Intelligent optimization design method and system for hydraulic mass concrete mix proportion

Abstract

The invention discloses a hydraulic mass concrete mix proportion intelligent optimization design method and system, and the method comprises the following steps: S1, inputting actually needed concrete indexes into a pre-trained and optimized intelligent hybrid prediction model, the concrete indexes at least comprising indexes related to strength and durability; s2, outputting an optimal concrete mix proportion through the intelligent mixing prediction model; the intelligent hybrid prediction model is an index feature weight determined according to the feature difference of concrete indexes, and is a high-generalization concrete macroscopic performance intelligent hybrid prediction model which is pre-constructed by fusing a plurality of neural network learning algorithm models in a weighting mode. The intelligent design of the concrete mix proportion can be rapidly and efficiently completed, and the scientificity and feasibility of the hydraulic mass concrete mix proportion design are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent design of hydraulic concrete, and in particular to an intelligent optimization design method and system for mix ratio of hydraulic mass concrete. Background Art

[0002] Mix ratio is a key factor affecting concrete performance, and a reasonable and reliable mix ratio design is the fundamental guarantee for concrete quality and service life. Existing concrete mix ratio design mainly includes two forms: one is a semi-quantitative design method based on experience, which uses empirical formulas, combines engineering requirements, and obtains the best concrete mix ratio through a large number of experiments; the other is to use optimization algorithms to calculate the optimal mix ratio, combine existing concrete mix ratio data with engineering overviews, and calculate the best concrete mix ratio to improve the scientificity and rationality of concrete mix ratio design.

[0003] However, with the continuous increase in the scale, difficulty and quality of engineering construction, higher requirements are placed on the design of concrete mix proportions, especially for hydraulic structures, which require large concrete volumes, high concrete performance requirements and harsh service conditions, further increasing the complexity of concrete mix design. In addition, different hydraulic structures have different requirements for concrete performance and mix design; the same hydraulic structure has different requirements for concrete performance in different construction environments. At the same time, with the continuous development of my country's water conservancy and hydropower projects, engineering construction has also put forward higher requirements for the durability of concrete, which has led to the complexity of concrete mix design.

[0004] The existing mix design method based on expert experience cannot efficiently meet the diverse performance requirements of large-volume concrete for hydraulic structures, and it is difficult to ensure the accuracy of mix design, which greatly increases the workload of concrete mix design and affects the progress of project construction.

[0005] In addition, the existing mix design methods based on optimization algorithms are highly targeted, but have weak generalization capabilities. The accuracy rates vary greatly under different construction environments and working conditions, and they cannot be applied to the complex and diverse mix design of large-volume concrete for hydraulic structures.

[0006] Based on this, in view of the problems that the existing mix design mostly relies on empirical formulas, has a long cycle and a large amount of experimental workload, it is urgent to build an intelligent mix design method suitable for large-volume concrete of hydraulic structures, so as to accurately, efficiently and scientifically obtain the concrete mix ratio, greatly reduce the workload and improve the efficiency of concrete mix design. Summary of the invention

[0007] The main purpose of the present invention is to provide a hydraulic mass concrete mix intelligent optimization design method and system which has strong applicability and can improve the concrete mix design efficiency.

[0008] The technical solution adopted by the present invention is:

[0009] Provided is a method for intelligent optimization design of hydraulic mass concrete mix ratio, comprising the following steps:

[0010] S1. Inputting the concrete index actually required into the pre-trained and optimized intelligent hybrid prediction model, wherein the concrete index at least includes indexes related to strength and durability;

[0011] S2, output the optimal concrete mix ratio through the intelligent mixing prediction model;

[0012] Among them, the intelligent hybrid prediction model is a highly generalized intelligent hybrid prediction model of concrete macro-performance that is pre-constructed by integrating multiple neural network learning algorithm models in a weighted manner according to the characteristic weights of the indicator determined according to the characteristic differences of the concrete indicators; the input of the intelligent hybrid prediction model is the concrete indicators related to strength and durability, and the output is the corresponding concrete mix ratio; based on the intelligent hybrid prediction model that has completed training and testing, with concrete preparation materials and mix ratio as independent variables and concrete indicators as dependent variables, the corresponding correlation function is generated, and the unilateral concrete production cost function is combined to form a concrete mix ratio optimization design objective function set, and according to the actual engineering conditions and relevant standards and specifications and the relationship between the range of values ​​of concrete preparation materials and mix ratio, a concrete mix ratio optimization design constraint function set is established, and the intelligent hybrid prediction model is iteratively optimized by combining the objective function set and the constraint function set to obtain the final optimized intelligent hybrid prediction model.

