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

Abstract

This invention relates to the field of precast concrete mix design technology, and discloses an artificial intelligence-based method and system for optimizing precast concrete mix design. The system includes a mix design scheme acquisition module, which categorizes material proportions according to component engineering requirements to obtain a first mix design scheme; an environmental parameter acquisition module, which collects real-time environmental data from the production area; a performance benchmark prediction module, which calculates benchmark parameters based on the material performance characteristic matrix in the first scheme to determine a first performance benchmark vector; an environmental evolution prediction module, which predicts trends based on real-time data to obtain a multi-stage production environment parameter sequence; a performance deviation assessment module, which combines the benchmark vector with a prediction model to obtain multi-stage performance deviation vectors; an optimization decision generation module, which optimizes mix design parameters and generates an optimization strategy; and a mix design control execution module, which adjusts material batching equipment parameters according to the strategy. The system adapts to dynamic production environments, ensuring accurate mix design and stable quality.

Description

Technical Field

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

[0002] In the field of precast concrete component production, the rationality of material proportions directly affects the quality, cost, and production efficiency of the components. Traditional precast concrete mix design relies heavily on the experience of technical personnel, determining the final mix proportion by referring to historical engineering data or industry standard formulas, combined with manual trial mixing adjustments. This approach has significant limitations: the manual trial mixing process is time-consuming, requiring repeated adjustments to material dosages and performance testing, making it difficult to meet the efficiency requirements of modern production; the reference value of historical data is limited by its completeness and timeliness, and when project requirements change or new materials are used, existing experience cannot accurately match actual needs, easily leading to a disconnect between the mix design and the actual project; environmental factors are often considered rather superficially during the manual design process, usually only referring to fixed environmental parameter ranges while ignoring the dynamic changes in real-time environmental conditions in the production area, such as fluctuations in temperature, humidity, and wind speed, which directly affect the mixing, setting, and hardening processes of concrete, thus causing deviations between actual performance and design expectations.

[0003] With the increasing demands for quality in precast concrete components and the widespread application of new building materials in the construction industry, the drawbacks of traditional mix design methods are becoming increasingly apparent. In some projects, unreasonable mix design has resulted in key performance indicators such as concrete strength and durability failing to meet requirements, necessitating rework. This not only increases production costs but also delays project progress. To ensure performance compliance, some companies adopt conservative mix design strategies, excessively increasing the use of expensive materials such as cement, leading to resource waste and contradicting current trends in green building development. Furthermore, while some existing mix optimization methods incorporate simple computer-aided calculations, they remain at the level of static parameter analysis, failing to achieve dynamic responses to environmental changes throughout the production process or to adjust mix parameters in advance based on environmental evolution trends. This fails to fundamentally address the impact of environmental factors on concrete performance. In large-scale precast concrete production scenarios, environmental conditions vary significantly across different production stages. Traditional methods cannot achieve precise control of mix parameters at each stage, resulting in insufficient component quality stability and hindering the high-quality development of the precast concrete industry. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for optimizing the mix proportion of precast concrete based on artificial intelligence, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an artificial intelligence-based precast concrete mix design optimization system, the system comprising:

[0006] The mix design scheme acquisition module is used to divide the material proportion levels according to the engineering requirements of precast concrete components and obtain the first precast concrete mix design scheme.

[0007] The environmental parameter acquisition module is used to collect real-time environmental monitoring data of the precast concrete production area;

[0008] The performance benchmark prediction module is used to calculate the benchmark performance parameters based on the material performance characteristic matrix in the first precast concrete mix design scheme, and determine the first precast concrete performance benchmark vector.

[0009] The environmental evolution prediction module is used to predict the environmental evolution trend of the precast concrete production area based on the real-time environmental monitoring data, and obtain a multi-stage production environment parameter sequence.

[0010] The performance deviation assessment module is used to predict the performance deviation of the multi-stage production environment parameter sequence based on the first precast concrete performance benchmark vector and the precast concrete performance prediction model, and generate a multi-stage performance deviation vector.

[0011] The optimization decision generation module is used to intelligently optimize and adjust the material proportioning parameters according to the multi-stage performance deviation vector, and generate a precast concrete proportioning optimization strategy.

[0012] The proportioning control execution module is used to adjust the parameters of the batching equipment in the precast concrete production area according to the precast concrete proportioning optimization strategy.

[0013] Preferably, the performance deviation evaluation module includes:

[0014] An environmental stage matching unit is used to perform a time-series mapping of the multi-stage production environment parameter sequence with each production stage in the first precast concrete mix design scheme.

[0015] The predictive performance calculation unit is used to input the material ratio parameters and environmental parameters corresponding to each production stage into the precast concrete performance prediction model to obtain the predictive performance vector for each production stage.

[0016] The deviation analysis unit is used to compare the predicted performance vectors of each production stage with the first precast concrete performance benchmark vector to generate a performance deviation vector for each production stage.

[0017] Preferably, the predictive performance calculation unit specifically includes:

[0018] The flow performance prediction subunit is used to process the material proportioning parameters and the environmental parameters through the precast concrete flow performance prediction model to obtain the flow performance prediction value.

[0019] The strength development prediction subunit is used to process the material proportioning parameters and the environmental parameters through the precast concrete strength development prediction model to obtain the strength development prediction value.

[0020] The durability performance prediction subunit is used to process the material proportioning parameters and the environmental parameters through the precast concrete durability performance prediction model to obtain the durability performance prediction value.

[0021] A vector construction subunit is used to construct the predicted performance vector based on the predicted flow performance value, the predicted strength development value, and the predicted durability performance value.

