Artificial Intelligence-Based Method and System for Concrete Mix Proportion Optimization Design

By optimizing the concrete mix proportion using artificial intelligence, the problems of drastic short-term slump loss and abrupt setting in small-scale construction have been solved, enabling dynamic adaptive adjustment of the mix proportion and improving construction stability and early performance consistency.

CN123072571APending Publication Date: 2026-11-10BEIJING MUNICIPAL ENG RES INST +3
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610818660.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-06-08
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing concrete mix design methods are insufficient to cope with the coupled effects of aggregate moisture content fluctuations, admixture lag effects, and early hydration heat release mutations in small-scale, short-term construction, leading to a sharp loss of slump and simultaneous occurrence of microcracks, resulting in insufficient optimization accuracy and stability.

Method used

The AI-based concrete mix design optimization method generates a set of candidate mix proportions by acquiring raw material information and environmental parameters, performs virtual mix design simulations, and dynamically adjusts the water-cement ratio and admixture dosage based on on-site feedback to achieve real-time optimization of the mix proportions.

Benefits of technology

It significantly reduces the risk of fluid instability and early defects, ensures construction stability and consistency of early structural performance, and achieves dynamic adaptive adjustment of mix proportions through real-time feedback correction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123072571A_ABST
    Figure CN123072571A_ABST
Patent Text Reader

Abstract

This invention discloses an artificial intelligence-based method and system for optimizing concrete mix design, belonging to the field of concrete material design technology. It acquires raw material information and environmental parameters from the target construction scenario to form a basic mix design dataset, and extracts performance response feature sequences from historical construction records. Based on this, a candidate mix design set is generated. Through virtual trial mixing, slump retention curves and early strength growth curves are obtained, and a comprehensive evaluation is conducted to select the optimal mix design. Furthermore, during the actual mixing process, mixing power, slump, and environmental changes are collected to form on-site feedback parameters, which are compared with the optimal mix design. When deviations from preset thresholds occur, the water-cement ratio and admixture dosage are automatically adjusted, and an updated mix design is output. This invention can reflect the coupling relationship between the evolution of concrete fluidity and the hydration reaction process, achieving dynamic optimization of the mix design and improving construction stability and early performance consistency.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of concrete material design technology, specifically to a method and system for optimizing concrete mix design based on artificial intelligence. Background Technology

[0002] In small-scale, short-term concrete applications such as precast component repair and winter construction, existing mix proportions rely heavily on experience-based trial mixing and limited experimental feedback. This makes it difficult to address the coupled effects of aggregate moisture content fluctuations, admixture lag effects, and sudden changes in early hydration heat release. Especially in small components or local reinforcement, the small mixing volume and large environmental disturbances can easily lead to abnormal problems such as initial slump meeting the standard but rapid loss within 30 minutes, and simultaneous occurrence of local bleeding and micro-cracks, resulting in increased strength dispersion and potential durability risks. Existing methods lack the ability to optimize the instantaneous state of raw materials in conjunction with the construction rhythm, making it difficult to correct key mix proportion parameters in a timely manner, resulting in insufficient optimization accuracy and stability. Summary of the Invention

[0003] The purpose of this invention is to provide a concrete mix design optimization method and system based on artificial intelligence to address the shortcomings of the prior art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a concrete mix design optimization method based on artificial intelligence, comprising: The raw material information and environmental parameters of the target construction scene are obtained and preprocessed to form a basic mix proportion dataset. The raw material information includes cement type, aggregate gradation and moisture content, admixture type and dosage range, and environmental parameters include temperature and humidity. Based on the aforementioned basic mix design dataset, and combined with slump evolution, setting time, and strength results from historical construction records, a performance response feature sequence is extracted. Based on the performance response characteristic sequence, a candidate mix design set is generated, and a virtual trial mix design is performed on each candidate mix design to obtain the corresponding slump retention curve and early strength growth curve. The slump retention curve and the early strength growth curve are comprehensively evaluated, the fit score index of each candidate ratio is calculated, and the optimal ratio is selected. During the actual mixing process, the mixing power, slump at the outlet and environmental changes are collected in real time to form on-site feedback parameters, which are then compared with the matching score index corresponding to the optimal mix ratio. When the on-site feedback parameters deviate from the preset threshold, the water-cement ratio and the amount of admixture are automatically adjusted, and an updated mix design is output.

[0005] Preferably, the steps for forming the basic proportion dataset include: The effective water usage is calculated based on the aggregate moisture content, and the water content brought in by the aggregate is combined with the design mixing water for verification to obtain the corrected water usage parameters. The aggregate gradation is merged into particle size ranges, and the boundaries of admixture type and dosage ranges are sorted out to form continuous gradation records and adjustable admixture range records, respectively. The corrected water parameters, gradation continuous records, admixture adjustable range records, and environmental status records are summarized according to the same mixing batch to form a basic mix proportion dataset.

[0006] Preferably, the steps for extracting the performance response feature sequence include: The basic mix design dataset and historical construction records are aligned in time according to mixing batch, mixing start time and discharge time to construct a multi-stage response record unit that includes slump evolution, setting time and strength results. The multi-stage response recording unit is segmented and the difference is analyzed to identify the slump drop interval and the condensation inflection point, forming a key response segment identifier set. The intensity results are back-screened based on the key response section identifier set to extract the performance response feature sequence that matches the slump stability and condensation coordination.

[0007] Preferably, the step of forming the key response segment identifier set includes: Calculate the slump loss between adjacent collection times and the slump loss per unit time. When both reach the preset loss condition at the same time, determine the corresponding time period as the slump drop interval. The slump evolution before the initial setting time is divided into continuous time periods. When the slump loss intensity of the later time period is higher than that of the previous time period and continues to reach the preset duration, the corresponding starting time is determined as the setting inflection point. The slump drop intervals and setting transition points are identified according to mixing batch, start time, end time, and section type to obtain a set of key response section identifiers.

[0008] Preferably, the step of generating the candidate formulation set includes: The slump decay rate constraint range and the early intensity development rate constraint range are extracted from the performance response feature sequence to form a dual performance boundary. Historical mix proportions that meet the target strength level and whose slump drop interval does not cross the solidification inflection point are selected as seed mix proportions, and verified using the aforementioned dual performance boundaries to form an initial candidate mix proportion group; The initial candidate formulation group is perturbed and expanded according to the linkage adjustment relationship between water-cement ratio and admixture dosage to form a candidate formulation set.

[0009] Preferably, the step of obtaining the slump retention curve includes: According to the mixing sequence and time interval corresponding to the target construction scenario, the process of each candidate proportion in the candidate proportion set is reproduced; The target slump value at the outlet is used as the initial value, and the foundation loss is determined segment by segment according to the slump decay rate per unit time. When the current time segment enters the slump drop interval or is within the preset time range before the solidification inflection point, the base loss is corrected in segments, and the slump values ​​at each time point are connected in chronological order to obtain the slump maintenance curve.

[0010] Preferably, the steps for obtaining the early intensity growth curve include: The mixing state of each candidate ratio is divided into time periods according to a preset time interval, forming a reaction stage division sequence covering the period before and after initial setting. Within each reaction stage, reaction activity indicators are extracted based on changes in mixing temperature rise and water consumption, and these are accumulated in chronological order to form a reaction cumulative index sequence. Based on the cumulative response index sequence, the accelerated growth segment, the steady growth segment, and the stable segment of intensity formation are identified, and the intensity change trend of each segment is spliced ​​together in chronological order to obtain the early intensity growth curve.

[0011] Preferably, the steps for selecting the optimal ratio include: A piecewise stationarity analysis was performed on the slump retention curve to extract a sequence of slump stability segments that meet the requirements of duration and lower limit of construction slump. The early intensity growth curve is analyzed for stage growth rate to form a sequence of intensity development segments that include rapid intensity development intervals and slow intensity growth intervals; Based on the temporal connection between the slump stability zone sequence and the intensity development zone sequence, a performance synergy evaluation value is generated, and the matching score index and optimal ratio for each candidate mix ratio are determined accordingly.

