Intelligent generation method, device and equipment for multi-condition design formula of civil engineering

Through spectral clustering and decision tree division design conditions, combined with intelligent models of physical constraints, a multi-condition design formula is generated in civil engineering, which solves the complex and time-consuming problem of design formula generation in traditional methods, and realizes efficient and accurate design formula generation.

CN118052035BActive Publication Date: 2024-11-26TSINGHUA UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311722000.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-12-14
Publication Date
2024-11-26
Estimated Expiration
2043-12-14

AI Technical Summary

Technical Problem

The generation of civil engineering design formulas is complex and time-consuming, especially difficult to obtain accurately under multiple conditions. Traditional methods rely on theoretical derivation and experimental calibration, resulting in low accuracy and a large amount of time.

Method used

The design conditions are divided by spectral clustering and decision tree methods, and sub-data sets are generated, and the formula that considers physical constraints is used to generate intelligent models and automatically generate multi-condition design formulas in civil engineering. By combining sub-computing formulas, efficient and accurate design formulas are achieved. generate.

Benefits of technology

It realizes efficient and accurate generation of design formulas under multiple conditions of civil engineering, simplifies the design process, greatly saves manpower and time costs, and improves the accuracy and mechanical significance of design calculations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118052035B_ABST
    Figure CN118052035B_ABST
Patent Text Reader

Abstract

The present application relates to a method, device, computer equipment, storage medium and computer program product for intelligently generating multi-condition design formulas for civil engineering, wherein the method includes: obtaining civil engineering test data and design conditions; using spectral clustering and decision tree methods to divide the design conditions to obtain design condition classification results, classifying the civil engineering test data based on the design condition classification results to obtain multiple sub-data sets, using a formula considering physical constraints to generate an intelligent model for each sub-data set to obtain a sub-calculation formula, and then merging these sub-calculation formulas to generate a multi-condition design formula for civil engineering. In the whole process, the civil engineering test data and design conditions are automatically processed, corresponding sub-data sets are obtained for different design conditions, sub-calculation formulas are generated in a targeted manner, and an intelligent model is generated using a formula considering physical constraints to automatically generate formulas.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, device, computer equipment, storage medium and computer program product for intelligently generating multi-conditional design formulas for civil engineering. Background Art

[0002] Civil engineering design formulas are the cornerstone of civil engineering. They help engineers perform accurate calculations and analyses during design to ensure the quality of engineering projects. They combine scientific principles, empirical data, and engineering practices to provide a solid foundation for the reliability and safety of infrastructure such as buildings and bridges. Nowadays, various new materials and new structures are beginning to be applied to civil engineering projects, but they are difficult to promote due to the lack of clear design specifications, and a large number of design formulas are the core part of the design specifications. The search for design formulas for civil engineering material components is a complex multi-mechanical-geometric variable regression problem, and the same material or component will have different design formulas under different design conditions, which makes the generation of formulas extremely difficult.

[0003] Traditional design formula discovery mostly adopts the method of "theoretical derivation + experimental calibration". Researchers first establish various possible mathematical models based on theoretical derivation, then observe the laws and calibrate the models from the experimental data, and finally obtain the design formula. However, it is difficult to establish accurate constitutive models for most civil engineering materials and components. When there are many variables, it is also difficult for researchers to observe explicit laws from high-dimensional data. This results in a lot of time and manpower spent on finding design formulas, and the accuracy of manually discovered design formulas is often not high.

[0004] Therefore, there is an urgent need for an efficient and accurate intelligent generation solution for multi-condition design formulas in civil engineering. Summary of the invention

[0005] Based on this, it is necessary to provide an efficient and accurate civil engineering multi-condition design formula intelligent generation method, device, computer equipment, computer readable storage medium and computer program product to address the above technical problems.