[0022] Preferably, the optimization decision generation module includes:

[0023] A deviation weight configuration unit is used to obtain precast concrete performance deviation weight configuration information, which includes flow performance deviation weight, strength development deviation weight and durability performance deviation weight.

[0024] The comprehensive deviation calculation unit is used to perform weighted calculation on the performance deviation vector of each production stage according to the precast concrete performance deviation weight configuration information to obtain the comprehensive performance deviation coefficient of each production stage.

[0025] The decision-triggered judgment unit is used to compare the comprehensive performance deviation coefficient of each production stage with the preset performance deviation threshold.

[0026] The material ratio optimization decision unit is used to generate a material ratio optimization decision for a production stage based on the performance deviation vector of that production stage when the comprehensive performance deviation coefficient of a certain production stage is greater than or equal to the preset performance deviation threshold.

[0027] Preferably, the ratio optimization decision unit includes:

[0028] An initial decision space construction sub-unit is used to make adjustment decisions on the material ratio parameters based on the performance deviation vector, and to generate an initial ratio adjustment decision set that meets the preset number of decisions;

[0029] The decision optimization and screening subunit is used to optimize and screen the initial mix proportion adjustment decision set through the precast concrete performance prediction model and the precast concrete performance deviation weight configuration information to obtain an optimized mix proportion adjustment decision set.

[0030] The decision space extension subunit is used to perform mutation and expansion processing on the optimized ratio adjustment decision set to generate an extended ratio adjustment decision set.

[0031] The optimal decision determination subunit is used to select the proportion adjustment decision with the smallest performance deviation from the extended proportion adjustment decision set as the final proportion optimization decision.

[0032] Preferably, the decision optimization screening subunit specifically includes:

[0033] The decision effect prediction component is used to predict changes in environmental parameters for each initial ratio adjustment decision and obtain predicted environmental parameter change data.

[0034] A performance prediction component is used to input the adjusted material ratio parameters and the predicted environmental parameter change data into the precast concrete performance prediction model to obtain a decision prediction performance vector.

[0035] A deviation calculation component is used to perform weighted calculations on the decision prediction performance vector based on the precast concrete performance deviation weight configuration information to obtain the decision prediction performance deviation coefficient.

[0036] The filtering execution component is used to retain the proportion adjustment decisions whose decision prediction performance deviation coefficient is less than the preset performance deviation threshold in the optimized proportion adjustment decision set.

[0037] Preferably, the decision space extension subunit includes:

[0038] The variation value assessment component is used to construct a decision variation value assessment system based on historical decision variation effect record data.

[0039] The coefficient of variation calculation component is used to input the predictive performance deviation coefficient of each optimized ratio adjustment decision into the decision variation value assessment system to obtain the variation value coefficient of each decision.

[0040] A mutation operation component is used to perform mutation operations on the optimized ratio adjustment decision set according to the mutation value coefficient, and generate a primary mutation decision set.

[0041] The variation decision screening component is used to screen the primary variation decision set through the precast concrete performance prediction model and the precast concrete performance deviation weight configuration information to obtain an optimized variation decision set.

[0042] An extension component is used to merge the optimized variation decision set into the optimized ratio adjustment decision set to form the extended ratio adjustment decision set.

[0043] Preferably, the performance benchmark prediction module includes:

[0044] The sample data retrieval unit is used to retrieve historical mix proportion performance data similar to the first precast concrete mix proportion design scheme from the historical precast concrete mix proportion database based on material performance characteristic indicators.

[0045] The benchmark value calculation unit is used to perform central trend analysis on the retrieved historical proportion performance data and calculate the benchmark value of each material performance characteristic index.

[0046] The benchmark vector construction unit is used to construct the first precast concrete performance benchmark vector based on the benchmark value of each material performance characteristic index.

[0047] Preferably, the material performance characteristics include flow performance indicators, strength development indicators, and durability performance indicators.

[0048] Preferably, the present invention also includes an artificial intelligence-based method for optimizing the mix proportion of precast concrete, the method comprising all the modules and process flow of the artificial intelligence-based precast concrete mix proportion optimization system described above.

[0049] Compared with the prior art, the beneficial effects of the present invention are:

[0050] The mix design scheme acquisition module can divide the material proportions into levels according to the specific engineering requirements of precast concrete components. Compared with the traditional manual design method that relies on historical experience, it can achieve precise breakdown of engineering requirements, making the mix design more targeted and avoiding the situation where the mix scheme is out of touch with the actual project due to insufficient understanding of the requirements. This makes the initial mix scheme more in line with the performance requirements of the components.

[0051] The environmental parameter acquisition module collects environmental monitoring data from the production area in real time, breaking the limitations of the traditional rough consideration of environmental factors in the proportion design. It can dynamically capture changes in environmental parameters such as temperature, humidity, and wind speed, providing comprehensive and real-time environmental data support for subsequent proportion optimization. It no longer relies on fixed environmental parameter ranges for design, allowing proportion adjustments to be closely combined with the actual environmental conditions of the production site, reducing performance deviations caused by ignoring or lagging environmental parameters.

[0052] The performance benchmark prediction module calculates benchmark performance parameters based on the material performance characteristic matrix in the first mix design scheme and constructs the first precast concrete performance benchmark vector. Compared with the traditional method of determining the performance benchmark only after manual trial mixing, the performance benchmark is determined in advance through matrix calculation, which greatly shortens the time for determining the performance benchmark. At the same time, the analysis using the material performance characteristic matrix can more comprehensively consider the correlation between the properties of different materials, making the determination of the performance benchmark more scientific and accurate, and providing a reliable reference standard for subsequent performance deviation evaluation.