[0012] Preferably, the steps for outputting the updated mix design include: The on-site feedback parameter sequence is aligned with the slump retention curve corresponding to the optimal mix ratio in time. Based on the difference between the measured slump value and the expected slump value, a slump deviation characteristic sequence is formed. Combined with the environmental change trend and the mixing power fluctuation characteristics, a real-time adaptation evaluation value is obtained. When the real-time adaptation evaluation value is lower than the preset threshold, or when the slump deviation feature sequence shows continuous deviation, it is determined that there is insufficient fluidity, excessive fluidity, abnormal mixing resistance, or environmentally induced deviation, thus forming a deviation type discrimination result. Based on the deviation type identification results, the water-cement ratio and admixture dosage are adjusted step by step, and the slump retention curve and early strength growth curve are updated simultaneously. The updated mix proportion scheme is output, which includes the corrected water parameters, the corrected water-cement ratio, the corrected admixture dosage, and the verification conditions.

[0013] This invention also provides an artificial intelligence-based concrete mix design optimization system, comprising: The raw material data acquisition module acquires raw material information and environmental parameters of the target construction scenario, and performs preprocessing to form a basic mix proportion dataset. The raw material information includes cement type, aggregate gradation and moisture content, admixture type and dosage range, and environmental parameters include temperature and humidity. The feature extraction module, based on the basic mix design dataset and combined with slump evolution, setting time and strength results from historical construction records, extracts performance response feature sequences. The virtual mix design module generates a set of candidate mix proportions based on the performance response characteristic sequence, and performs virtual mix design simulations on each candidate mix proportion to obtain the corresponding slump retention curve and early strength growth curve. The optimal mix ratio screening module comprehensively evaluates the slump retention curve and the early strength growth curve, calculates the fit score index of each candidate mix ratio, and screens out the optimal mix ratio. The comparison module collects mixing power, slump at the outlet and environmental changes in real time during the actual mixing process, forms on-site feedback parameters, and compares them with the adaptation score index corresponding to the optimal ratio. The proportioning update module automatically adjusts the water-cement ratio and admixture dosage when the on-site feedback parameters deviate from the preset threshold, and outputs the updated proportioning scheme.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention introduces a continuous physical correlation method—combining modified water parameters, key response section identifiers, and response coupling weights—to uniformly characterize the effects of aggregate moisture content fluctuations, admixture lag, and hydration exothermic processes on fluidity and setting behavior. Compared to existing technologies that rely solely on static mix proportions or single test results, this approach starts with the slump drop interval and setting inflection point, capturing the critical stages of rapid remodeling of cement particle flocculation structure and consumption of free water. Furthermore, it establishes a coupling relationship through the synchronous changes in raw material moisture content and admixture dosage, thereby accurately reflecting the physical evolution process between the decline in slurry lubrication capacity and the acceleration of structure formation. This invention directly solves the problem of unpredictable short-term drastic slump loss and setting mutations in small-batch construction, enabling mix proportion optimization to constrain and control sensitive hydration reaction intervals, significantly reducing the risk of fluidity instability and early defects.

[0015] This invention achieves dynamic adaptive adjustment of the mix proportion during actual mixing through a closed-loop control method consisting of a reaction accumulation index sequence, a slump maintenance curve, and on-site feedback correction. Its core lies in using mixing temperature rise and water consumption to characterize the cement hydration rate, transforming the microscopic hydration reaction process into a quantifiable reaction activity, and establishing a linkage correction relationship with real-time collected mixing power and slump deviation. Unlike existing technologies that adjust the water-cement ratio or admixture dosage solely based on experience, this scheme can determine the synergistic adjustment path of the water-cement ratio and admixtures based on the direction of influence of fluidity deviation and environmental disturbances on the hydration process, ensuring a stable match between the slurry cohesive structure and the aggregate skeleton. This invention can maintain a gradual decrease in slump over time during construction, while ensuring that the hydration reaction rapidly enters the strength development stage during the critical demolding or forming period, thereby improving construction stability and the consistency of early structural performance. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a flowchart of the method of the present invention.

[0018] Figure 2 This is a flowchart of the system modules of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1, please refer to Figure 1 As shown in this embodiment, the artificial intelligence-based concrete mix design optimization method includes: The raw material information and environmental parameters of the target construction scenario are obtained and preprocessed to form a basic mix proportion dataset. The raw material information includes cement type, aggregate gradation and moisture content, admixture type and dosage range, and environmental parameters include temperature and humidity.

[0021] In one specific embodiment, firstly, the concrete application location, pouring time window, and mixing batch corresponding to the target construction scenario are determined, and this construction scenario is used as the data acquisition boundary. Subsequently, the cement type, aggregate gradation and moisture content, admixture type and dosage range of the corresponding batch are collected, and the temperature and humidity at the construction site are collected simultaneously. The cement type is used to define the reaction basis of the cementitious material, the aggregate gradation is used to characterize the aggregate filling state, the aggregate moisture content is used to correct the actual mixing water, and the admixture type and dosage range are used to define the subsequent mix proportion adjustment range.

[0022] After obtaining the above raw data, the aggregate moisture content is effectively converted by water, and the water adhering to the aggregate surface is merged and checked with the designed mixing water to obtain the corrected water parameters, so as to avoid the actual water-cement ratio deviating from the design value due to fluctuations in aggregate moisture content. At the same time, the aggregate gradation is processed by merging the particle size range, and different screening records are unified into continuous gradation records of the same particle size range to eliminate the differences in particle size markings between different test batches.

[0023] The boundaries of admixture types and dosage ranges are sorted out, and dosage values ​​exceeding the allowable range for construction are eliminated. Adjustable admixture range records are formed according to the minimum adjustable dosage unit. Temperature and humidity are linked to the collection time and mixing batch to form environmental status records, so that each set of raw material information has its corresponding on-site environmental conditions.

[0024] Finally, the corrected water parameters, continuous gradation records, adjustable admixture range records, and environmental condition records are summarized according to the same mixing batch. Missing, duplicate, and obviously abnormal items are checked for consistency, resulting in a basic mix design dataset that reflects the material state and environmental disturbances under the target construction scenario. This basic mix design dataset serves as input data for subsequent extraction of performance response feature sequences, ensuring that subsequent slump evolution, setting time, and strength results accurately correspond to the material state.

[0025] Based on the aforementioned basic mix design dataset, and combined with slump evolution, setting time, and strength results from historical construction records, a performance response feature sequence is extracted.

[0026] In one implementation, when extracting performance response feature sequences based on a basic mix design dataset, the basic mix design dataset is first aligned with historical construction records in chronological order.

[0027] Each record in the basic mix design dataset uses the mixing batch, mixing start time, and discharge time as the time reference. Slump evolution, setting time, and strength results in historical construction records are merged according to the same time reference. Slump evolution includes discharge slump and slump values ​​collected at predetermined time intervals after discharge, which can be set to 5 to 15 minutes. Setting time includes initial setting time and final setting time, or includes continuous monitoring time points that can reflect changes in the setting process. Strength results include compressive strength records of the same mixing batch at 1 day, 3 days, or 7 days of age.

[0028] If the collection time in the historical construction records is not completely consistent with the time reference in the basic mix design data, they will be aligned according to the principle of prioritizing adjacent times. When the time deviation does not exceed half of the predetermined time interval, they will be included in the same stage. When the time deviation exceeds half of the predetermined time interval, they will not participate in the calculation of the current stage.

[0029] After time alignment is completed, multi-stage response recording units are constructed for the same mixing batch. Each multi-stage response recording unit includes at least the raw material moisture content, admixture dosage, ambient temperature, ambient humidity, slump value at each sampling time, setting time, and strength results.

[0030] To ensure comparability between different batches, slump values ​​are grouped into stages based on discharge time, holding stage, and near-initial setting stage. Discharge time is 0 to 5 minutes after the concrete leaves the mixing plant, holding stage is 5 to 30 minutes after discharge, and near-initial setting stage is 30 minutes after discharge until the initial setting time.

[0031] If the required construction time for the target construction scenario is less than 30 minutes, the stage end time is kept to be the target construction time; if the required construction time for the target construction scenario is greater than 60 minutes, the stage end time is kept to be 60 minutes. The resulting multi-stage response recording unit can simultaneously retain the correspondence between the mix proportion input, construction environment, and performance output.

