CN-121075518-B - System and method for optimizing concrete mix proportion based on industrial data analysis

CN121075518BCN 121075518 BCN121075518 BCN 121075518BCN-121075518-B

Abstract

Compared with the prior art which mainly relies on laboratory trial and design and engineer experience, the invention has the defects of long period, high cost and difficulty in coping with raw material fluctuation, the scheme adopts a machine learning-based performance prediction model and multisource real-time industrial data fusion analysis, can rapidly and accurately predict the performance of concrete, remarkably improves the design efficiency and scientificity of the mixing ratio, and realizes the radical transition from experience driving to data driving.

Inventors

GAO YUAN
FENG LIANGPING
HAN XIAO
HU SHAOPENG
REN JINGHUA
GAO ZHIHAO
Xie Enhui

Assignees

中交公路长大桥建设国家工程研究中心有限公司

Dates

Publication Date: 20260512
Application Date: 20250908

Claims (5)

1. The system for optimizing the concrete mixing proportion based on the industrial data analysis is characterized by comprising the following components: The data acquisition and perception module is used for acquiring multisource industrial data of concrete production; the data storage and processing module is used for cleaning, data fusion and characteristic engineering of the multi-source industrial data acquired by the data acquisition and sensing module; The core analysis module is used for predicting the performance of the concrete through a machine learning model, balancing cost, performance and environmental protection indexes by utilizing a multi-objective optimization engine, and finally outputting an optimal mixing ratio scheme; the application output module is used for outputting the optimized mixing proportion scheme and the control instruction; The data acquisition and perception module specifically comprises: a11, a near infrared spectrometer is used for monitoring the water content of the aggregate in real time; a12, a laser particle size analyzer for detecting the fineness of the powdery raw materials; A13, a densimeter is used for detecting the concentration and the solid content of the water reducer; a14, a machine vision system, which is used for analyzing the grain size distribution and the grain shape of the aggregate in real time through image processing and a deep learning algorithm; a pH value sensor for monitoring the pH value of the water reducing agent solution; A16, a temperature sensor array for monitoring the temperature of the raw materials; A17, an online viscometer, which is used for monitoring rheological property of the mixture in real time; The core analysis module specifically comprises: a21, a machine learning-based performance prediction model is used for training the model by utilizing historical data to rapidly and accurately predict the strength and slump index of the concrete under a specified mixing ratio; A22, a multi-objective optimization engine, which is used for carrying out weighing and searching under the conditions of multiple objectives and constraint and automatically generating an optimal matching ratio scheme set; a23, a digital twin simulation unit, which is used for simulating the hydration process and microstructure evolution of the concrete based on a physical mechanism model and predicting the long-term durability and service performance of the concrete; The data storage and processing module comprises: S11, receiving an original data stream from a data acquisition and sensing module, wherein the original data stream comprises real-time raw material characteristic data from an on-line monitoring sensor, process time sequence data from production equipment, mix proportion and performance data from a laboratory and macroscopic supply chain data accessed through a data interface; s12, cleaning the data of the original data stream, including processing missing values, eliminating abnormal values generated by sensor faults or transmission interference, and carrying out unit unification and format standardization on data of different sources; S13, performing space-time alignment and fusion on the cleaned multi-source data, wherein the space-time alignment comprises marking all data with uniform time stamps, and correlating based on formula numbers and production batch numbers to construct a complete data record for analysis; s14, executing characteristic engineering operation on the fused data, wherein the operation comprises the following steps: A. calculating derived features, wherein the derived features comprise water-gel ratio, total cementing material and sand ratio; B. extracting statistical features from the process time sequence data, wherein the statistical features comprise an average value, a variance and an integral value of a specific time interval of stirring current; C. preprocessing macroscopic supply chain data to generate features reflecting future trend of raw material prices or inventory cost; s15, outputting the high-quality data set processed by the feature engineering to a core analysis module for training and reasoning based on a machine learning performance prediction model; The performance prediction model based on machine learning specifically comprises: S21, receiving a data set output by a data storage and processing module, wherein the data set comprises historical mix proportion data, corresponding real-time raw material characteristic data, process data characteristics and macroscopic characteristics; s22, dividing the data set into a training set and a testing set according to the time stamp, and training a preset machine learning algorithm by using the training set to fit a complex nonlinear mapping relation between the mixing proportion characteristic and the concrete performance index, wherein when the preset machine learning algorithm is trained by using the training set, the objective function is as follows: ; Wherein, the For the parameters of the optimal model to be chosen, As a parameter of the model, it is possible to provide, The concrete performance index actual value of the i-th sample, As a term of experience loss, Representing model M at given parameters Next, for the ith input feature vector Is used for the prediction result of (a), In order to regularize the term(s), As a function of the regularization, Is a regularization coefficient; S23, deploying the trained model in a production environment, receiving a new mixing ratio scheme and corresponding real-time sensor data, and generating a predicted value of a concrete key performance index; s24, outputting the predicted value to the multi-objective optimization engine as a core basis for scheme optimization and evaluation; the multi-objective optimization engine specifically comprises: s31, receiving performance predicted values of the machine learning-based