[0053] The environmental evolution prediction module predicts the environmental evolution trend of the production area based on real-time environmental monitoring data, and obtains a multi-stage production environment parameter sequence. This changes the traditional method of statically viewing environmental factors and enables the prediction of environmental changes at different production stages. This allows the ratio optimization to no longer be limited to the response to the current environment, but to formulate adjustment strategies in advance based on future environmental evolution trends. This avoids the problem of the ratio parameters not being able to be adjusted in time due to sudden environmental changes, and provides forward-looking environmental data support for subsequent multi-stage performance deviation prediction and ratio optimization.

[0054] The performance deviation assessment module, combined with the first performance benchmark vector, uses a precast concrete performance prediction model to predict performance deviations in a multi-stage production environment parameter sequence, generating a multi-stage performance deviation vector. Compared to the traditional method of discovering performance deviations only after manual inspection, this module enables early prediction of performance deviations. It allows for the early detection of potential performance deviations at different environmental stages during concrete production, rather than discovering problems only after component molding. This facilitates early adjustment measures and reduces rework or quality issues caused by performance deviations. Furthermore, the application of the performance prediction model allows deviation prediction to fully integrate the correlation between material properties and environmental changes, improving the accuracy of deviation prediction.

[0055] The optimization decision generation module intelligently optimizes and adjusts material proportioning parameters based on multi-stage performance deviation vectors, generating proportion optimization strategies. Compared to traditional manual trial mixing and adjustment, the intelligent optimization algorithm can quickly traverse multiple combinations of proportioning parameters to find a better adjustment scheme, avoiding the blindness and limitations of manual adjustment. At the same time, it formulates adjustment strategies for multi-stage performance deviations, achieving precise optimization of proportioning parameters at different production stages. This allows the proportioning parameters to dynamically adapt to environmental changes and performance requirements, moving away from a single fixed proportioning scheme and improving the flexibility and applicability of the mix design.

[0056] The proportioning control execution module adjusts the parameters of the batching equipment according to the optimization strategy, realizing the automatic implementation of the proportioning optimization strategy without manual intervention to adjust equipment parameters. This reduces errors and delays caused by manual operation, improves the automation and efficiency of the production process, and ensures the timeliness and accuracy of proportioning parameter adjustments. This allows the quality of the produced precast concrete components to remain stable, avoiding quality fluctuations caused by untimely or inaccurate equipment parameter adjustments. This helps to improve the overall quality level of precast concrete components, meets the needs of modern production and green building development, and promotes technological progress and high-quality development in the precast concrete industry. Attached Figure Description

[0057] Figure 1 This is a timing diagram of the artificial intelligence-based precast concrete mix design optimization system described in this invention.

[0058] Figure 2 A flowchart illustrating the operation of the performance deviation assessment module;

[0059] Figure 3 A flowchart for optimizing the decision generation module. Detailed Implementation

[0060] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0061] Please see Figure 1 This invention provides an artificial intelligence-based precast concrete mix design optimization system. The system integrates materials science, environmental monitoring, and machine learning technologies to dynamically adjust the mix design during precast concrete production. The system acquires an initial mix design scheme based on engineering requirements, collects real-time environmental data and predicts environmental evolution trends, evaluates performance deviations under different environmental conditions based on a performance prediction model, and generates a mix design adjustment strategy and controls the execution device through an intelligent optimization algorithm. The system includes a mix design scheme acquisition module, an environmental parameter acquisition module, a performance benchmark prediction module, an environmental evolution prediction module, a performance deviation evaluation module, an optimization decision generation module, and a mix design control execution module. These modules work collaboratively to form a closed-loop control logic, enabling them to adapt to changes in the production environment and maintain the stability of concrete performance.

[0062] Example 1: See Figure 2 The performance deviation assessment module achieves accurate performance prediction by matching multi-stage environmental parameters with the production stages in a time sequence. The environmental stage matching unit establishes a mapping relationship between the multi-stage production environmental parameter sequence and the production stages in the mix design scheme according to the time axis. Each stage has a set of environmental parameters, which includes real-time monitoring data such as temperature, humidity, wind speed, and atmospheric pressure. The mapping process uses a time window alignment algorithm to ensure that the starting point of the production stage matches the time stamp of the environmental parameter collection. The predicted performance calculation unit receives the time-series matched environmental parameters and the corresponding material proportion parameters for the stage. The material proportion parameters cover core variables such as cement dosage, water-cement ratio, aggregate gradation, and admixture dosage. These parameters, along with the environmental parameters, are input into the precast concrete performance prediction model for processing.

[0063] The precast concrete performance prediction model consists of three parallel sub-prediction units. The flowability prediction sub-unit uses a deep neural network structure to process input parameters. The network contains five hidden layers and employs the ReLU activation function. The input layer receives twenty-seven feature dimensions, including ambient temperature, cement fineness, and water-reducing agent type. The output layer generates two key flowability indicators: slump prediction and spread prediction. The strength development prediction sub-unit is built based on a gradient boosting regression tree model. This model is trained using 300 sets of historical mix design experimental data. During the feature engineering stage, an interaction variable between ambient humidity and cement hydration heat is introduced. The model outputs predicted compressive strength and flexural strength values ​​for three ages: 3 days, 7 days, and 28 days. The durability performance prediction sub-unit runs a random forest algorithm framework. An ensemble model composed of two hundred decision trees analyzes the correlation between chloride ion permeability coefficient and carbonation depth. Feature importance analysis shows that aggregate mud content and ambient carbon dioxide concentration have the highest weighting on the prediction results. This sub-unit ultimately outputs predicted chloride ion diffusion coefficient and carbonation depth values.