[0013] According to the above technical solution, the indicators related to strength and durability include concrete compressive strength, relative dynamic elastic modulus and average water seepage height.

[0014] Following the above technical solution, the algorithm models of multiple neural network learning include support vector machine algorithm SVM, random forest algorithm RF, and multi-layer perceptron algorithm MLP.

[0015] Following the above technical solution, the data set used for training and testing of the intelligent hybrid prediction model is: according to various factors in the concrete index, the original sample data set is constructed using the existing hydraulic mass concrete mix ratio data and macro-performance quantitative data, and the K-fold cross-validation method is used to divide the original sample data set into K sub-samples, one of the sub-samples is selected as the test set, and the remaining K-1 sub-samples are used as the training set.

[0016] Following the above technical solution, after pre-intelligent hybrid prediction model, the tuna school algorithm TSO is used and combined with the data features of the algorithm model learned by multiple neural networks to optimize the model parameters.

[0017] Following the above technical solution, the NSGA-III algorithm is specifically used to iteratively optimize the intelligent hybrid prediction model.

[0018] Following the above technical solution, the process of determining concrete indicators is: selecting indicators that characterize the strength and durability of large-volume hydraulic concrete, determining concrete materials and mix ratio factors, and establishing a concrete mix ratio design indicator system to determine concrete indicators.

[0019] The present invention also provides a hydraulic mass concrete mix ratio intelligent optimization design system, comprising:

[0020] The prediction model construction module is used to construct an intelligent hybrid prediction model, which is a highly generalized intelligent hybrid prediction model of concrete macro-performance pre-constructed by integrating multiple neural network learning algorithm models in a weighted manner according to the characteristic weights of the indicators determined by the characteristic differences of the concrete indicators. The input of the intelligent hybrid prediction model is the concrete indicators related to strength and durability, and the output is the corresponding concrete mix ratio. Based on the intelligent hybrid prediction model that has completed training and testing, with concrete preparation materials and mix ratio as independent variables and concrete indicators as dependent variables, the corresponding correlation relationship function is generated, and the unilateral concrete production cost function is combined to form a concrete mix ratio optimization design objective function set, and according to the actual engineering conditions and relevant standards and specifications and the relationship between the range of values ​​of concrete preparation materials and mix ratio, a concrete mix ratio optimization design constraint function set is established, and the intelligent hybrid prediction model is iteratively optimized by combining the objective function set and the constraint function set to obtain the final optimized intelligent hybrid prediction model.

[0021] An input module, used for inputting actually required concrete indicators into the pre-trained and optimized intelligent hybrid prediction model, wherein the concrete indicators at least include indicators related to strength and durability;

[0022] Output module, used to output the optimal concrete mix ratio through the intelligent mix prediction model.

[0023] According to the above technical solution, the indicators related to strength and durability include concrete compressive strength, relative dynamic elastic modulus and average water seepage height.

[0024] The present invention also provides a computer storage medium, which stores a computer program that can be executed by a processor, and the computer program executes the intelligent optimization design method for hydraulic mass concrete mix ratio described in the above technical solution.

[0025] The beneficial effects of the present invention are as follows: on the premise of ensuring the strength, durability and economy of concrete, the present invention pre-constructs an intelligent hybrid prediction model, and based on the correlation function between concrete preparation materials and mix ratio concrete indicators, combines the unilateral concrete production cost function to form a concrete mix ratio optimization design objective function set, and at the same time constructs a concrete mix ratio optimization design constraint function set, iteratively optimizes the intelligent hybrid prediction model, and obtains the final optimized intelligent hybrid prediction model; when the actually required concrete indicators are input into the intelligent hybrid prediction model, the optimal concrete mix ratio can be directly output, thereby quickly and efficiently completing the intelligent design of the concrete mix ratio, and improving the scientificity and feasibility of the large-volume concrete mix ratio design of hydraulic engineering.

[0026] Furthermore, the indicators related to strength and durability include concrete compressive strength, relative dynamic elastic modulus and average water seepage height. Compressive strength is an important indicator for characterizing the mechanical properties of concrete, and relative dynamic elastic modulus and average water seepage height are important indicators for characterizing the durability of concrete.

[0027] Furthermore, the algorithm models of multiple neural network learning include support vector machine algorithm SVM, random forest algorithm RF, and multi-layer perceptron algorithm MLP. The SVM algorithm has strong adaptability and generalization, high robustness, and can effectively deal with the noise in complex multi-source concrete mix data; the RF algorithm has strong feature selection ability, can evaluate the importance of each concrete mix factor, and perform effective feature selection; the MLP algorithm can handle nonlinear problems and high-dimensional data, has strong expression ability, and can efficiently obtain the complex correlation between concrete mix data.