[0032] When performing piecewise difference analysis on multi-stage response recording units, the slump difference between adjacent acquisition times is first calculated. The calculation process is as follows: subtract the slump value of the previous acquisition time from the slump value of the next acquisition time to obtain the adjacent slump change; then divide the adjacent slump change by the time interval between the two acquisition times to obtain the slump change per unit time. A negative slump change per unit time indicates a slump loss.

[0033] The determination of sudden slump drops is based on a dual-condition approach: both a pre-set threshold of 10 mm to 30 mm for adjacent slump losses and a pre-set threshold of 0.5 mm / min to 2.0 mm / min for slump losses per unit time. The threshold can be determined based on the allowable fluctuations in slump flow required for the target construction scenario. For localized repairs of precast components, a threshold of 10 mm to 15 mm is used; for routine small-batch cast-in-place construction, a threshold of 15 mm to 25 mm is used; and for scenarios with significant allowable slump fluctuations, a threshold of 25 mm to 30 mm is used.

[0034] The setting point is determined by its relative position to the slump evolution. The slump evolution before the initial setting time is divided into several consecutive time periods, and the change in slump per unit time is calculated for each period. If the slump loss per unit time in a later time period is not less than 1.5 times that of the previous time period, and the duration is between 10 and 20 minutes, the starting time of the later time period is designated as a candidate setting point. If the initial setting time also exists in the setting time record, the candidate setting point closest to and before the initial setting time is designated as the setting point. If no candidate setting point meets the criteria, the time of collection with the largest slump loss within the 10 to 20 minutes before the initial setting time is designated as the setting point.

[0035] The critical response section identifier set is formed by the slump drop interval and the solidification transition node. Each identifier includes at least the corresponding mixing batch, start time, end time, section type, and slump loss within the section.

[0036] After forming the key response segment identifier set, the response coupling weight sequence is obtained by combining the synchronous change relationship between the moisture content of raw materials and the amount of admixture in each segment.

[0037] During the calculation, the starting and ending batches corresponding to the critical response sections are first determined, and the moisture content of the raw materials and the dosage of the admixtures for the two batches are obtained respectively.

[0038] The degree of change in raw material moisture content is obtained by subtracting the raw material moisture content of the starting batch of the section from the moisture content of the last batch of the section, taking the absolute value, and then comparing it with the difference between the maximum and minimum raw material moisture content under the same target construction scenario in the basic mix data; the degree of change in admixture dosage is obtained by subtracting the admixture dosage of the starting batch of the section from the admixture dosage of the last batch of the section, taking the absolute value, and then comparing it with the width of the admixture dosage range.

[0039] If there are multiple collection batches within the same segment, the degree of change of adjacent batches is calculated in chronological order, and the average value is taken as the degree of change of the segment.

[0040] The response coupling weight is set to a range of 0.5 to 1.5. The initial value is 1. When the change in raw material moisture content is greater than 0.3% and the change in admixture dosage is greater than 0.3%, the weight is increased by 0.2 to 0.4. When only one of these changes is greater than 0.3, the weight is increased by 0.1 to 0.2. When neither change is greater than 0.1, the weight is decreased by 0.1 to 0.2. Values ​​below 0.5 after the increase or decrease are counted as 0.5, and values ​​above 1.5 are counted as 1.5.

[0041] If the critical response segment falls within a sudden drop in slump, and the endpoint of this segment is within 10 minutes before or after the condensation inflection point, then the value is increased by 0.1 based on the aforementioned values, but not exceeding 1.5. Arranging the values ​​corresponding to each critical response segment in chronological order yields the response coupling weight sequence.

[0042] When performing graded backtracking screening of strength results based on the response coupling weight sequence, the strength results of the same batch of mixing are used as the backtracking object, and historical construction records that correspond to the key response sections are searched along the time sequence.

[0043] The strength results are graded based on the degree of conformity to the target strength: records with an age strength of 95% or higher of the target strength value are classified as stable; records with a strength of 85% or higher but lower than 95% are classified as adjustable; and records with a strength lower than 85% are classified as rejectable. The target strength value can be the corresponding age strength required in the construction mix design, or it can be the average strength of qualified batches under the same target construction scenario.

[0044] For stable grade records, the slump loss, condensation inflection point, and response coupling weight within the corresponding key response segment are retained; for adjustable grade records, they are only retained when the response coupling weight is not higher than 1.2 and the slump drop interval does not cross the condensation inflection point; for discarded grade records, they are not included in subsequent performance response feature extraction.

[0045] After screening, the retained records will be organized in chronological order of mixing batches, and performance response feature sequences that match slump stability and setting coordination will be extracted.

[0046] The performance response characteristic sequence includes at least the slump loss per unit time at each stage, the location of the slump drop interval, the location of the condensation inflection point, the response coupling weight, and the corresponding intensity level.

[0047] If the difference in response coupling weight between two adjacent retained records is less than 0.1, and the position deviation of the slump drop interval is less than a predetermined time interval, they are merged into the same performance response feature; if the difference reaches 0.1 or more, or the position deviation of the slump drop interval reaches more than a predetermined time interval, they are retained separately.

[0048] The performance response feature sequence obtained after the above processing can serve as the data basis for subsequent generation of candidate mix design sets and for deducing slump retention curves and early intensity growth curves.

[0049] Based on the performance response characteristic sequence, a candidate mix design set is generated, and virtual trial mix design is performed on each candidate mix design to obtain the corresponding slump retention curve and early strength growth curve.

[0050] In one implementation, when generating a candidate mix design set and performing virtual mix design simulation, the unit time slump loss, slump drop interval location, condensation inflection point location, response coupling weight, and corresponding strength level of each retained record are first read from the performance response feature sequence, and the workable time and early strength requirements required by the target construction scenario are used as the screening boundary.

[0051] The constraint range for the slump decay rate is determined based on qualified historical records. During calculation, the slump loss of each retained record during the retention period is first divided by the corresponding duration to obtain the slump loss per unit time for each retained record. Then, the highest and lowest 5% of abnormal records are removed, and the minimum and maximum values ​​among the remaining records are taken as the basic range for the slump decay rate. If fewer than 20 records remain, proportional removal is not performed; instead, records that significantly deviate from the allowable slump fluctuation range for construction are removed. The basic range for the slump decay rate is typically limited to 0.2 mm / min to 1.5 mm / min; values ​​below 0.2 mm / min are counted as 0.2 mm / min, and values ​​above 1.5 mm / min are counted as 1.5 mm / min, to avoid candidate mix proportions deviating from the fluidity retention requirements of the target construction scenario.

[0052] The constraint range of the early intensity development rate is determined based on the intensity results and age time. During the calculation, the intensity difference between adjacent ages is first divided by the time interval between adjacent ages to obtain the intensity development rate corresponding to each time period; then, stable grade records and adjustable grade records that meet the retention conditions are screened according to intensity level, and the median value of the intensity development rate in the screened records is taken as the benchmark value.

[0053] The lower limit of the early strength development rate can be taken as 80% to 90% of the benchmark value, and the upper limit can be taken as 110% to 125% of the benchmark value. When the target construction scenario is the repair of precast components or short-term demolding construction in winter, the lower limit should be taken as 90% of the benchmark value; when the target construction scenario has lower requirements for early strength but higher requirements for maintaining fluidity, the lower limit should be taken as 80% of the benchmark value. The slump decay rate constraint range and the early strength development rate constraint range together form a dual performance boundary.

[0054] When constructing the initial candidate mix group, the water-cement ratio, cementitious material dosage, sand ratio, aggregate gradation, and adjustable range of admixtures after effective water correction are recorded in the basic mix data.

[0055] First, a historical mix design that meets the target strength level and whose slump drop does not cross the setting inflection point is selected as the seed mix design. Then, the water-cement ratio, admixture dosage, and corrected water parameters in the seed mix design are respectively substituted into the dual performance boundary for verification. The verification process includes: judging whether the slump loss per unit time falls within the slump decay rate constraint range based on the performance response characteristics corresponding to the seed mix design; and judging whether the early strength development rate is not lower than the lower limit of the early strength development rate.