[0064] The vector construction subunit standardizes the three types of predicted values: flowability prediction values ​​use slump values ​​in millimeters, strength development prediction values ​​retain compressive strength data in megapascals, and durability prediction values ​​are converted into diffusion coefficients expressed in scientific notation. The standardized data are combined into a six-dimensional vector in a fixed order of flowability, strength, and durability indices, forming a complete predicted performance vector for this production stage. The deviation analysis unit uses the first precast concrete performance benchmark vector as a reference. This benchmark vector contains flowability benchmark values, strength benchmark values, and durability benchmark values ​​of the same dimension. The calculation process performs component-by-component difference operations between the predicted performance vector and the benchmark vector: flowability deviation is calculated using absolute differences, strength deviation using relative percentage differences, and durability deviation using logarithmic differences. The final generated performance deviation vector retains the original dimensional structure but the numerical values ​​represent the magnitude of the deviation.

[0065] The performance deviation assessment module establishes a dynamic feedback mechanism. When the environmental evolution prediction module updates the multi-stage production environment parameter sequence, this module automatically triggers a recalculation process. The time-series matching stage adopts a sliding time window mechanism to adapt to environmental parameter updates, with the window width dynamically adjusted according to the initial setting time of concrete. The predictive performance calculation unit has a model hot update function. When a new admixture or special aggregate type is input, it automatically retrieves the corresponding predictive model from the cloud model library to ensure the accuracy of predictions for special material formulations. The deviation analysis unit sets up an abnormal deviation early warning mechanism. When the deviation in any dimension exceeds three times the standard deviation of the historical deviation distribution, the system automatically marks the production stage as a high-risk node and prioritizes its push to the optimization decision module. The entire performance deviation assessment process adopts a distributed computing architecture. The environmental stage matching unit runs on edge computing devices to achieve real-time response, the predictive performance calculation unit is deployed on a cloud computing platform to complete complex model calculations, and the deviation analysis unit executes on a local server to ensure data security. An internal data verification mechanism is established within the module. When the input environmental parameters exceed the sensor range or there are logical contradictions in the material ratio parameters, the data cleaning program is automatically triggered, and historical averages are used to replace outliers. At the same time, a data anomaly alarm is sent to the monitoring terminal. The performance deviation vector output format uses JSON structured data encapsulation, which includes metadata information such as timestamp, production stage number, deviation values ​​of each dimension and data confidence level, for downstream modules to parse and use.

[0066] Example 2: See Figure 3 The optimized decision generation module implements an intelligent decision-making triggering mechanism based on the precast concrete performance deviation weight configuration information. The deviation weight configuration unit loads a preset set of weight parameters from the system configuration library. This set is classified and set according to the technical requirements of the engineering contract and the service environment of the components. For example, the configuration scheme for bridge precast components is 0.5 for automatic loading strength development deviation, 0.3 for durability performance deviation, and 0.2 for flow performance deviation, while the configuration for building exterior wall panels is differentiated with 0.4 for durability performance deviation, 0.4 for strength development deviation, and 0.2 for flow performance deviation. The weight parameters are stored using an encrypted data structure, and each call requires authorization verification through the project number. The configuration interface allows the chief engineer of the construction party to adjust the weight ratio online, and the adjustment record is synchronized to the blockchain evidence storage system in real time. The comprehensive deviation calculation unit receives multi-stage performance deviation vectors transmitted by the performance deviation evaluation module. Each vector contains three dimensions of data: fluidity deviation, strength deviation, and durability deviation. The calculation process multiplies the deviation of each dimension by its corresponding weighting coefficient: the fluidity deviation is multiplied by the fluidity performance deviation weight, the strength deviation by the strength development deviation weight, and the durability deviation by the durability performance deviation weight. The three products are then added together to obtain the comprehensive performance deviation coefficient for that production stage. The calculation process is accelerated using a fixed-point processor, with the processing time for a single vector controlled within 5 milliseconds. The calculation results are appended with a timestamp and stage identifier to form a structured record.

[0067] The decision triggering unit maintains a preset performance deviation threshold parameter table. This table stores threshold data according to concrete strength grade, with a threshold of 0.15 for C30 concrete and 0.08 for C50 concrete. The threshold data originates from the company's quality control standards. A comparator circuit monitors the relationship between the comprehensive performance deviation coefficient and the threshold in real time. When the coefficient value is detected to be greater than or equal to the threshold at a certain stage, a trigger command containing the stage number, deviation type distribution, and exceedance magnitude is immediately sent to the mix design optimization decision unit. The trigger status is transmitted to the central control console via the industrial bus, and the abnormal stage is marked with a flashing red light on the 3D production progress view. Upon receiving the trigger command, the mix design optimization decision unit initiates the decision analysis process. First, it analyzes the component structure of the performance deviation vector to identify the dominant deviation type: when the flowability deviation component accounts for more than 60%, it is determined to be a workability anomaly; when the strength deviation component exceeds the standard for two consecutive stages, it is determined to be a strength development anomaly; and when the absolute value of the durability deviation component is greater than 0.2, it is determined to be a durability risk. For workability anomalies, a water-cement ratio adjustment scheme is generated; for strength development anomalies, cementitious material composition optimization is triggered; and for durability risks, admixture ratio adjustment is initiated. The decision output is encapsulated in XML format, including the cement adjustment amount (…). The module includes executable parameters such as the range of water-reducing agent dosage changes (±0.2%) and mineral admixture replacement rate (±8%). A decision traceability mechanism is established, linking all generated proportion optimization decisions to original environmental monitoring data and performance deviation data, forming a decision traceability chain. When three consecutive decision adjustments are triggered in the same production stage, the system automatically upgrades to a level-two warning state, initiates expert consultation mode, and uploads a complete data package to the remote technical support center. The decision execution effect feedback channel receives the execution results from the batching equipment in real time, compares the actual adjusted parameters with the decision parameters, and triggers a calibration procedure to re-verify sensor accuracy when the deviation exceeds 2%. The entire optimization decision-making process adopts a fault-tolerant design; when the weight configuration unit transmission is interrupted, the most recently valid configuration is automatically called, and the comprehensive deviation calculation unit has a built-in overflow protection mechanism to prevent extreme values ​​from causing system crashes. The decision trigger judgment unit sets a dual-threshold comparison mode, adding a dynamic threshold option in addition to a fixed threshold. The dynamic threshold automatically floats and adjusts based on the historical deviation data of the current month. The allocation optimization decision unit integrates a case reasoning engine. When a new deviation pattern is detected, it automatically searches the historical case library for similar solutions. If no valid case is found, it activates a decision exploration mode based on reinforcement learning.