[0028] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0030] Figure 1 It is a flow chart of a method for intelligent optimization design of hydraulic mass concrete mix ratio according to an embodiment of the present invention;

[0031] Figure 2 It is a schematic diagram of the process of building, training, testing and optimizing an intelligent hybrid prediction model according to an embodiment of the present invention;

[0032] Figure 3is a schematic diagram of compressive strength prediction performance according to an embodiment of the present invention;

[0033] Figure 4 It is a schematic diagram of the prediction performance of relative dynamic elastic modulus according to an embodiment of the present invention;

[0034] Figure 5 This is a schematic diagram of the average water seepage height prediction performance of an embodiment of the present invention;

[0035] Figure 6 1 is a schematic diagram for comparing the performance of models according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0037] It should be noted that the illustrations provided in the embodiments of the present invention are only used to illustrate the basic concept of the present invention in a schematic manner. Therefore, the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout type may also be more complicated.

[0038] In the present invention, it is also necessary to explain that, if the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. appear, the orientation or position relationship indicated is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present application. In addition, if the terms "first" and "second" appear, they are only used for description and distinction purposes, and cannot be understood as indicating or implying relative importance.

[0039] The intelligent optimization design method for the mix ratio of large-volume concrete of hydraulic structures proposed in the present invention can solve the problem of difficult selection of the mix ratio of large-volume concrete of hydraulic structures under complex and diverse construction environments, and efficiently and accurately realizes the intelligent and refined design of the concrete mix ratio; on the premise of ensuring the strength, durability and economy of the concrete, the intelligent design of the concrete mix ratio can be completed quickly and efficiently, thereby improving the effectiveness and reliability of the mix ratio design of large-volume concrete of hydraulic structures.

[0040] like Figure 1 As shown, the intelligent optimization design method for hydraulic mass concrete mix ratio according to the embodiment of the present invention comprises the following steps:

[0041] S1. Inputting the concrete index actually required into the pre-trained and optimized intelligent hybrid prediction model, wherein the concrete index at least includes indexes related to strength and durability;

[0042] S2. Output the optimal concrete mix ratio through the intelligent mixing prediction model.

[0043] Among them, the intelligent hybrid prediction model is a highly generalized intelligent hybrid prediction model of concrete macro-performance that is pre-constructed by integrating multiple neural network learning algorithm models and determining the indicator feature weights according to the characteristic differences of concrete indicators; the input of the intelligent hybrid prediction model is concrete indicators related to strength and durability, and the output is the corresponding concrete mix ratio; based on the intelligent hybrid prediction model that has completed training and testing, with concrete preparation materials and mix ratio as independent variables and concrete indicators as dependent variables, the corresponding correlation function is generated, and the unilateral concrete production cost function is combined to form a concrete mix ratio optimization design objective function set, and according to the actual engineering conditions and relevant standards and specifications, as well as the relationship between the range of values ​​of concrete preparation materials and mix ratio, a concrete mix ratio optimization design constraint function set is established; combined with the objective function set and the constraint function set, the intelligent hybrid prediction model is iteratively optimized to obtain the final optimized intelligent hybrid prediction model.

[0044] Among them, the indicators related to strength and durability include concrete compressive strength, relative dynamic elastic modulus and average water seepage height.

[0045] In a preferred embodiment of the present invention, the compatibility of the characteristics of each indicator with the algorithm is analyzed based on the concrete strength and durability indicators, and the support vector machine algorithm (SVM), random forest algorithm (RF), and multi-layer perceptron algorithm (MLP) are selected as the basic algorithms. The data features calculated by the three basic algorithms are fused in a weighted manner to construct an intelligent hybrid prediction model for concrete macroscopic performance, improve the generalization and robustness of the model, and realize the effective combination of different learning features, as shown in the following formula:

[0046] f(x)=w1SVM(x,s,r)+w2RF(x,t,u)+w3MLP(x,g,e) (1)

[0047] Where: f(x) is the intelligent hybrid prediction algorithm for concrete macro-performance; w1, w2, w3 are the weights of data features extracted by each algorithm; SVM(x), RF(x), MLP(x) are the data features extracted by each model; x is the input variable; s, r, t, u, g, e are algorithm parameters.

[0048] The data set used for training and testing of the intelligent hybrid prediction model is as follows: According to various factors in the concrete indicators, the original sample data set is constructed using the existing hydraulic mass concrete mix ratio data and macro-performance quantitative data. The K-fold cross-validation method is used to divide the original sample data set into K sub-samples, one of which is selected as the test set, and the remaining K-1 sub-samples are used as the training set.