[0056] If both conditions are met, the seed mix ratio is retained to form the initial candidate mix ratio group; if only one condition is met, it enters the adjustable record before disturbance expansion; if neither condition is met, it does not participate in the formation of subsequent candidate mix ratios. If the number of records in the initial candidate mix ratio group is less than 3, the top 3 groups with the lowest response coupling weight can be selected from the stability level records to supplement them, but the supplemented records still need to meet the construction-allowed water-cement ratio and admixture dosage boundaries.

[0057] When performing a perturbation expansion process on the initial candidate mix proportions, the water-cement ratio and admixture dosage are adjusted using a linkage rule, without unconstrained changes to a single parameter. The perturbation step size for the water-cement ratio can be set to 0.005 to 0.02, and the perturbation step size for the admixture dosage can be set to 0.02% to 0.10% of the mass of the cementitious material.

[0058] When the performance response characteristic sequence shows that the slump decay rate is close to the upper limit and the early strength development rate is more than 10% higher than the lower limit, priority should be given to reducing the impact of actual water fluctuations per unit of cementitious material, and the admixture dosage should be increased within the adjustable range of the admixture; when the slump decay rate is in the middle of the constraint range and the early strength development rate is close to the lower limit, priority should be given to reducing the water-cement ratio, while keeping the admixture dosage not exceeding the middle to upper part of the adjustable range of the admixture; when the slump decay rate is lower than the middle of the constraint range and the strength development rate is higher than the upper limit, the admixture dosage should be reduced or the water-cement ratio should be appropriately increased to keep the fluidity of the candidate mix proportion in harmony with the early strength development.

[0059] After each disturbance, the water usage parameters are recalculated and corrected. During the calculation, the amount of water carried in by the aggregate is first determined based on the aggregate moisture content, and then the amount of water carried in by the aggregate is subtracted from the design mixing water to obtain the actual amount of water added. If the actual amount of water added is less than 0, the disturbance record is deemed invalid.

[0060] The amount of cementitious materials is kept within the allowable range of the basic mix design dataset, and the sand ratio disturbance does not exceed ±3 percentage points of the original sand ratio to avoid masking the influence of the linkage between water-cement ratio and admixture dosage due to changes in the skeleton. After the above perturbation expansion, the extended candidate mix designs are arranged in order of water-cement ratio from low to high, admixture dosage from low to high, and corrected water parameters from small to large, forming an extended candidate mix design set with a gradient change relationship.

[0061] There must be at least one identifiable ratio gradient between any two adjacent candidate ratios, and the ratio gradient must be no less than the corresponding perturbation step size to avoid duplicate recordings in the virtual trial matching simulation.

[0062] When reproducing the process for the extended candidate mix design, the mixing sequence and time intervals should be consistent with the target construction scenario. The mixing sequence can be set as follows: dry mix coarse and fine aggregates for 10 to 30 seconds; add some water corresponding to the corrected water parameters and mix for 20 to 40 seconds; add cement and other cementitious materials and mix for 30 to 60 seconds; add the remaining water and admixtures and continue mixing for 60 to 120 seconds. When different construction scenarios have fixed mixing procedures, the time for each stage should be determined according to the on-site procedures. Process reproduction should not change the mix proportion gradient in the extended candidate mix design; it should only read the factors that may cause changes in slump at each stage according to the established mixing sequence, including the admixture entry time, aggregate moisture content correction results, mixing time, and ambient temperature and humidity.

[0063] When projecting the slump change trend, the target slump value at the discharge point is used as the initial value. The initial value can be the median design slump, with an allowable fluctuation range generally between ±10 mm and ±30 mm. Subsequently, the slump loss is deducted segment by segment at predetermined time intervals. The slump loss for each time segment is determined jointly by the base loss and the segment correction.

[0064] The base loss is the product of the slump decay rate per unit time corresponding to the candidate mix ratio and the duration of the time period. The segment correction is determined based on the set of key response segment identifiers. If the current time period falls within the slump drop range, the base loss is increased by 10% to 35%. If the current time period is within 10 to 20 minutes before the condensation inflection point, the base loss is increased by 5% to 20%. If both conditions exist, the increase is the sum of the higher of the two values ​​and half the difference between them, with the final increase not exceeding 40%.

[0065] The slump values ​​after deduction for each time period are connected in chronological order to obtain the slump retention curve corresponding to the candidate mix proportion. A slump value below 0 is counted as 0; if the slump value is below the lower limit of construction within the target workable time, the candidate mix proportion is marked as insufficient fluidity, but its curve is still retained for comprehensive evaluation.

[0066] When conducting a phased cumulative evaluation of the reaction process of cementitious materials with various candidate proportions, the mixing state is first divided into time periods according to preset time intervals, forming a reaction stage division sequence covering the period before and after initial setting. The preset time interval can be set to 5 to 15 minutes; it should include at least 3 time periods before initial setting and at least 2 time periods after initial setting. If the initial setting time is less than 30 minutes, the preset time interval should preferably be 5 minutes; if the initial setting time is greater than 90 minutes, the preset time interval can be 10 to 15 minutes. The starting point of the reaction stage division sequence is the moment when water is added to contact the cementitious material, and the ending point can be the preliminary evaluation time corresponding to the target early strength age, such as 6 hours, 12 hours, or 24 hours.

[0067] During each reaction stage, reaction activity indicators are extracted based on changes in mixing temperature rise and water consumption.

[0068] The temperature rise during mixing is obtained by subtracting the initial temperature of the current stage from the end temperature of the current stage; the water consumption process is obtained by comparing the decrease in available water in the current stage with the initial available water after water addition.

[0069] The available water volume at the current stage can be obtained by subtracting the estimated value of aggregate adsorbed water and the estimated value of water consumed by cementitious materials from the corrected water use parameters.

[0070] The estimated water consumption of the cementitious materials can be determined by multiplying the mass of the cementitious materials by the stage consumption ratio. The stage consumption ratio is determined based on the time of the reaction stage: 0.5% to 3% before initial setting, 3% to 8% from initial setting to final setting, and 8% to 18% after final setting to 24 hours. The calculation process for the reaction activity indicator value is as follows: divide the change in mixing temperature rise by the upper limit of allowable temperature rise in the target construction scenario to obtain the temperature rise ratio; divide the current reduction in available water volume by the initial available water volume to obtain the water consumption ratio; multiply the temperature rise ratio and water consumption ratio by their respective influence ratios and add them together to obtain the reaction activity indicator value. The influence ratio of the temperature rise ratio can be taken as 0.4 to 0.6, and the influence ratio of the water consumption ratio can be taken as 0.4 to 0.6, with the sum of the two being 1. A reaction activity indicator value less than 0 is counted as 0, and a value greater than 1 is counted as 1.

[0071] When forming a reaction cumulative index sequence, the reaction activity indicator value of the first reaction stage is used as the starting cumulative value, and the cumulative value of the next reaction stage is equal to the cumulative value of the previous reaction stage plus the reaction activity indicator value of the next reaction stage.

[0072] To avoid cumulative deviations caused by varying time periods, the reaction activity indicator value for each reaction stage is first multiplied by the ratio of the stage duration to a preset time interval before being accumulated. The accumulated values ​​are then arranged chronologically to obtain the reaction cumulative index sequence. If the ambient temperature is below 5 degrees Celsius, the reaction activity indicator value for each reaction stage can be multiplied by a low-temperature correction ratio of 0.6 to 0.8; if the ambient temperature is above 30 degrees Celsius, it can be multiplied by a high-temperature correction ratio of 1.1 to 1.3, but the corrected reaction activity indicator value will still not exceed 1.

[0073] When identifying the acceleration and stability zones formed by the intensity based on the cumulative response index sequence, the increment between two adjacent cumulative values ​​is compared.

[0074] If the cumulative increase over two or more consecutive time periods is higher than 120% of the average increase over all time periods, then that consecutive time period is classified as an accelerating intensity formation zone. If the cumulative increase over two or more consecutive time periods is between 80% and 120% of the average increase over all time periods, then it is classified as a steady growth zone. If the cumulative increase over two or more consecutive time periods is lower than 80% of the average increase over all time periods, and the zone has already reached its final solidification stage, then it is classified as a stable zone. Isolated points that do not meet the requirement of two consecutive time periods are assigned to a zone with a longer adjacent time period. Each zone corresponds to a different intensity change slope: the accelerating zone is defined as 110% to 125% of the baseline intensity development rate after screening, the steady growth zone as 90% to 110% of the baseline value, and the stable zone as 50% to 90% of the baseline value.