performance predicted model for a set of candidate matching ratio schemes and long-term durability index predicted values from the digital twin simulation unit; S32, defining a plurality of optimization objective functions, wherein the defined optimization objective functions specifically comprise a cost objective function, a performance objective function and an environment-friendly objective function; s33, defining a constraint condition set, wherein the defined constraint condition set comprises performance constraints, durability constraints and formula constraints; s34, adopting a multi-objective optimization algorithm, and carrying out iterative search in a formula solution space based on the objective function and constraint conditions to generate a group of pareto optimal solution sets, wherein each solution in the pareto optimal solution sets represents an optimal balance scheme among a plurality of targets; And S35, fusing the cost and the performance by adopting a weighted fusion mode to obtain the priority index of each solution in the optimal solution set, and sequencing and outputting the pareto optimal solution set according to the calculated priority index.
2. The system for optimizing concrete mix based on industrial data analysis according to claim 1, wherein when defining a plurality of optimization objective functions, the defined optimization objective functions specifically include: a31 cost objective function: ; Wherein, the For the price of the j-th raw material, The unit dosage of the j-th raw material is M, and the number of the raw material types is M; A32, performance objective function: ; Wherein, the Taking a negative value for the predicted value of 28-day compressive strength of the scheme x as the performance prediction model indicates that the pursuit of the maximization of the strength is sought; a33, environmental protection objective function: ; Wherein, the Is the unit carbon emission factor of the j-th raw material.
3. The system for optimizing concrete mix based on industrial data analysis according to claim 2, wherein when the cost and the performance are fused by adopting a weighted fusion method, a priority index of each solution in the optimal solution set is obtained, a principle formula is as follows: ; Wherein, the For the predefined weighting coefficients of the cost objective function, The predefined weighting coefficients for the performance objective function, For the minimum of cost objective function values in all solutions in the pareto optimal solution set, For the maximum of the cost objective function values in all solutions in the pareto optimal solution set, For the minimum of the performance objective function values in all solutions in the pareto optimal solution set, For the maximum value of the performance objective function values in all solutions in the pareto optimal solution set, To adopt the ith optimal mixing proportion scheme When the estimated total cost of raw materials required for producing a unit of concrete is reached, For the ith optimum mix scheme Performance values of (2).
4. The system for optimizing concrete mix based on industrial data analysis according to claim 3, wherein the digital twin simulation unit comprises: S41, receiving standardized mixing ratio data from the data storage and processing module, wherein the standardized mixing ratio data comprises a water-gel ratio, chemical components and proportions of all cementing materials, aggregate characteristics and initial maintenance conditions; S42, calling and initializing a microscopic model based on a physical mechanism, and setting initial boundary conditions and material parameters of the model according to input mix ratio data; S43, iteratively executing simulation calculation of the microscopic model within a set time step, and simulating hydration reaction dynamics of cement particles, an evolution process of a microscopic pore structure and transmission diffusion of moisture and ions in a pore network; S44, calculating and outputting a predicted value of the long-term performance and the durability index based on the simulation result; And S45, conveying the long-term performance and the predicted value of the durability index to the multi-objective optimization engine as a durability constraint condition or an optimization objective when the multi-objective optimization is performed.
5. The method for optimizing the mix proportion of the concrete based on the industrial data analysis is applied to the system as claimed in any one of claims 1 to 4, and is characterized by comprising the following steps: s51, collecting and sensing data, namely collecting raw material characteristics, process time sequence data of a production process, concrete mix proportion and performance actual measurement data of a laboratory and supply chain information of raw material futures market price, stock level and logistics information in real time through a sensor and an interface; S52, preprocessing and fusing data, cleaning, standardizing, aligning time and space and engineering characteristics of multi-source data, and constructing a high-quality data set containing water-gel ratio, stirring current statistical characteristics and supply chain characteristics of inventory cost; s53, predicting performance, namely rapidly predicting the 28-day compressive strength and slump of the concrete under the specified mixing proportion based on the processed data by using a machine learning model and taking the water-cement ratio, the cementing material composition, the real-time aggregate water content, the real-time powder fineness and the environment temperature and humidity as input characteristics; s54, durability simulation, namely simulating the hydration process and microstructure evolution of the concrete by a digital twin technology, and predicting long-term durability indexes of the concrete; s55, multi-objective optimization, wherein under multiple constraints of performance, cost, environmental protection and the like, an optimization algorithm is adopted to automatically search and generate an optimal matching ratio scheme set; S56, scheme decision sorting, namely respectively normalizing the cost objective function value and the performance objective function value, multiplying the normalized cost objective function value and the normalized performance objective function value by a predefined cost weight coefficient and a predefined performance weight coefficient, summing the pre-defined cost weight coefficient and the pre-defined performance weight coefficient to obtain a priority index, and sorting the optimized scheme according to the priority index; and S57, outputting and applying, outputting the optimized proportioning scheme, the control instruction and the purchasing strategy, and guiding production and operation.