[0068] The module's operational status is integrated with the factory's MES system via the OPCUA protocol, uploading operational metrics such as the number of decision triggers, the distribution of deviation types, and decision response times in real time. The safety protection system implements tiered access control for modifying decision parameters; ordinary operators can only view the decision results, and any adjustments exceeding the preset safety range must be confirmed with a digital signature by the certified laboratory director. When night mode is activated, decision sensitivity is automatically increased by 20%, compensating for measurement errors from environmental monitoring equipment under low-temperature conditions.

[0069] Example 3: The mix proportion optimization decision unit implements a dynamic adjustment strategy through a multi-stage decision generation and screening mechanism. The initial decision space construction sub-unit receives the performance deviation vector and initiates the decision generation program. This program determines the main adjustment direction based on the component structure of the deviation vector: when fluidity deviation is dominant, a water-cement ratio adjustment scheme is generated; when strength deviation is significant, a cementitious material dosage adjustment scheme is generated; and when durability deviation is prominent, an admixture ratio correction scheme is formulated. Each adjustment scheme includes core parameters such as changes in cement dosage, admixture dosage adjustment, and changes in mineral admixture substitution rate. The parameter variation range is strictly controlled within the allowable fluctuation range of the material specifications. The number of generated decisions is determined through an adaptive algorithm, with a base number of 20. When the comprehensive deviation coefficient exceeds a threshold of 50%, the number is increased to 30, forming the initial mix proportion adjustment decision set.

[0070] The decision optimization and screening subunit performs multiple rounds of evaluation and screening on the initial set. The decision effect prediction component first analyzes the environmental parameter changes that each adjustment decision may cause. For example, increasing the water-reducing agent dosage may lead to a 0.3-0.8 degree Celsius increase in the mixture temperature, while increasing the fly ash dosage may decrease the mixture viscosity by 5-10%. The performance prediction component inputs the adjusted material parameters and predicted environmental change data into the precast concrete performance prediction model to obtain the predicted performance vector corresponding to each decision. This vector includes the adjusted fluidity index, strength development index, and durability index. The deviation calculation component applies the precast concrete performance deviation weight configuration information to perform a weighted calculation on the predicted performance vector. The calculation formula is as follows:

[0071]

[0072] in: Indicates the deviation coefficient of decision prediction performance. , , These represent the weights for deviations in flow performance, strength development, and durability performance, respectively. , , The corresponding predicted performance vectors include flow performance values, strength development values, and durability performance values. , , This represents the reference value in the performance benchmark vector. The selection and execution component sets the acceptance threshold to 0.8 times the original preset performance deviation threshold, and the calculated value is then used. Compare the value with the acceptance threshold and retain all. Decisions with values ​​below the threshold form an optimized allocation adjustment decision set.

[0073] The decision space expansion subunit employs multiple mutation strategies to enhance decision diversity, applying random perturbation operations to each decision in the optimization set: cement usage in Fluctuations within a certain range, with the water-to-binder ratio adjustment range controlled within... Within the specified range, the admixture dosage variation shall not exceed ±0.15%. An adaptive mutation rate mechanism is introduced for the mutation operation: a high mutation rate of 0.3 is used when the number of decisions in the optimization set is less than 10, and a conservative mutation rate of 0.15 is used when the number exceeds 15. Each new decision generated by mutation must pass a consistency check to ensure that the material parameters meet the basic constraints of the mix proportion. The number of decisions after mutation is expanded to 1.5 times that of the original set. The optimal decision determination sub-unit performs final selection from the expanded decision set, first excluding all decisions whose material parameters exceed the allowable range of the construction mix proportion, and then calculating the prediction performance deviation coefficient of the remaining decisions. Value, selection The decision with the minimum value is taken as the final allocation optimization decision. When multiple decisions have the same minimum value... When calculating the value, the decision with the smallest change in cement usage is prioritized, followed by the option with the smallest adjustment in admixture dosage. The final decision output includes a complete list of material proportioning parameter adjustments, with each parameter marked with the baseline value before adjustment and the target value after adjustment, along with data on the predicted performance improvement.

[0074] The entire decision generation process employs a cyclical optimization mechanism. If performance deviations exceed limits after the final decision is executed, the system automatically initiates a new round of decision generation, but the adjustment range is limited to 50% of the previous adjustment. The decision verification phase utilizes a virtual simulation mode. Decisions regarding mix proportion adjustments at critical engineering nodes must be verified through virtual trial mixing using a digital twin model. Only after successful verification can the decisions be sent to the execution equipment. The decision tracking system records data such as the generation time, execution effect, and degree of deviation improvement for each decision, forming a historical database of decision effects for optimizing subsequent decision generation algorithms.