[0075] When generating segmented change trends, the starting point is either 0 initial strength or the equivalent strength corresponding to the initial penetration resistance under the same conditions. The end strength of each segment is calculated based on its duration and the corresponding strength change slope. The calculation process is as follows: multiply the current segment duration by the corresponding strength change slope to obtain the current segment strength increase; then add this strength increase to the segment's starting strength to obtain the segment's ending strength. The starting point of the next segmented change trend is the ending point of the previous segmented change trend, ensuring continuity between adjacent segments. All segmented change trends are spliced ​​together in chronological order to form an early strength growth curve reflecting the relationship between strength and time. This early strength growth curve, along with the slump retention curve for the same candidate mix design, is used in subsequent comprehensive evaluation to calculate the fit scoring index and select the optimal mix design.

[0076] The slump retention curve and the early strength growth curve are comprehensively evaluated, the fit score index of each candidate ratio is calculated, and the optimal ratio is selected.

[0077] In one implementation, when performing a segmented stability analysis on the slump retention curves corresponding to each candidate mix proportion, the analysis range is defined by the target workable time, and the slump retention curves are divided into multiple continuous time periods according to a predetermined time interval.

[0078] The predetermined time interval is consistent with the time interval in the aforementioned virtual trial matching simulation, and can be 5 to 15 minutes; when the target construction time is less than 30 minutes, the predetermined time interval is 5 minutes; when the target construction time is not less than 30 minutes and not more than 60 minutes, the predetermined time interval is 10 minutes; when the target construction time is more than 60 minutes, the predetermined time interval is 15 minutes.

[0079] Within each time period, the starting slump value and the ending slump value of that time period are read. First, the starting slump value is subtracted from the ending slump value to obtain the slump loss of the section. Then, the slump loss of the section is divided by the duration of that time period to obtain the slump loss rate of the section.

[0080] When the slump loss rate of a section is less than or equal to 0.5 mm per minute, the time period is designated as a stable section; when it is greater than 0.5 mm per minute but less than or equal to 1.0 mm per minute, the time period is designated as a controllable attenuation section; when it is greater than 1.0 mm per minute, the time period is designated as a non-stable section.

[0081] For local repairs of precast components, the threshold for determining a stable section can be 0.4 mm / min; for ordinary small-batch cast-in-place projects, the threshold for determining a stable section can be 0.5 mm / min to 0.7 mm / min.

[0082] After determining a single time period, adjacent time periods that are both stable or controllable attenuation periods are merged to obtain a continuous maintenance period.

[0083] A slump-stable section is defined as one whose continuous duration is not less than 30% of the target workable time, and whose slump value at the end of the section is not lower than the lower limit of the allowable slump for construction. Sections whose duration is less than 30% of the target workable time are not considered slump-stable sections. Multiple slump-stable sections are arranged in chronological order to form a slump-stable section sequence.

[0084] For each slump-stabilized section, at least the start time, end time, section duration, average slump loss rate, and end slump value should be recorded.

[0085] When performing stage growth rate analysis on the early intensity growth curve, adjacent moments in the reaction stage division sequence are used as the calculation boundary.

[0086] Read the intensity values ​​of two adjacent moments, subtract the intensity value of the previous moment from the intensity value of the later moment to obtain the increase in intensity during a phase; then divide the increase in intensity during a phase by the duration of that phase to obtain the phase growth rate. Arrange all phase growth rates in chronological order and take the median value as the baseline rate for the intensity phase.

[0087] A period of rapid intensity development is defined as one where the growth rate reaches 120% or more of the baseline rate for the intensity stage and continues for two or more consecutive time periods. A period of gradual increase is defined as one where the growth rate is below 80% of the baseline rate for the intensity stage and continues for two or more consecutive time periods. A period of stable development is defined as one where the growth rate is between 80% and 120% of the baseline rate for the intensity stage. Individual time periods that do not meet the continuous time requirement are grouped into adjacent time periods with longer durations.

[0088] The rapid development interval, stable development interval, and slow increase interval are arranged according to their occurrence time to form a sequence of intensity development intervals.

[0089] For each intensity development segment, at least the start time, end time, segment duration, average growth rate of each stage, and intensity value at the end of the segment should be recorded.

[0090] If the target early intensity requirement corresponds to a 6-hour or 12-hour age, then the intensity development segment sequence must extend to at least the corresponding age; if the target early intensity requirement corresponds to a 24-hour age, then the intensity development segment sequence must extend to 24 hours, and the slow increase intervals that appear before 24 hours must not exceed 40% of the target evaluation time consecutively, otherwise the candidate ratio will be marked as an insufficient intensity development record.

[0091] When calculating the degree of synergy based on the time matching relationship between the slump stability zone sequence and the strength development zone sequence, the effective construction window should be determined first.

[0092] The effective construction window begins at the moment of machine deployment and ends at the target constructable time. The slump stability zone within the effective construction window is used to evaluate construction adaptability, while the rapid strength development zone after the effective construction window and close to the initial setting time is used to evaluate structural formation coordination.

[0093] When calculating the construction adaptability score, the total duration of the slump-stable section within the effective construction window is divided by the target workable time, and then multiplied by 40 to obtain the fluidity retention score. If the total duration of the slump-stable section exceeds the target workable time, the score is calculated based on the target workable time. If a non-stable section appears within the effective construction window, 2 to 5 points are deducted for each non-stable section. The deduction is determined by the slump loss rate of that section; the higher the loss rate, the greater the deduction, with a cumulative deduction not exceeding 15 points.

[0094] When calculating the structural compatibility value, first determine the relationship between the starting time of the rapid strength development interval and the target constructable time.

[0095] When the start time of the rapid development of intensity is later than the end time of the target construction period, and the interval between the two is no more than 30 minutes, it is considered coordinated; the corresponding score is 25 to 30 points.

[0096] When the starting point of the rapid strength development interval falls within the effective construction window, it indicates that the fluidity maintenance stage and the rapid strength formation stage overlap prematurely, and the corresponding score is 10 to 20 points; the longer the overlap time, the lower the score.

[0097] If no rapid strength development zone appears within 60 minutes after the construction period ends, the corresponding score is 5 to 15 points.

[0098] If the intensity development segment sequence is continuously in a slow increase range before the early intensity age of the target, and the duration exceeds 40% of the target evaluation period, the structural formation coordination value shall not exceed 10 points.

[0099] The time matching relationship also includes the judgment of the connection between the stable slump section and the gradually increasing section.

[0100] If the slump stability zone ends and the rapid intensity development zone begins within 30 minutes, and the intensity value at the end of this rapid intensity development zone reaches more than 60% of the target early intensity value, 5 to 10 points will be added as a transition bonus. If the rapid intensity development zone begins before the slump stability zone ends, 5 to 10 points will be deducted. If the slump stability zone ends and the slump remains in a slow-increasing zone for more than 60 minutes, 5 to 15 points will be deducted. The time matching score after bonuses and deductions is limited to the range of 0 to 30 points.

[0101] The performance synergy evaluation score is based on a 100-point scale. The maximum score for liquidity maintenance is 40 points, the maximum score for structural coordination is 30 points, and the maximum score for time matching is 30 points.

[0102] During calculation, the scores of the three items are added together to obtain the performance synergy evaluation value corresponding to the candidate ratio. If the candidate ratio is marked as insufficient liquidity, the performance synergy evaluation value is reduced by 10 to 20 points from the sum of the scores; if the candidate ratio is marked as insufficient intensity development, 10 to 25 points are reduced; if both types of records exist, the cumulative reduction shall not exceed 35 points. If the result after deduction is lower than 0 points, it is counted as 0 points; if it is higher than 100 points, it is counted as 100 points.