Description

System and method for optimizing concrete mix proportion based on industrial data analysis Technical Field The invention relates to the technical field of industrial data analysis, in particular to an optimization system and method for analyzing concrete mixing ratio based on industrial data. Background The design of the concrete mixing ratio is a core link for determining the performance, cost and environmental protection benefits of the concrete, and aims to determine the optimal combination proportion of raw materials such as water, cement, aggregate, admixture, additive and the like. The traditional method mainly depends on the experience of engineers, a large number of laboratory trial-and-error and table look-up according to specifications, and the whole process is tedious and time-consuming and more seriously depends on subjective judgment of people. With the increasing demands of the construction industry for concrete performance, cost control and green production, this experience-oriented traditional model has made it difficult to meet the demands of modern concrete intelligent manufacturing and fine management. The prior art is generally limited to single-objective optimization, e.g., pursuing only strength compliance or minimal cost, lacking comprehensive tradeoffs for multiple objectives of workability, long-term durability, and carbon emissions. The dependent historical data are mostly static laboratory records, and the real-time fluctuation information in the production process, such as the change of aggregate water content, cannot be effectively fused, so that the actual water-gel ratio is directly deviated from a design value, and the performance prediction distortion is caused. In addition, the existing method almost completely ignores the influence of raw material supply chain fluctuation on cost optimization, and lacks scientific prediction means on the microstructure and long-term performance of concrete, so that a mix proportion design scheme is often feasible in a short term but insufficient in long-term durability, or truly global optimum cannot be realized, and the application requirements of high-performance concrete and complex engineering are difficult to support. Aiming at the problems, the invention constructs a concrete mixing proportion optimizing system based on industrial data analysis. The method comprises the steps of capturing raw material characteristics and production process data in real time through deployment of an online sensor, fusing supply chain information to construct a high-quality data set, accurately predicting concrete performance by using a machine learning model, simulating long-term durability by combining a digital twin technology, finally automatically generating an optimal proportioning scheme by adopting a multi-objective optimization algorithm under multiple constraints of performance, cost, environmental protection and the like, and outputting recommended sequences, thereby realizing innovative breakthrough of spanning of the value of the proportioning from experience driving to data driving, single objective to global optimization and short-term performance to full life cycle. Disclosure of Invention In order to overcome the problems in the prior art, the invention provides an optimization system and a method for analyzing concrete mixing proportion based on industrial data. The technical scheme of the invention is that the concrete mix proportion optimizing system based on industrial data analysis comprises: The data acquisition and perception module is used for acquiring multisource industrial data of concrete production; the data storage and processing module is used for cleaning, data fusion and characteristic engineering of the multi-source industrial data acquired by the data acquisition and sensing module; The core analysis module is used for predicting the performance of the concrete through a machine learning model, balancing cost, performance and environmental protection indexes by utilizing a multi-objective optimization engine, and finally outputting an optimal mixing ratio scheme; and the application output module is used for outputting the optimized mixing ratio scheme and the control instruction. Preferably, the data acquisition and sensing module specifically includes: a11, a near infrared spectrometer is used for monitoring the water content of the aggregate in real time; a12, a laser particle size analyzer for detecting the fineness of the powdery raw materials; A13, a densimeter is used for detecting the concentration and the solid content of the water reducer; a14, a machine vision system, which is used for analyzing the grain size distribution and the grain shape of the aggregate in real time through image processing and a deep learning algorithm; a pH value sensor for monitoring the pH value of the water reducing agent solution; A16, a temperature sensor array for monitoring the temperature of the raw materials; a17, an on-li