[0075] Example 4: The decision space extension subunit implements value-oriented mutation operations based on the optimized mix proportion adjustment decision set. The mutation value assessment component accesses the historical decision database to extract mutation operation records from the past six months. This database stores key information such as the original decision code, mutation operation type, mutation magnitude, and post-execution deviation improvement rate. The assessment system is constructed using a multi-dimensional analysis framework, classifying mutation operations into three categories based on material parameter types: cement adjustment mutation, aggregate gradation mutation, and admixture mutation. Each type of mutation has a success rate index and an improvement efficiency index. The success rate index counts the proportion of decisions of this type of mutation that were ultimately adopted in historical records. The improvement efficiency index calculates the ratio of the reduction in the comprehensive performance deviation coefficient after the mutation decision is executed to the mutation cost. The mutation value coefficient is calculated using a composite algorithm, with the base value being the product of the success rate and the improvement efficiency, and then dynamically weighted and corrected based on the predicted performance deviation coefficient of the current decision. The mutation coefficient calculation component receives each decision data from the optimized mix proportion adjustment decision set, extracts its predicted performance deviation coefficient δ value, and inputs it into the assessment system. The system automatically matches the adjustment type of the decision and calls the corresponding mutation category's value calculation model. For decisions primarily driven by cement usage adjustments, the system retrieves historical data on cement adjustment variation categories to calculate value coefficients; for aggregate gradation adjustment decisions, it matches aggregate variation category data. The value coefficient output range is set between 0 and 1. When the δ value is higher than 0.2, the value gain mode is automatically activated, multiplying the base value coefficient by the δ value amplification factor. The variation value coefficient calculation results are appended with a timestamp and decision code to form the basic parameters for variation operation instructions.

[0076] The mutation operation component performs differentiated processing based on the mutation value coefficient. Decisions with a value coefficient higher than 0.7 employ a refined mutation strategy: cement usage... Random fluctuations are allowed within a certain range, with admixture dosage variations controlled within ±0.1%. A material compatibility verification mechanism is introduced into the variation operation to automatically detect the compatibility between cement and admixtures. Exploratory variation is implemented for decisions with a value coefficient below 0.3: cement dosage is allowed. The variation in the mineral admixture substitution rate has expanded to ±10%, and the mutation process breaks through the conventional mix proportion constraints while maintaining the minimum total amount of cementitious materials. Each decision generates three mutation schemes to form a primary mutation decision set, with the set size controlled at twice the original optimized decision set. The mutation decision screening component performs a two-stage filtering on the primary mutation decision set. The first stage performs logical rule verification, eliminating decisions with a water-cement ratio higher than 0.5 or lower than 0.3, and excluding schemes where the total amount of fly ash and mineral powder admixtures exceeds 40% of the cementitious materials. The second stage calls the precast concrete performance prediction model for virtual verification, inputting the mutated material parameters and environmental parameter prediction values, obtaining the predicted performance vector, and calculating... Values. Filter and retain. Decisions with values ​​below 0.9 times the preset performance deviation threshold are re-screened when the number of retained decisions is less than 30% of the initial set, with the threshold relaxed to 1.1 times the original threshold. Decisions that pass the screening are marked as effective mutation decisions and are then processed according to... The values ​​are sorted in ascending order to generate an optimized mutation decision set.

[0077] The extended component performs data fusion, combining the optimized mutation decision set with the original optimized proportion adjustment decision set, and using decision code hash value comparison to remove complete duplicates. Decisions with material parameter similarity exceeding 95% are automatically merged, retaining... The version with the lowest value. The expanded set implements capacity control, and when the total number of decisions exceeds 50, a priority sorting mechanism is activated, based on... The top 50 decisions, ranked from lowest to highest, form an expanded allocation adjustment decision set. The set data is stored in a tree structure, with the root node representing the original optimization decision and the child nodes representing the derived variation decisions, facilitating the tracing of the decision evolution path (see Table 1).

[0078] Table 1: Decision Variation Operation Table

[0079]

[0080] The system establishes a mutation knowledge update mechanism, automatically collecting actual performance test data after each decision execution, calculating the true deviation improvement rate, and transmitting it back to the historical decision database. The mutation value assessment system performs monthly parameter calibration, adjusting the success rate and improvement efficiency indicators of various mutation operations based on new data. When the success rate of a mutation for a specific material type falls below 20% for five consecutive times, the system automatically freezes that type of mutation operation and sends a technical alert to the materials laboratory. The mutation operation component has seasonal adjustment parameters, automatically narrowing the mutation range during winter construction to prevent large-span adjustments from causing concrete frost damage. The decision expansion process is monitored in real time; when a single decision continuously generates more than five levels of mutation, a mutation path review is automatically triggered to prevent invalid mutations from consuming computational resources. A mutation path graph is attached to the expanded set output, graphically displaying the parameter evolution trajectory from the original decision to the final mutation decision. The data security module encrypts and stores the mutation decision set, with access permissions bound to the project's security level. For high-security projects, mutation decisions automatically use 256-bit AES encryption to protect core parameters.

