate formation was 15.8%, which basically corresponds to the 17.4% carbon fixation after the carbonation reaction, verifying the effectiveness of the carbon fixation. Figure 5The image shows a comparison of X-ray diffraction patterns before and after the carbonization reaction. After the carbonization reaction... The characteristic peak intensity was significantly higher than before the carbonization reaction, indicating that... and A complete carbonization reaction occurred.

[0106] The technical team continued to conduct the same carbon sequestration experiments on waste aerated concrete blocks and waste concrete mortar. The chemical composition analysis results of the waste aerated concrete blocks showed... The mass percentage content was 30.82%, the theoretical maximum carbon fixation was 27.06%, and the specific surface area increase ratio after pre-crushing was 3.58. The intelligent carbonation parameter control model output process parameters based on the chemical composition characteristics of the waste aerated concrete blocks: initial carbon dioxide concentration of 25%, reaction temperature of 28℃, reaction pressure of 0.25MPa, and moisture content of 11%. After 180 minutes of carbonation, the carbonation depth was measured to be 26.3 mm, the carbon fixation was 23.1%, reaching 85% of the theoretical maximum carbon fixation, and the calcium carbonate formation was 21.2%. The chemical composition test results of the waste concrete mortar showed... The mass percentage content was 26.19%, the theoretical maximum carbon fixation was 26.58%, and the specific surface area increase ratio after pre-crushing treatment was 4.21. The intelligent control model of carbonization parameters output the initial carbon dioxide concentration of 23%, the reaction temperature of 31℃, the reaction pressure of 0.23MPa, and the water content of 10.5%. After 165 minutes of carbonization reaction, the carbonization depth was measured to be 25.1 mm, the carbon fixation was 22.4%, reaching 84% of the theoretical maximum carbon fixation, and the calcium carbonate formation was 20.6%.

[0107] The technical team summarized and analyzed the carbon sequestration results of three different types of waste concrete raw material samples. They found that the intelligent control model for carbonation parameters can automatically adjust the process parameters according to the differences in chemical composition of different waste concrete raw material samples, so that the carbon sequestration efficiency is maintained at a high level. Table 2 shows the comparison of the carbon sequestration results of the three types of waste concrete raw material samples.

[0108] Table 2 Comparison of carbon sequestration results for three types of waste concrete raw material samples

[0109]

[0110] The technological advancement of this invention compared to traditional fixed-parameter carbon fixation processes lies in the fact that traditional carbon fixation processes use a uniform... The concentration, reaction temperature, reaction pressure, and moisture content parameters were applied to all waste concrete raw material samples, but this approach failed to accommodate the differences in chemical composition and pore structure among waste concrete raw material samples from different sources, resulting in high... The raw materials with low content were not fully carbonized due to insufficient process parameters. The raw materials with excessive carbonization content suffer intensity loss due to overly stringent process parameters. This invention addresses this issue by establishing an intelligent carbonization parameter control model to achieve intelligent mapping from the chemical composition of waste concrete raw material samples to optimal process parameters. This intelligent mapping mechanism can identify the carbon sequestration potential of each waste concrete raw material sample and match it with corresponding process intensities, ensuring that waste concrete raw material samples with different chemical compositions can undergo carbonization reactions under optimal process conditions. The parameter regularization framework based on adaptive weight decay employed in this invention dynamically adjusts the regularization intensity of each layer of the neural network, enabling the intelligent carbonization parameter control model to learn from high-quality sample features in the training dataset. While maintaining the ability to generalize to unknown component raw materials, the generalization ability ensures that the model can still output reasonable process parameters when facing waste concrete raw material samples with fluctuating components in actual applications. The dual threshold judgment and iterative optimization closed-loop control strategy for carbonation depth and carbon fixation amount designed in this invention breaks through the limitation of traditional one-time fixed parameter processing methods that cannot correct errors by monitoring the carbonation effect in real time and adjusting the process parameters according to the monitoring results. The closed-loop control strategy ensures that the carbon fixation process achieves a dynamic balance between carbonation depth and carbon fixation amount, avoiding the problems of excessive carbonation caused by unilaterally pursuing high carbon fixation amount or insufficient carbon fixation amount caused by unilaterally pursuing shallow carbonation.

[0111] It should be noted that the variables involved in this invention are explained in detail in Tables 3 and 4.