[0049] After pre-intelligent hybrid prediction model, the tuna school algorithm TSO is used and combined with the data characteristics of the algorithm model learned by multiple neural networks to optimize the model parameters.

[0050] Furthermore, the NSGA-III algorithm is specifically used to iteratively optimize the intelligent hybrid prediction model.

[0051] Preferably, the process of determining the concrete index is: selecting indexes characterizing the strength and durability of hydraulic mass concrete, determining concrete materials and mix ratio factors, and establishing a concrete mix ratio design index system to determine the concrete index.

[0052] like Figure 2 As shown, this embodiment provides a preferred intelligent hybrid prediction model construction, training, testing and optimization process, which is as follows:

[0053] Step 1: Combined with the design data of large-volume concrete mix ratio of hydraulic engineering, with strength guarantee and durability as the priority, select indicators that characterize the strength and durability of concrete, determine the concrete materials and mix ratio factors, and establish a concrete mix ratio design indicator system.

[0054] Step 2: According to the concrete mix design index system, collect sample data of each variable in the concrete mix design index system, establish the original sample data set, and divide it into a training data set and a test data set.

[0055] Step 3: Considering the characteristic differences of concrete strength and durability, the support vector machine algorithm (SVM), random forest algorithm (RF) and multi-layer perceptron algorithm (MLP) are integrated to build a highly generalized intelligent hybrid prediction model for concrete macro-performance.

[0056] Step 4: According to the characteristics and fusion forms of each algorithm, extract the parameters to be optimized in the intelligent hybrid prediction model of concrete macro-performance, use the tuna school algorithm (TSO) and combine the data characteristics to optimize the model parameters to achieve automatic and high-precision prediction of concrete compressive strength, relative dynamic elastic modulus, and average water seepage height.

[0057] Step 5: Use the training data set to train the intelligent prediction model of concrete macro-performance, determine the optimal model parameters corresponding to different prediction targets, and learn to extract data features; use the test data set to evaluate the prediction effect of the model, characterize the correlation between concrete materials, mix factors and compressive strength, relative dynamic elastic modulus, and average water seepage height, and generate the corresponding correlation function.

[0058] Step 6: Based on the correlation function corresponding to compressive strength, relative dynamic elastic modulus, and average water seepage height, the concrete production cost is integrated to establish the concrete mix optimization design objective function. The NSGA-II optimization model is constructed with the concrete material dosage range and mix ratio relationship as constraints.

[0059] Step 7: Set the model optimization parameters and iteration termination conditions, and conduct multi-objective global optimization of the hydraulic mass concrete mix ratio to generate the Pareto optimal solution of the objective function and obtain the mix ratio optimization design results under different preference forms.

[0060] Specifically, step 1 includes the following steps:

[0061] Step 1-1, collect concrete material books, mix ratio test reports and other information for hydraulic dams, select compressive strength as the concrete strength evaluation index, and select relative dynamic elastic modulus and average water seepage height as concrete durability evaluation indexes.

[0062] Step 1-2, fully consider the characteristics of large-volume hydraulic concrete, extract concrete preparation materials and mix ratio factors such as water consumption, sand consumption, cement consumption, fly ash consumption, extra-large stone consumption, large stone consumption, medium stone consumption, small stone consumption, water reducer dosage, air entraining agent dosage, etc.

[0063] Step 1-3, based on compressive strength, relative dynamic elastic modulus, average water seepage height, combined with concrete preparation materials and mix ratio factors, establish a concrete mix ratio design index system.

[0064] Step 2 specifically includes the following steps:

[0065] Step 2-1, according to various factors in the concrete mix design index system, the original sample data set is constructed using the existing hydraulic mass concrete mix data and macro-performance quantitative data (the macro-performance of concrete mainly includes durability and mechanical properties, among which mechanical properties include compressive strength; durability performance includes relative dynamic elastic modulus and average water seepage height).

[0066] In step 2-2, the K-fold cross-validation method is used to divide the original sample data set into K sub-samples, one of which is selected as the test set, and the remaining K-1 sub-samples are used as the training set to form a training data set and a test data set.

[0067] Step 3 also includes the following steps:

[0068] In step 3-1, for concrete strength and durability indicators, analyze the adaptability of the characteristics of each indicator and the algorithm, and select support vector machine algorithm (SVM), random forest algorithm (RF), and multi-layer perceptron algorithm (MLP) as the basic algorithms.