[0103] When sorting and screening candidate formulations, the performance synergy evaluation value is used as the main value of the adaptation scoring index, and they are arranged from high to low. Candidate formulations with a performance synergy evaluation value of not less than 80 points are included in the preferred range, candidate formulations with a score between 60 and 80 points are included in the alternative range, and candidate formulations with a score below 60 points are not considered as the optimal formulation.

[0104] If the performance synergy evaluation value difference among multiple candidate formulations is less than 3 points, the candidate formulation with the water-cement ratio located in the middle of the extended candidate formulation set, the admixture dosage not reaching the upper limit of the admixture adjustable range, and the slump stability zone lasting for a longer period of time shall be given priority.

[0105] After sorting, the candidate mix proportion with the highest ranking that meets the lower limit of allowable slump and the target early strength requirements is determined as the optimal mix proportion. The corresponding slump retention curve, early strength growth curve, slump stability zone sequence, strength development zone sequence and adaptation scoring index are retained for subsequent on-site feedback parameter comparison.

[0106] During the actual mixing process, the mixing power, slump at the outlet, and environmental changes are collected in real time to form on-site feedback parameters, which are then compared with the adaptation score index corresponding to the optimal mix ratio.

[0107] In one implementation, when generating and comparing on-site feedback parameters, before the actual mixing begins, the adaptation score index corresponding to the optimal mix ratio, the slump retention curve, and the ambient temperature and humidity corresponding to the basic mix ratio data are read as on-site comparison benchmarks.

[0108] After the actual mixing begins, the power value of the mixing equipment is continuously recorded at intervals of 1 to 5 seconds, and the times of material feeding completion, admixture addition, mixing end, and discharge are recorded simultaneously. The mixing power change is calculated based on the no-load power of the mixing equipment. First, the no-load power is subtracted from the power value at each collection time to obtain the net power value. Then, the power is smoothed by averaging over 3 to 5 consecutive collection points to obtain the net power change record.

[0109] If the power difference between two adjacent sampling points exceeds 50% of the average net power in the previous 30 seconds and the duration is less than 5 seconds, the sampling point is removed as an instantaneous disturbance point; if the duration reaches more than 5 seconds, it is retained as a mixing power fluctuation characteristic.

[0110] The slump test results are collected within 0 to 5 minutes after the machine leaves the machine; if the target construction scenario requires secondary testing, 1 to 3 additional tests can be conducted within 10 to 30 minutes after the machine leaves the machine.

[0111] Ambient temperature and humidity data are recorded at 5 to 10-minute intervals. When the temperature change in the construction area exceeds 2 degrees Celsius or the humidity change exceeds 5 percentage points, an additional record is immediately added.

[0112] The mixing power variation, slump test results, and ambient temperature and humidity data are arranged according to the time of collection and bound to the mixing batch number to form a corresponding time sequence of on-site feedback parameters.

[0113] Each record in the on-site feedback parameter sequence should include at least the acquisition time, net power value, net power change range, measured slump value, on-site temperature value, and on-site humidity value.

[0114] Time alignment processing uses the discharge time as a common zero point. In the slump retention curve corresponding to the optimal mix ratio, each time point after the discharge time is used as the reference time point; the slump detection time in the field feedback parameter sequence is converted into elapsed time relative to the discharge time.

[0115] If the on-site testing time coincides with the reference time point in the slump maintenance curve, the expected slump value at that reference time point is directly read. If they do not coincide, linear supplementation is performed between two adjacent reference time points. The supplementation process is as follows: first, calculate the time ratio between the on-site testing time and the previous reference time point; then, subtract the expected slump value of the previous reference time point from the expected slump value of the next reference time point to obtain the reference difference; multiply the reference difference by the time ratio and add it to the expected slump value of the previous reference time point to obtain the expected slump value corresponding to the on-site testing time.

[0116] The real-time flowability offset is obtained by comparing the measured slump value with the expected slump value at the corresponding on-site testing time. The calculation process is as follows: subtract the expected slump value from the measured slump value to obtain the single flowability offset; a positive value indicates that the on-site flowability is higher than expected, and a negative value indicates that the on-site flowability is lower than expected. Then, divide the single flowability offset by the upper limit of the allowable slump fluctuation during construction to obtain the normalized offset. The upper limit of the allowable slump fluctuation during construction can be taken as 10 mm to 30 mm, 10 mm to 15 mm for local repair of precast components, and 20 mm to 30 mm for ordinary small-batch cast-in-place scenarios. The normalized offsets obtained from multiple testing times are arranged in chronological order to form a slump deviation characteristic sequence; if the absolute value of the normalized offset at two or more consecutive testing times is greater than 1, it is marked as a continuous deviation record; if only one testing time exceeds 1, and the values ​​of the preceding and following testing times do not exceed 0.5, it is marked as an occasional deviation record.

[0117] The trend of environmental change is determined based on the difference between the on-site temperature and humidity values ​​and the corresponding environmental state records in the baseline mix data.

[0118] The temperature difference is calculated by subtracting the baseline temperature from the on-site temperature, and the humidity difference is calculated by subtracting the baseline humidity from the on-site humidity. When the absolute value of the temperature difference is less than 2 degrees Celsius and the absolute value of the humidity difference is less than 5 percentage points, the environmental change trend is recorded as slight; when the absolute value of the temperature difference is between 2 and 5 degrees Celsius, or the absolute value of the humidity difference is between 5 and 15 percentage points, it is recorded as moderate; when the absolute value of the temperature difference is greater than 5 degrees Celsius, or the absolute value of the humidity difference is greater than 15 percentage points, it is recorded as significant. Increased temperature and decreased humidity have an accelerating effect on slump loss; decreased temperature and increased humidity have a delaying effect on early strength growth.

[0119] When dynamically mapping and correcting the adaptation score index corresponding to the optimal mix ratio, the adaptation score index is used as the initial score. First, calculate the slump deviation deduction value: take the average of the absolute values ​​of each normalized offset in the slump deviation characteristic sequence, multiply by 15 to 25 to obtain the slump deviation deduction value; if there are continuous deviation records, add 5 to 10 points for deduction; if there are only occasional deviation records, no deduction is added. Next, calculate the power fluctuation deduction value: take the ratio of the net power change amplitude to the average net power of the previous 30 seconds. If the ratio is 10% to 20%, deduct 2 to 5 points; if it is 20% to 35%, deduct 5 to 10 points; if it exceeds 35%, deduct 10 to 15 points. Then calculate the environmental correction value: slight changes are assigned 0 points, moderate changes are deducted 3 to 8 points, and significant changes are deducted 8 to 15 points; when the temperature rises and the humidity decreases and is accompanied by a negative flow offset, the environmental correction value is taken as the upper limit of the corresponding range. The initial score is subtracted sequentially from the slump deviation deduction, power fluctuation deduction, and environmental correction value to obtain the real-time adaptation evaluation value; if the real-time adaptation evaluation value is lower than 0, it is counted as 0 points, and if it is higher than 100 points, it is counted as 100 points.

[0120] When comparing the real-time adaptation evaluation value with the preset threshold range, the preset threshold range can be set to 70 to 100 points. When the real-time adaptation evaluation value is not lower than 80 points, it is determined that the on-site condition is consistent with the optimal ratio; when the real-time adaptation evaluation value is between 70 and 80 points, it is determined to be a slight deviation, and it is necessary to continue tracking the next collection cycle; when the real-time adaptation evaluation value is lower than 70 points, it is determined that the ratio adjustment is required.

[0121] If the real-time adaptation assessment value is not lower than 70 points, but there are two consecutive negative shifts in the slump deviation characteristic sequence and the absolute value of each shift is greater than 80% of the upper limit of the allowable slump fluctuation during construction, the mix proportion adjustment will also be initiated. This comparison result, together with the slump deviation characteristic sequence, environmental change trend, and mixing power fluctuation characteristics, serves as the basis for automatically adjusting the water-cement ratio and admixture dosage.

[0122] When the on-site feedback parameters deviate from the preset threshold, the water-cement ratio and the amount of admixture are automatically adjusted, and an updated mix design is output.