[0081] Example 5: The performance benchmark prediction module establishes a performance evaluation benchmark for precast concrete through historical data mining. After receiving the first precast concrete mix design scheme, the sample data retrieval unit initiates a similar case matching process. This scheme includes material characteristic parameters such as cement type and grade, aggregate gradation curve, water-cement ratio range, and admixture type. The retrieval engine converts the material characteristic parameters into feature vector form. The cement characteristic dimension includes sub-parameters such as specific surface area, alkali content, and C3A content; the aggregate characteristic dimension covers attributes such as fineness modulus, mud content, and crushing index; and the admixture characteristic dimension includes indicators such as solid content, chloride ion content, and water reduction rate. The feature vector adopts an eight-dimensional data structure, and each dimension is normalized to eliminate the influence of dimensions. The similarity calculation uses an improved cosine similarity algorithm, setting a similarity threshold of 0.85 as a screening condition. The historical precast concrete mix database uses a distributed architecture for storage, containing data from 2,400 engineering cases completed in the past five years, with each case storing 375 technical parameters. The retrieval process employs multi-level index optimization. The first-level index categorizes cement strength grades, the second-level index is stratified by maximum aggregate size, and the third-level index is partitioned according to ambient temperature range. Cases with successful similarity matches are added to the candidate set. The system automatically excludes cases with quality accidents in the construction records and filters records with incomplete detection data. The candidate set size is controlled between thirty and fifty cases. When there are fewer than twenty matching cases, the similarity threshold is automatically relaxed to 0.75, and the cross-regional database retrieval function is activated.

[0082] The benchmark calculation unit performs statistical processing on the candidate set data. For flow performance indicators, the slump test values ​​are primarily analyzed. After removing outlier records with test times exceeding 30 minutes, the arithmetic mean of the slump data from all cases is calculated as the benchmark value. The strength development indicator focuses on 28-day compressive strength data, using the median statistic to avoid interference from extreme values. Samples with curing temperatures exceeding the standard range of 20±2℃ are excluded during the calculation process. The durability performance indicator focuses on the chloride ion diffusion coefficient. The mode is calculated after transforming the test value using the natural logarithm, and then restored to the benchmark value through exponential calculation. All statistical calculations implement a triple verification mechanism: the first round excludes outliers outside the range of ±3 standard deviations; the second round verifies the normality of the data distribution; and the third round uses bootstrap sampling to verify stability. The benchmark vector construction unit receives three core indicators: flow performance benchmark, strength development benchmark, and durability performance benchmark. The flow performance benchmark integrates the slump and spread benchmarks to form a two-dimensional sub-vector. The strength development benchmark includes compressive strength benchmarks for three ages (3 days, 7 days, and 28 days) to form a three-dimensional sub-vector. The durability performance benchmark integrates the chloride ion diffusion coefficient benchmark and the carbonization depth benchmark to form a two-dimensional sub-vector. These three sub-vectors are concatenated to construct a seven-dimensional performance benchmark vector. Vector elements retain two decimal places of precision, and the units of measurement for each dimension are labeled. The vector data structure is encapsulated in JSON format with timestamps, and includes the project number set and case quantity information of the data source.

[0083] The module establishes a dynamic update mechanism, automatically triggering a baseline recalculation process after a new project completes quality acceptance. The update cycle is set to quarterly rolling updates. The data reliability assessment system assigns a confidence level to each baseline value: baseline values ​​with more than forty cases are marked as Level A confidence, those with twenty to forty cases are marked as Level B confidence, and those with fewer than twenty cases are marked as Level C confidence requiring manual review. An anomaly data processing channel monitors historical database updates; when the deviation between the detected value of a new case and the existing baseline value exceeds 25%, a data anomaly alarm is automatically triggered, prompting technicians to verify the consistency of the testing methods. The performance baseline vector output interface is equipped with multiple protections: baseline vector data packets are transmitted via the OPCDA protocol during industrial field use, and a VPN encrypted channel is enabled for remote access. The baseline value version management system records the vector change trajectory during each update and supports historical baseline value backtracking query functionality. In special material formulation scenarios, when similar case matching fails, the prediction mode based on material composition calculation is activated. The hydration heat model of cementitious materials and the aggregate interface effect model are called to perform theoretical calculations to generate temporary benchmark values. The temporary benchmark values ​​can only be officially entered into the database after laboratory trial mixing verification.

[0084] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0085] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An artificial intelligence-based precast concrete proportioning optimization system, characterized by, The system comprises: The mixing ratio design scheme acquisition module is configured to perform material mixing ratio level division according to engineering requirements of the prefabricated concrete component, and obtain a first prefabricated concrete mixing ratio design scheme; The environmental parameter acquisition module is configured to acquire real-time environmental monitoring data of a prefabricated concrete production area; The performance benchmark prediction module is configured to calculate a benchmark performance parameter according to a material performance characteristic matrix in the first prefabricated concrete mixing ratio design scheme, and determine a first prefabricated concrete performance benchmark vector; The environmental evolution prediction module is configured to predict an environmental evolution trend of the prefabricated concrete production area based on the real-time environmental monitoring data, and obtain a multi-stage production environmental parameter sequence; The performance deviation evaluation module is configured to predict a performance deviation of the multi-stage production environmental parameter sequence based on the first prefabricated concrete performance benchmark vector and a prefabricated concrete performance prediction model, and generate a multi-stage performance deviation vector; The performance deviation evaluation module comprises: The environmental stage matching unit is configured to time-correspond the multi-stage production environmental parameter sequence with each production stage in the first prefabricated concrete mixing ratio design scheme; The predicted performance calculation unit is configured to input material mixing ratio parameters and environmental parameters corresponding to each production stage into the prefabricated concrete performance prediction model, and obtain a predicted performance vector of each production stage; The deviation analysis unit is configured to compare the predicted performance vector of each production stage with the first prefabricated concrete performance benchmark vector, and generate a performance deviation vector of each production stage; The optimization decision generation module is configured to intelligently optimize and adjust the material mixing ratio parameters based on the multi-stage performance deviation vector, and generate a prefabricated concrete mixing ratio optimization strategy; The optimization decision generation module comprises: The deviation weight configuration unit is configured to acquire prefabricated concrete performance deviation weight configuration information, wherein the prefabricated concrete performance deviation weight configuration information comprises a flow performance deviation weight, a strength development deviation weight, and a durability performance deviation weight; The comprehensive deviation calculation unit is configured to weight calculate the performance deviation vector of each production stage based on the prefabricated concrete performance deviation weight configuration information, and obtain a comprehensive performance deviation coefficient of each production stage; The decision trigger judgment unit is configured to compare the comprehensive performance deviation coefficient of each production stage with a preset performance deviation threshold; The mixing ratio optimization decision unit is configured to generate a material mixing ratio optimization decision of a production stage according to a performance deviation vector of the production stage when the comprehensive performance deviation coefficient of the production stage is greater than or equal to the preset performance deviation threshold; The mixing ratio control execution module is configured to adjust parameters of a batching equipment of the prefabricated concrete production area according to the prefabricated concrete mixing ratio optimization strategy.