[0069] Step 3-2, the data features obtained by the three basic algorithms are integrated in a weighted manner to construct an intelligent hybrid prediction model for concrete macroscopic performance, improve the generalization and robustness of the model, and achieve an effective combination of different learning features, as shown in the following formula:

[0070] f(x)=w1SVM(x,s,r)+w2RF(x,t,u)+w3MLP(x,g,e) (1)

[0071] Where: f(x) is the intelligent hybrid prediction algorithm for concrete macro-performance; w1, w2, w3 are the weights of data features extracted by each algorithm; SVM(x), RF(x), MLP(x) are the data features extracted by each model; x is the input variable; s, r, t, u, g, e are algorithm parameters.

[0072] Step 4 also includes the following steps:

[0073] Step 4-1, combined with the theoretical characteristics of each algorithm, analyze the calculation process of the intelligent hybrid prediction model of concrete macro-performance. According to formula (1), extract the weights of each algorithm in the model, the penalty factor s and kernel function coefficient r in the SVM algorithm, the number of decision trees t and the maximum depth of the decision tree u in the RF algorithm, and the maximum number of iterations g and the number of hidden layer nodes e in the MLP algorithm as the parameters to be optimized.

[0074] Step 4-2, use the tuna school algorithm (TSO) to optimize the intelligent hybrid prediction model of concrete macro performance. Initialize the hybrid prediction model parameters and determine the parameter value range; then, select the spiral search method to update the hybrid prediction model parameters, as shown in the following formula:

[0075]

[0076] Where: is the value of parameter i at time t+1; is the optimal value of the parameter at time t; is the actual value of parameter i at time t; N is the number of parameters. According to step 4-1, N is equal to 6; c1, c2, and β are the information interaction coefficients between parameters, i.e., the spiral search parameters of tuna schools.

[0077] Step 4-3, define the spiral search related parameters, use formula (2) to continuously update the values ​​of parameters such as penalty factor and kernel function coefficient, and use the change in parameter value between time t and time t+1 to calculate the fitness of parameter optimization. When the fitness value is constantly changing, it indicates that it is in the optimization process; when the fitness value tends to be stable, it indicates that the optimal parameter value has been obtained, that is, the tuna school has reached the optimal position.

[0078] Step 4-4, after multiple iterative calculations, the most stable fitness value and the best position information are obtained, and the final parameters of the hybrid prediction model are determined according to the best position information to achieve hybrid prediction model parameter optimization and improve calculation accuracy.

[0079] Step 5 also includes the following steps:

[0080] Step 5-1, using the training data set described in claim 3 to train the intelligent hybrid prediction model of concrete macroscopic properties, according to step 4-2, optimize the parameters in the model, obtain the optimal parameter values, extract the relationship between data features, and realize the intelligent prediction of compressive strength, relative dynamic elastic modulus, and average water seepage height.

[0081] Step 5-2, to verify the accuracy of the model, combined with the trained and optimized model parameters, with the test set data as input, and the compressive strength, relative dynamic elastic modulus, and average water seepage height as output, the accuracy of the model was evaluated using goodness of fit and root mean square error.

[0082] Step 5-3, based on the trained and tested intelligent hybrid prediction model of concrete macro-performance, with concrete preparation materials and mix ratio factors as independent variables, and compressive strength, relative dynamic elastic modulus, and average water seepage height as dependent variables, generate the corresponding correlation function to characterize the relationship between the independent variable and the dependent variable, as shown in the following formula:

[0083]

[0084] Where: CCS, RDM, WSE are the compressive strength, relative dynamic elastic modulus, and average water seepage height after standard curing for 28 days, respectively. CCS is the correlation function between compressive strength and concrete preparation materials and mix ratio factors; f RDM is the correlation function between relative dynamic elastic modulus and concrete preparation materials and mix ratio factors; f WSE is the correlation function between the average water seepage height and the concrete preparation materials and mix ratio factors;

[0085] x1, x2, x3, x4, x5, x6, x7, x8, x9, x 10 They are water consumption, sand consumption, cement consumption, fly ash consumption, extra large stone consumption, large stone consumption, medium stone consumption, small stone consumption, water reducing agent consumption, and air entraining agent consumption.

[0086] Step 6 also includes the following steps:

[0087] Step 6-1, consider the production cost of one cubic meter of concrete, collect the market unit price of concrete preparation materials, and establish the production cost function of one cubic meter of concrete, as follows:

[0088]

[0089] Where: PCC is the production cost of one cubic meter of concrete; i is the unit price of the i-th concrete preparation material and mix ratio factor; x i is the amount of materials and mix factors for preparing the i-th concrete; p0 is the mixing and transportation cost of one cubic meter of concrete.