[0123] In one implementation, when automatically adjusting the water-cement ratio and admixture dosage and outputting an updated mix design, the comparison between the real-time adaptation evaluation value and a preset threshold range is used as the activation condition. The preset threshold range can be 70 to 100 points; when the real-time adaptation evaluation value is below 70 points, the mix design correction process begins. The mix design correction process also begins when the real-time adaptation evaluation value is between 70 and 80 points but the slump deviation characteristic sequence shows a continuous negative shift; a continuous negative shift refers to two or more consecutive tests where the measured slump value is lower than the expected slump value at the corresponding moment in the slump maintenance curve, and the absolute value of each shift reaches more than 80% of the upper limit of the allowable slump fluctuation during construction.

[0124] The deviation type is determined based on the slump deviation characteristic sequence. At each detection time, the measured slump value is subtracted from the expected slump value to obtain the single flowability offset.

[0125] A single instance of fluidity offset is recorded as a point of insufficient fluidity when its absolute value is negative and exceeds the upper limit of the allowable slump fluctuation during construction; a single instance of fluidity offset is recorded as a point of excess fluidity when its absolute value is positive and exceeds the upper limit of the allowable slump fluctuation during construction; and a single instance of fluidity offset is recorded as an acceptable point when its absolute value does not exceed the upper limit of the allowable slump fluctuation during construction.

[0126] The upper limit of the allowable slump fluctuation during construction is based on the value obtained when comparing the on-site feedback parameters. For local repair of precast components, the upper limit is 10 mm to 15 mm, and for ordinary small-batch cast-in-place scenarios, the upper limit is 20 mm to 30 mm.

[0127] When the proportion of insufficient flowability points in the slump deviation characteristic sequence reaches 60% or more of the total number of detection points, or when two consecutive detection points are insufficient flowability points, the deviation type is determined to be insufficient flowability. When the proportion of excessive flowability points reaches 60% or more of the total number of detection points, or when two consecutive detection points are excessive flowability points, the deviation type is determined to be excessive flowability. When neither insufficient flowability points nor excessive flowability points meet the above conditions, but the net power change exceeds 20% of the average net power value of the previous 30 seconds, the deviation type is determined to be abnormal mixing resistance. When all detection points are within the acceptable range, but the real-time adaptation evaluation value is below 70 points, the deviation type is determined to be environmentally induced deviation.

[0128] When determining the direction of the linkage adjustment between the water-cement ratio and the admixture dosage, first read the water-cement ratio, admixture dosage, corrected water parameters, and the adjustable range of the admixture in the current candidate mix.

[0129] When fluidity is insufficient and the environmental trend shows an increase in temperature and a decrease in humidity, the admixture dosage should be increased first; the single increase should be 0.02% to 0.06% of the mass of the cementitious material. If the admixture dosage has reached more than 90% of the upper limit of the adjustable range of the admixture, the water-cement ratio should be increased slightly, with a single increase of 0.005 to 0.010, and the early strength growth curve should be calculated simultaneously to avoid the early strength being lower than the target requirement.

[0130] If the fluidity is insufficient and the mixing power fluctuation characteristics show that the net power continues to rise, first increase the amount of admixture, and then check and correct the water parameters; if the actual amount of added water is less than 95% of the designed amount of added water, compensate the amount of added water to 95% to 100% of the designed amount of added water.

[0131] When there is excess liquidity, the adjustment direction is opposite to that when there is insufficient liquidity. If the measured slump value is consistently higher than the expected slump value, and the early strength growth curve is close to the lower limit of the strength development rate, the water-cement ratio should not be reduced. Instead, the admixture dosage should be reduced first, with a single reduction of 0.02% to 0.05% of the cementitious material mass. If the early strength growth curve is still more than 10% higher than the lower limit of the strength development rate, the water-cement ratio can be reduced by 0.005 to 0.010, while keeping the admixture dosage in the middle of the adjustable range of the admixture.

[0132] When there is excess fluidity and the environmental trend is characterized by decreasing temperature and increasing humidity, the admixture dosage reduction value should be taken at the lower limit of the above range to prevent the subsequent slump from falling rapidly.

[0133] Adjustments to abnormal mixing resistance do not directly increase the variation in the water-cement ratio. When the net power variation is between 20% and 35%, first correct according to the lower limit of the single adjustment amount of the admixture dosage; when the net power variation exceeds 35%, check the direction of the slump deviation characteristic sequence. If it is simultaneously negative, increase the admixture dosage and compensate for the correction water parameters; if it is simultaneously positive, decrease the admixture dosage and keep the water-cement ratio unchanged.

[0134] The adjustment corresponding to the environmentally induced deviation is determined based on the trend of environmental change. When the temperature rises and the humidity decreases, the amount of admixture is increased. When the temperature decreases and the humidity rises, the water-cement ratio or the amount of admixture is decreased. The change in water-cement ratio in a single instance shall not exceed 0.005.

[0135] The ratio correction instruction sequence records the content of each adjustment in chronological order. Each record includes at least the deviation type judgment result, the adjustment object, the adjustment direction, the adjustment range, the execution batch, and the review time.

[0136] To avoid mix ratio fluctuations caused by continuous adjustments, an observation interval is set between two adjacent adjustments; the observation interval can be one slump test cycle, usually 10 to 30 minutes. Within any mixing batch, the cumulative adjustment of the water-cement ratio should not exceed 0.020, and the cumulative adjustment of the admixture dosage should not exceed 0.15% of the mass of the cementitious material; when the cumulative boundaries are exceeded, automatic adjustment is stopped and the current mix ratio is retained as the object of verification.

[0137] When performing step-by-step corrections to the current candidate mix proportions according to the mix proportion correction instruction sequence, the water parameters for correction are recalculated after each correction. The calculation process is as follows: first, determine the amount of water carried over by the aggregate based on the aggregate moisture content; then, subtract the amount of water carried over by the aggregate from the target total water consumption to obtain the new actual added water consumption. If the new actual added water consumption is less than 0, the water-cement ratio increase instruction is invalid, and only the admixture dosage is adjusted. After the admixture dosage correction, if it exceeds the upper or lower limit recorded in the admixture adjustable range, the corresponding boundary value is applied.

[0138] Each time a step-by-step correction is completed, the slump retention curve and the early strength growth curve are updated again.

[0139] When updating the slump retention curve, the slump loss for subsequent time periods is recalculated based on the corrected water-cement ratio, admixture dosage, and environmental change trends. For measured time periods that have already occurred, the on-site test results are retained, while for time periods that have not occurred, the corrected predicted slump value is used.

[0140] When updating the early strength growth curve, the strength development rate is corrected according to the direction of water-cement ratio adjustment: when the water-cement ratio increases, the subsequent strength development rate decreases by 5% to 15%; when the water-cement ratio decreases, the subsequent strength development rate increases by 5% to 12%; when adjusting the admixture dosage alone, the change in the subsequent strength development rate is limited to within ±5%.

[0141] The updated two curves will continue to be used for the next round of on-site feedback parameter comparison.

[0142] The revised mix design parameter set includes the revised water-cement ratio, admixture dosage, revised water parameters, corresponding slump retention curve, corresponding early strength growth curve, and a mix design correction instruction sequence. When outputting the updated mix design, the applicable batch and verification conditions are also provided.

[0143] The verification conditions include that the slump test result at the mixer falls within the allowable fluctuation range for construction, the real-time adaptation evaluation value recovers to above 70 points, and the fluctuation characteristic of the mixing power does not continuously exceed 20%. When all the above conditions are met, the updated mix proportion scheme is used for subsequent mixing batches; when any condition is not met, the modified mix proportion parameter set is retained and revised again in the next observation interval.

[0144] Example 2, please refer to Figure 2As shown in this embodiment, the artificial intelligence-based concrete mix design system includes: The raw material data acquisition module acquires raw material information and environmental parameters of the target construction scenario, and performs preprocessing to form a basic mix proportion dataset. The raw material information includes cement type, aggregate gradation and moisture content, admixture type and dosage range, and environmental parameters include temperature and humidity. The feature extraction module, based on the basic mix design dataset and combined with slump evolution, setting time and strength results from historical construction records, extracts performance response feature sequences. The virtual mix design module generates a set of candidate mix proportions based on the performance response characteristic sequence, and performs virtual mix design simulations on each candidate mix proportion to obtain the corresponding slump retention curve and early strength growth curve. The optimal mix ratio screening module comprehensively evaluates the slump retention curve and the early strength growth curve, calculates the fit score index of each candidate mix ratio, and screens out the optimal mix ratio. The comparison module collects mixing power, slump at the outlet and environmental changes in real time during the actual mixing process, forms on-site feedback parameters, and compares them with the adaptation score index corresponding to the optimal ratio. The proportioning update module automatically adjusts the water-cement ratio and admixture dosage when the on-site feedback parameters deviate from the preset threshold, and outputs the updated proportioning scheme.