2. The artificial intelligence-based ready-mix concrete proportioning optimization system of claim 1, wherein, The predicted performance calculation unit specifically comprises: The flow performance prediction subunit is configured to process the material mixing ratio parameters and the environmental parameters through a prefabricated concrete flow performance prediction model, and obtain a flow performance prediction value; The strength development prediction subunit is configured to process the material mixing ratio parameters and the environmental parameters through a prefabricated concrete strength development prediction model, and obtain a strength development prediction value; a durability performance prediction subunit configured to process the material proportioning parameters and the environmental parameters by using a precast concrete durability performance prediction model to obtain a durability performance prediction value; a vector construction subunit configured to construct the performance prediction vector according to the flow performance prediction value, the strength development prediction value and the durability performance prediction value. 3.The artificial intelligence-based ready-mixed concrete proportioning optimization system of claim 1, wherein, The proportioning optimization decision unit comprises: an initial decision space construction subunit configured to make adjustment decisions on the material proportioning parameters according to the performance deviation vector to generate an initial proportioning adjustment decision set satisfying a preset decision quantity; a decision optimization screening subunit configured to perform optimization screening on the initial proportioning adjustment decision set by using the precast concrete performance prediction model and the precast concrete performance deviation weight configuration information to obtain an optimized proportioning adjustment decision set; a decision space expansion subunit configured to perform mutation expansion processing on the optimized proportioning adjustment decision set to generate an expanded proportioning adjustment decision set; an optimal decision determination subunit configured to select a proportioning adjustment decision with the minimum performance deviation from the expanded proportioning adjustment decision set as a final proportioning optimization decision.

4. The artificial intelligence-based ready-mix concrete proportioning optimization system of claim 3, wherein, The decision optimization screening subunit specifically comprises: a decision effect prediction component configured to perform environmental parameter change prediction on each initial proportioning adjustment decision to obtain predicted environmental parameter change data; a performance prediction component configured to input the adjusted material proportioning parameters and the predicted environmental parameter change data into the precast concrete performance prediction model to obtain a decision prediction performance vector; a deviation calculation component configured to perform weighted calculation on the decision prediction performance vector according to the precast concrete performance deviation weight configuration information to obtain a decision prediction performance deviation coefficient; a screening execution component configured to retain a proportioning adjustment decision with a decision prediction performance deviation coefficient less than the preset performance deviation threshold to the optimized proportioning adjustment decision set.

5. The artificial intelligence-based ready-mix concrete proportioning optimization system of claim 3, wherein, The decision space expansion subunit comprises: a mutation value evaluation component configured to construct a decision mutation value evaluation system according to historical decision mutation effect record data; a mutation coefficient calculation component configured to input the prediction performance deviation coefficient of each optimized proportioning adjustment decision into the decision mutation value evaluation system to obtain a mutation value coefficient of each decision; a mutation operation component configured to perform mutation operation on the optimized proportioning adjustment decision set according to the mutation value coefficient to generate a primary mutation decision set; a mutation decision screening component configured to perform screening on the primary mutation decision set by using the precast concrete performance prediction model and the precast concrete performance deviation weight configuration information to obtain an optimized mutation decision set; a set expansion component configured to merge the optimized mutation decision set into the optimized proportioning adjustment decision set to form the expanded proportioning adjustment decision set.

6. The artificial intelligence based ready mix concrete proportioning optimization system as claimed in claim 1 wherein, The performance benchmark prediction module comprises: a sample data retrieval unit configured to retrieve historical proportioning performance data similar to the first precast concrete proportioning design scheme from a historical precast concrete proportioning database according to material performance characteristic indexes; A reference value calculation unit is configured to perform centralized trend analysis on the retrieved historical proportioning performance data, and calculate a reference value of each material performance characteristic index; A reference vector construction unit is configured to construct the first prefabricated concrete performance reference vector according to the reference value of each material performance characteristic index.

7. The artificial intelligence based ready mix concrete proportioning optimization system as claimed in claim 6 wherein, The material performance characteristic indexes include a flow performance index, a strength development index, and a durability performance index.

8. An artificial intelligence-based precast concrete proportioning optimization method, characterized in that, All modules and method processes of the prefabricated concrete proportioning optimization system based on artificial intelligence according to any one of claims 1 to 7 are included.

Images