[0090] Step 6-2, integrate the correlation function corresponding to the compressive strength, relative dynamic elastic modulus, and average water seepage height mentioned in step 5-3 with the unit concrete production cost function to establish the concrete mix ratio optimization design objective function set. As shown in the following formula:

[0091]

[0092] Where: X is the set of concrete preparation materials and mix ratio factors, as shown below:

[0093] X={x i |i=1,2,3,...10} (6)

[0094] Step 6-3, according to the actual engineering conditions and relevant standards and specifications, obtain the relationship between the range of concrete preparation material usage and mix ratio, and establish the concrete mix ratio optimization design constraint function set, as shown in the following formula:

[0095]

[0096] Where: X l 、CCS l , RDM l 、WSE l They are concrete preparation materials and mix ratio factors, compressive strength, relative dynamic elastic modulus, and the lower limit of average water seepage height; X u 、CCS u , RDM u 、WSE u 、PCC u It is the upper limit allowed for concrete preparation materials and mix ratio factors, compressive strength, relative dynamic elastic modulus, and average water seepage height.

[0097] Step 6-4, combining the concrete mix optimization design objective function set and constraint condition set, using the NSGA-III algorithm, establishes a concrete mix optimization model and searches for the optimal concrete mix.

[0098] Step 6-4 specifically includes the following steps:

[0099] Step 6-4-1, initialize the population. According to the given concrete mix ratio factors and the range of values ​​of each indicator in the constraint conditions, randomly generate a set of initial solutions, and use the values ​​of the concrete mix ratio factors to represent each individual in the population.

[0100] Step 6-4-2, non-dominated sorting. Sort the individuals in the population according to the non-dominated sorting method, obtain the non-dominated solutions corresponding to the individuals, sort all non-dominated solutions, calculate the crowding degree of the non-dominated solutions in the objective function space, and estimate the density occupied by each individual to facilitate individual pairing during crossover.

[0101] Step 6-4-3, crossover and mutation. Use a random method to select two groups of individuals for crossover to obtain a new generation of population, that is, concrete mix parameter values, and perform mutation operations on each individual to increase population diversity.

[0102] Step 6-4-4, update the population. Merge the newly generated population with the original population, retain the non-dominated solutions in the current round, bring them into the objective function, and use the crowding degree to select the most suitable solution.

[0103] Step 6-4-5, iteratively output the results. Loop through all the above steps to check whether the termination condition is met. If not, continue to the next generation of evolution; if the termination condition is met, output the high-quality solution set on the Pareto frontier.

[0104] Step 7 also includes the following steps:

[0105] Step 7-1, determine the parameters in the NSGA-III algorithm, and define the number of iterations and termination conditions of the NSGA-III algorithm.

[0106] Step 7-2, perform concrete mix optimization calculation to obtain the Pareto optimal solution of the objective function and provide decision support for concrete mix selection under different engineering requirements.

[0107] In another embodiment of the present invention, a concrete dam construction is taken as an example to verify the effectiveness and advancement of the intelligent optimization design method for the mix ratio of hydraulic mass concrete. The concrete design requirements of the project are as follows: 28d compressive strength is between 30 and 40MPa, relative dynamic elastic modulus is between 80% and 100%, average water seepage height is between 0mm and 45mm, and unit production cost is between 650 yuan and 900 yuan. The amount limit of concrete preparation materials and the allowable range of relevant mix ratio parameters are shown in Table 1.

[0108] Table 1 Permissible range of relevant parameters of concrete mix ratio

[0109] Concrete components Unit price (yuan / t) Dosage lower limit (kg) Dosage lower limit (kg) water 10 70 120 sand 200 450 600 cement 800 100 200 Fly ash 50 0 150 Extra large stone 70 450 700 Dashi 60 450 700 Zhongshi 80 250 500 Xiaoshi 90 76 284 Water reducing agent 3000 0 5 Air Entraining Agent 2700 0 15

[0110] The concrete mix design report was summarized and the concrete mix design and macro performance test data were obtained to form 820 sets of original sample data sets. According to step 3, the training set and the test set were divided into 615 sets of data and the test set contained 205 sets of data. According to steps 3 to 7, the concrete mix optimization calculation is mainly divided into two stages: intelligent prediction of concrete compressive strength, relative dynamic elastic modulus, and average water seepage height, and concrete mix optimization.

[0111] 1. Intelligent prediction of concrete compressive strength, relative dynamic elastic modulus, and average water seepage height.