[0112] Table 3. Variable Explanation Table (Part 1)

[0113]

[0114] Table 4. Variable Explanation Table (Part Two)

[0115]

[0116] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for optimizing concrete carbon sequestration process parameters, characterized in that, Includes the following steps: Waste concrete raw material samples were collected and their chemical composition was analyzed to obtain the mass percentages of silica, calcium oxide, aluminum oxide, ferric oxide, potassium oxide, magnesium oxide, sulfur trioxide, and sodium oxide. Simultaneously, the initial porosity and initial specific surface area of ​​the waste concrete raw material samples were measured. The waste concrete raw material samples were pre-crushed, and the specific surface area after crushing was measured. The ratio of the crushed specific surface area to the initial specific surface area was calculated to obtain the specific surface area increase ratio. The crushed waste concrete raw material samples were placed in a carbonization reactor, and... The gas was monitored, and real-time reaction temperature, humidity, pressure, and humidity were collected every 30 minutes during the carbonization reaction. Concentration; real-time reaction temperature, real-time reaction humidity, real-time reaction pressure, real-time Concentration and chemical composition detection data are input into the intelligent carbonization parameter control model, and the intelligent carbonization parameter control model outputs... Concentration adjustment value, reaction temperature adjustment value, reaction pressure adjustment value, and moisture content adjustment value; based on the output adjustment values, adjust the real-time values ​​in the carbonization reactor. The concentration, real-time reaction temperature, real-time reaction pressure, and real-time reaction humidity are dynamically adjusted. Waste concrete raw material samples are taken out, and the carbonation depth and carbon fixation amount after carbonation reaction are measured to determine whether the carbonation depth after carbonation reaction is greater than 20 mm and whether the carbon fixation amount after carbonation reaction reaches more than 80% of the theoretical maximum carbon fixation amount. Waste concrete raw material samples are tested by X-ray diffraction analysis to obtain the amount of calcium carbonate generated. The carbonation depth, carbon fixation amount, and calcium carbonate generation amount after carbonation reaction are recorded as process optimization result data.

2. The method for optimizing concrete carbon sequestration process parameters according to claim 1, characterized in that, In the pre-crushing process, the particle size of the crushed particles is controlled within the range of 5 to 15 mm, and the specific surface area after crushing is measured by a laser particle size analyzer.

3. The method for optimizing concrete carbon sequestration process parameters according to claim 2, characterized in that, The step of placing the crushed waste concrete raw material sample into the carbonization reactor specifically involves setting the initial moisture content to 8 to 12%. The concentration is 15% to 25%, the reaction temperature is controlled at 20 to 40°C, and the reaction pressure is set at 0.1 to 0.3 MPa.

4. The method for optimizing concrete carbon sequestration process parameters according to claim 3, characterized in that, The real-time adjustment of the carbonization reactor based on the output adjustment value. In the step of dynamically adjusting the concentration, real-time reaction temperature, real-time reaction pressure, and real-time reaction humidity, the actual value after adjustment... Concentration and The deviation of the concentration adjustment value shall not exceed 2%.

5. The method for optimizing concrete carbon sequestration process parameters according to claim 4, characterized in that, In the dynamic adjustment step, the deviation between the adjusted actual reaction temperature and the adjusted reaction temperature value shall not exceed 3°C, the deviation between the adjusted actual reaction pressure and the adjusted reaction pressure value shall not exceed 0.02 MPa, and the deviation between the adjusted actual moisture content and the adjusted moisture content value shall not exceed 1%.

6. The method for optimizing concrete carbon sequestration process parameters according to claim 5, characterized in that, The step of taking out the waste concrete raw material sample specifically involves taking out the waste concrete raw material sample 120 to 180 minutes after the carbonization reaction has been carried out.

7. The method for optimizing concrete carbon sequestration process parameters according to claim 6, characterized in that, In the step of determining whether the carbonization depth after the carbonization reaction is greater than 20 mm and whether the carbon fixation amount after the carbonization reaction reaches more than 80% of the theoretical maximum carbon fixation amount, if the carbonization depth after the carbonization reaction is not greater than 20 mm or the carbon fixation amount after the carbonization reaction does not reach 80% of the theoretical maximum carbon fixation amount, then the initial... The carbonization reaction was repeated after increasing the concentration by 5% and increasing the reaction pressure by 0.05 MPa.

8. The method for optimizing concrete carbon sequestration process parameters according to claim 7, characterized in that, The input layer of the intelligent control model for carbonization parameters contains 12 neurons, corresponding to the mass percentages of silicon dioxide, calcium oxide, aluminum oxide, ferric oxide, potassium oxide, magnesium oxide, sulfur trioxide, sodium oxide, real-time reaction temperature, real-time reaction humidity, real-time reaction pressure, and real-time... concentration.

9. The method for optimizing concrete carbon sequestration process parameters according to claim 8, characterized in that, The hidden layer of the intelligent control model for carbonization parameters is a three-layer fully connected layer. The first hidden layer contains 64 neurons, the second hidden layer contains 32 neurons, and the third hidden layer contains 16 neurons. The activation function of each hidden layer is a modified linear unit function.

10. The method for optimizing concrete carbon sequestration process parameters according to claim 9, characterized in that, The output layer of the intelligent carbonization parameter control model contains four neurons, each corresponding to... Concentration adjustment value, reaction temperature adjustment value, reaction pressure adjustment value, and moisture content adjustment value.

Images