[0145] The above description is merely a specific embodiment of this application, but the scope of protection of this application 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 this application should be included within the scope of protection of this application.

Claims

1. A concrete mix design optimization method based on artificial intelligence, characterized in that, include: The raw material information and environmental parameters of the target construction scene are obtained and preprocessed to form a basic mix proportion dataset. The raw material information includes cement type, aggregate gradation and moisture content, admixture type and dosage range, and environmental parameters include temperature and humidity. Based on the aforementioned basic mix design dataset, and combined with slump evolution, setting time, and strength results from historical construction records, a performance response feature sequence is extracted. Based on the performance response characteristic sequence, a candidate mix design set is generated, and virtual trial mix design is performed on each candidate mix design to obtain the corresponding slump retention curve and early strength growth curve. The slump retention curve and the early strength growth curve are comprehensively evaluated, the fit score index of each candidate ratio is calculated, and the optimal ratio is selected. During the actual mixing process, the mixing power, slump at the outlet and environmental changes are collected in real time to form on-site feedback parameters, which are then compared with the matching score index corresponding to the optimal mix ratio. When the on-site feedback parameters deviate from the preset threshold, the water-cement ratio and the amount of admixture are automatically adjusted, and an updated mix design is output.

2. The artificial intelligence-based concrete mix design optimization method according to claim 1, characterized in that, The steps to create a basic formula dataset include: The effective water usage is calculated based on the aggregate moisture content, and the water content brought in by the aggregate is combined with the design mixing water for verification to obtain the corrected water usage parameters. The aggregate gradation is merged into particle size ranges, and the boundaries of admixture type and dosage ranges are sorted out to form continuous gradation records and adjustable admixture range records, respectively. The corrected water parameters, gradation continuous records, admixture adjustable range records, and environmental status records are summarized according to the same mixing batch to form a basic mix proportion dataset.

3. The artificial intelligence-based concrete mix design optimization method according to claim 2, characterized in that, The steps for extracting performance response feature sequences include: The basic mix design dataset and historical construction records are aligned in time according to mixing batch, mixing start time and discharge time to construct a multi-stage response record unit that includes slump evolution, setting time and strength results. The multi-stage response recording unit is segmented and the difference is analyzed to identify the slump drop interval and condensation inflection point, forming a key response segment identifier set. The intensity results are back-screened based on the key response section identifier set to extract the performance response feature sequence that matches the slump stability and condensation coordination.

4. The artificial intelligence-based concrete mix design optimization method according to claim 3, characterized in that, The steps to form a set of critical response segment identifiers include: Calculate the slump loss between adjacent collection times and the slump loss per unit time. When both reach the preset loss condition at the same time, determine the corresponding time period as the slump drop interval. The slump evolution before the initial setting time is divided into continuous time periods. When the slump loss intensity of the later time period is higher than that of the previous time period and continues to reach the preset duration, the corresponding starting time is determined as the setting inflection point. The slump drop intervals and setting transition points are identified according to mixing batch, start time, end time, and section type to obtain a set of key response section identifiers.

5. The artificial intelligence-based concrete mix design optimization method according to claim 1, characterized in that, The steps for generating a candidate mix design set include: The slump decay rate constraint range and the early intensity development rate constraint range are extracted from the performance response feature sequence to form a dual performance boundary. Historical mix proportions that meet the target strength level and whose slump drop interval does not cross the solidification inflection point are selected as seed mix proportions, and verified using the aforementioned dual performance boundaries to form an initial candidate mix proportion group; The initial candidate formulation group is perturbed and expanded according to the linkage adjustment relationship between water-cement ratio and admixture dosage to form a candidate formulation set.

6. The artificial intelligence-based concrete mix design optimization method according to claim 5, characterized in that, The steps to obtain the slump retention curve include: According to the mixing sequence and time interval corresponding to the target construction scenario, the process of each candidate proportion in the candidate proportion set is reproduced; The target slump value at the outlet is used as the initial value, and the foundation loss is determined segment by segment according to the slump decay rate per unit time. When the current time segment enters the slump drop interval or is within the preset time range before the solidification inflection point, the base loss is corrected in segments, and the slump values ​​at each time point are connected in chronological order to obtain the slump maintenance curve.

7. The artificial intelligence-based concrete mix design optimization method according to claim 6, characterized in that, The steps to obtain the early intensity growth curve include: The mixing state of each candidate ratio is divided into time periods according to a preset time interval, forming a reaction stage division sequence covering the period before and after initial setting. Within each reaction stage, reaction activity indicators are extracted based on changes in mixing temperature rise and water consumption, and these are accumulated in chronological order to form a reaction cumulative index sequence. Based on the cumulative response index sequence, the accelerated growth segment, the steady growth segment, and the stable segment of intensity formation are identified, and the intensity change trend of each segment is spliced ​​together in chronological order to obtain the early intensity growth curve.

8. The artificial intelligence-based concrete mix design optimization method according to claim 1, characterized in that, The steps for selecting the optimal ratio include: A piecewise stationarity analysis was performed on the slump retention curve to extract a sequence of slump stability segments that meet the requirements of duration and lower limit of construction slump. The early intensity growth curve is analyzed for stage growth rate to form a sequence of intensity development segments that include rapid intensity development intervals and slow intensity growth intervals; Based on the temporal connection between the slump stability zone sequence and the intensity development zone sequence, a performance synergy evaluation value is generated, and the matching score index and optimal ratio for each candidate mix ratio are determined accordingly.

9. The artificial intelligence-based concrete mix design method according to claim 8, characterized in that, The steps to output the updated mix design include: The on-site feedback parameter sequence is aligned with the slump retention curve corresponding to the optimal mix ratio in time. Based on the difference between the measured slump value and the expected slump value, a slump deviation characteristic sequence is formed. Combined with the environmental change trend and the mixing power fluctuation characteristics, a real-time adaptation evaluation value is obtained. When the real-time adaptation evaluation value is lower than the preset threshold, or when the slump deviation feature sequence shows continuous deviation, it is determined that there is insufficient fluidity, excessive fluidity, abnormal mixing resistance, or environmentally induced deviation, thus forming a deviation type discrimination result. Based on the deviation type identification results, the water-cement ratio and admixture dosage are adjusted step by step, and the slump retention curve and early strength growth curve are updated simultaneously. The updated mix proportion scheme is output, which includes the corrected water parameters, the corrected water-cement ratio, the corrected admixture dosage, and the verification conditions.

10. An artificial intelligence-based concrete mix design optimization system, used to implement the artificial intelligence-based concrete mix design optimization method according to any one of claims 1-9, characterized in that, include: The raw material data acquisition module acquires raw material information and environmental parameters of the target construction scenario, and performs preprocessing to form a basic mix proportion dataset. The raw material information includes cement type, aggregate gradation and moisture content, admixture type and dosage range, and environmental parameters include temperature and humidity. The feature extraction module, based on the basic mix design dataset and combined with slump evolution, setting time and strength results from historical construction records, extracts performance response feature sequences. The virtual mix design module generates a set of candidate mix proportions based on the performance response characteristic sequence, and performs virtual mix design simulations on each candidate mix proportion to obtain the corresponding slump retention curve and early strength growth curve. The optimal mix ratio screening module comprehensively evaluates the slump retention curve and the early strength growth curve, calculates the fit score index of each candidate mix ratio, and screens out the optimal mix ratio. The comparison module collects mixing power, slump at the outlet and environmental changes in real time during the actual mixing process, forms on-site feedback parameters, and compares them with the adaptation score index corresponding to the optimal ratio. The proportioning update module automatically adjusts the water-cement ratio and admixture dosage when the on-site feedback parameters deviate from the preset threshold, and outputs the updated proportioning scheme.