[0112] The TSO algorithm parameters are determined as follows: number of iterations = 200, c1 = 1, c2 = 2, β = 0.9; the initial values ​​of the parameters to be optimized are set as follows: w1 = 0.3, w2 = 0.3, w3 = 04, s = 0, r = 0, t = 0, u = 0, g = 0, e = 0. The correlation coefficient (R 2 ), root mean squared error (RMSE) and mean absolute error (MAE) are used to quantitatively evaluate the model performance. 2 The closer the value is to 1, the smaller the RMSE and MAE values ​​are, indicating that the model performance is better. Using the training set in the original sample data set, after 100 iterations of calculation, the model parameters are optimized, and the model performance is verified by the test set to obtain the model performance evaluation results (see Figure 3-Figure 5 ).

[0113] In order to further verify the reliability and superiority of the model, based on the original sample data set, the existing concrete performance prediction models were compared, including: multi-gene genetic programming (M-GGP) model, artificial neural network-support vector regression (ANN-SVR) integrated model, RF model, SVM model, MLP model, and the prediction effects of different models were obtained (see Figure 6 ). Figure 6 The R of the mixed model prediction is 2 It is 0.96. Compared with other models, the model accuracy is improved by 0.08, 0.13, 0.18, 0.21, and 0.25, respectively, which proves the superiority of the model in predicting concrete properties.

[0114] 2. Optimization of concrete mix ratio.

[0115] Combined with the concrete design requirements, the data in Table 1 are substituted into formula (6) to form a set of constraints. According to the process described in step 6, the parameters in the NSGA-III algorithm are defined: evolutionary generation = 100, crossover probability = 1, and mutation probability = 0.3. The NSGA-III algorithm is used to solve the mathematical model, and the Pareto frontier scatter distribution after multiple iterations of optimization is obtained (see Table 2), and the optimal concrete mix ratio set is obtained (see Table 3).

[0116] Table 2 Pareto frontier data

[0117] index 1 2 3 … 34 35 36 Compressive strength(MPa) 38.08 34.33 31.36 … 30.15 34.11 39.02 Relative dynamic elastic modulus (%) 97.11 90.67 94.34 … 97.17 96.56 91.31 Average water seepage height (mm) 40 33 18 … 15 39 29 Cost of concrete per cubic meter (yuan) 837.37 899.64 823.03 … 777.71 835.02 682.37

[0118] Table 3 Concrete mix ratio

[0119]

[0120] The present invention also provides a hydraulic mass concrete mix ratio intelligent optimization design system, comprising:

[0121] The prediction model construction module is used to construct an intelligent hybrid prediction model, which is a highly generalized intelligent hybrid prediction model of concrete macro-performance pre-constructed by integrating multiple neural network learning algorithm models in a weighted manner according to the characteristic weights of the indicators determined by the characteristic differences of the concrete indicators. The input of the intelligent hybrid prediction model is the concrete indicators related to strength and durability, and the output is the corresponding concrete mix ratio. Based on the intelligent hybrid prediction model that has completed training and testing, the concrete preparation materials and mix ratio are used as independent variables, and the concrete indicators are used as dependent variables to generate corresponding correlation relationship functions, and the unilateral concrete production cost function is combined to form a concrete mix ratio optimization design objective function set, and according to the actual engineering conditions and relevant standards and specifications, and the relationship between the range of values ​​of concrete preparation materials and mix ratio, a concrete mix ratio optimization design constraint function set is established. Combined with the objective function set and the constraint function set, the intelligent hybrid prediction model is iteratively optimized to obtain the final optimized intelligent hybrid prediction model.

[0122] An input module, used for inputting actually required concrete indicators into the pre-trained and optimized intelligent hybrid prediction model, wherein the concrete indicators at least include indicators related to strength and durability;

[0123] Output module, used to output the optimal concrete mix ratio through the intelligent mix prediction model.

[0124] Among them, the indicators related to strength and durability include concrete compressive strength, relative dynamic elastic modulus and average water seepage height.

[0125] The various modules of the system in this embodiment are mainly used to implement the various steps of the above method embodiment, which will not be described one by one here.

[0126] The present application also provides a computer-readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a disk, an optical disk, a server, an App application store, etc., on which a computer program is stored, and when the program is executed by a processor, a corresponding function is realized. When the computer-readable storage medium of this embodiment is executed by a processor, the intelligent optimization design method for the mix ratio of hydraulic mass concrete of the method embodiment is realized.

[0127] It should be pointed out that, according to the needs of implementation, the various steps / components described in this application can be split into more steps / components, and two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0128] The order of execution of each step in the above embodiment does not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application.

[0129] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should fall within the scope of protection of the appended claims of the present invention.

Claims

1. An intelligent optimization design method for the mix ratio of hydraulic mass concrete. It is characterized in that The following steps are involved: S1. Inputting the concrete index actually required into the pre-trained and optimized intelligent hybrid prediction model, wherein the concrete index at least includes indexes related to strength and durability; S2, output the optimal concrete mix ratio through the intelligent mixing prediction model; Among them, the intelligent hybrid prediction model is a highly generalized intelligent hybrid prediction model of concrete macro-performance that is pre-constructed by using a weighted method to fuse multiple neural network learning algorithm models based on the characteristic weights of the indicator determined according to the characteristic differences of the concrete indicators; the input of the intelligent hybrid prediction model is the concrete indicators related to strength and durability, and the output is the corresponding concrete mix ratio; Based on the intelligent hybrid prediction model that has completed training and testing, with concrete preparation materials and mix ratio as independent variables and concrete indicators as dependent variables, the corresponding correlation function is generated, and the unilateral concrete production cost function is combined to form a concrete mix ratio optimization design objective function set. According to the actual engineering conditions, relevant standards and specifications, and the relationship between the range of values ​​of concrete preparation materials and mix ratio, a concrete mix ratio optimization design constraint function set is established. Combining the objective function set with the constraint function set, the intelligent hybrid prediction model is iteratively optimized to obtain the final optimized intelligent hybrid prediction model.

2. According to claim 1, the intelligent optimization design method for hydraulic mass concrete mix ratio, It is characterized in that Indicators related to strength and durability include concrete compressive strength, relative dynamic elastic modulus and average water seepage height.

3. According to claim 1, the intelligent optimization design method for hydraulic mass concrete mix ratio, It is characterized in that The algorithm models for multiple neural network learning include support vector machine algorithm SVM, random forest algorithm RF, and multi-layer perceptron algorithm MLP.

4. The intelligent optimization design method for hydraulic mass concrete mix ratio according to claim 1, It is characterized in that The data set used for training and testing of the intelligent hybrid prediction model is as follows: According to various factors in the concrete indicators, the original sample data set is constructed using the existing hydraulic mass concrete mix ratio data and macro-performance quantitative data. The K-fold cross-validation method is used to divide the original sample data set into K sub-samples, one of which is selected as the test set, and the remaining K-1 sub-samples are used as the training set.

5. The intelligent optimization design method for hydraulic mass concrete mix ratio according to claim 1, It is characterized in that After pre-intelligent hybrid prediction model, the tuna school algorithm TSO is used and combined with the data characteristics of the algorithm model learned by multiple neural networks to optimize the model parameters.

6. The intelligent optimization design method for hydraulic mass concrete mix ratio according to claim 1, It is characterized in that Specifically, the NSGA-III algorithm is used to iteratively optimize the intelligent hybrid prediction model.

7. The intelligent optimization design method for hydraulic mass concrete mix ratio according to claim 1, It is characterized in that The process of determining concrete indicators is: selecting indicators that characterize the strength and durability of large-volume hydraulic concrete, determining concrete materials and mix ratio factors, and establishing a concrete mix ratio design indicator system to determine concrete indicators.

8. An intelligent optimization design system for hydraulic mass concrete mix ratio, It is characterized in that include: The prediction model building module is used to build an intelligent hybrid prediction model, which is based on the characteristic weights of the indicators determined by the characteristic differences of the concrete indicators, and adopts a weighted method to fuse multiple neural network learning algorithm models to build a highly generalized intelligent hybrid prediction model for concrete macro-performance. The input of the intelligent hybrid prediction model is concrete indicators related to strength and durability, and the output is the corresponding concrete mix ratio. Based on the intelligent hybrid prediction model that has completed training and testing, with concrete preparation materials and mix ratio as independent variables and concrete indicators as dependent variables, the corresponding correlation function is generated, and the unilateral concrete production cost function is combined to form a concrete mix ratio optimization design objective function set. According to the actual engineering conditions and relevant standards and specifications, as well as the relationship between the range of concrete preparation material usage and mix ratio, a concrete mix ratio optimization design constraint function set is established. Combined with the objective function set and the constraint function set, the intelligent hybrid prediction model is iteratively optimized to obtain the final optimized intelligent hybrid prediction model. An input module, used for inputting actually required concrete indicators into the pre-trained and optimized intelligent hybrid prediction model, wherein the concrete indicators at least include indicators related to strength and durability; Output module, used to output the optimal concrete mix ratio through the intelligent mix prediction model.

9. The intelligent optimization design system for hydraulic mass concrete mix ratio according to claim 8, It is characterized in that Indicators related to strength and durability include concrete compressive strength, relative dynamic elastic modulus and average water seepage height.

10. A computer storage medium, It is characterized in that A computer program executable by a processor is stored therein, and the computer program executes the intelligent optimization design method for hydraulic mass concrete mix ratio according to any one of claims 1 to 7.

Images