[0006] In a first aspect, the present application provides a method for intelligently generating multi-condition design formulas for civil engineering. The method comprises:

[0007] Obtain civil engineering test data and design conditions;

[0008] The design conditions are divided by using spectral clustering and decision tree methods to obtain a design condition classification result;

[0009] Classifying the civil engineering test data based on the design condition classification result to obtain multiple sub-data sets;

[0010] Inputting each of the sub-data sets into a formula generation intelligent model taking into account physical constraints to obtain a sub-calculation formula;

[0011] The sub-calculation formulas are combined to generate a civil engineering multi-condition design formula.

[0012] In one embodiment, the design conditions are divided by using spectral clustering and decision tree methods to obtain the design condition classification results, including:

[0013] Selecting a set of candidate reference quantities for dividing design conditions;

[0014] Calculating alternative reference quantities from the set of alternative reference quantities according to the civil engineering test data;

[0015] Performing spectral clustering on the candidate reference amount to obtain a spectral clustering classification result;

[0016] Using the candidate reference amount as input and the spectral clustering classification result as a category identifier, a decision tree model is constructed;

[0017] The design conditions are divided based on the constructed decision tree model to obtain a design condition classification result.

[0018] In one embodiment, the set of candidate reference quantities for selecting the partitioning design conditions includes:

[0019] Obtain the dimensional matrix of the mechanical performance prediction problem in civil engineering formula design and the target quantity of the prediction problem;

[0020] Based on the dimensional matrix and the target quantity, a set of candidate reference quantities for dividing design conditions is selected.

[0021] In one embodiment, the obtaining of the dimensional matrix of the mechanical property prediction problem in the civil engineering formula design and the target quantity of the prediction problem includes:

[0022] Obtain the problem of mechanical performance prediction in civil engineering formula design;

[0023] Identifying the characteristics of the mechanical property prediction problem and the target quantity of the prediction problem;

[0024] Construct a dimensional matrix of the features.

[0025] In one embodiment, before performing spectral clustering on the candidate reference amount to obtain the spectral clustering classification result, the method further includes:

[0026] The alternative reference amounts are subjected to maximum-minimum normalization.

[0027] In one embodiment, the step of inputting each of the sub-data sets into a formula that takes physical constraints into consideration to generate an intelligent model to obtain a sub-calculation formula includes:

[0028] Generate corresponding symbol sequences according to each of the sub-data sets respectively;

[0029] Inputting the symbol sequence into a formula generation intelligent model taking into account physical constraints to obtain an output result, and extracting the calculation results carried in each of the sub-data sets;

[0030] Iteratively updating the formula generation intelligent model considering physical constraints based on the output result and the calculation result to obtain an updated formula generation intelligent model considering physical constraints;

[0031] Each of the sub-data sets is input into an updated formula generation intelligent model taking physical constraints into consideration to obtain a sub-calculation formula.

[0032] In one embodiment, before inputting the symbol sequence into a formula considering physical constraints to generate an intelligent model to obtain an output result, the method further includes:

[0033] Obtaining grammatical rules of Polish expressions and local and global dimensional constraints, wherein the local and global dimensional constraints are used to restrict symbol selection so that the model generates dimensionally balanced formulas;

[0034] Based on the grammatical rules of the Polish expression and the local and global dimensional constraints, a long short-term memory model is used to generate a formula generation intelligent model that takes physical constraints into consideration.

[0035] In a second aspect, the present application also provides a device for intelligently generating multi-condition design formulas for civil engineering. The device comprises:

[0036] Data acquisition module, used to obtain civil engineering test data and design conditions;

[0037] A partitioning module, used to partition the design conditions using spectral clustering and decision tree methods to obtain a classification result of the design conditions;

[0038] A classification module, used for classifying the civil engineering test data based on the design condition classification result to obtain a plurality of sub-data sets;

[0039] A sub-formula generation module, used for inputting each of the sub-data sets into a formula generation intelligent model taking physical constraints into consideration to obtain a sub-calculation formula;

[0040] The merging module is used to merge the sub-calculation formulas to generate a civil engineering multi-condition design formula.

[0041] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0042] Obtain civil engineering test data and design conditions;

[0043] The design conditions are divided by using spectral clustering and decision tree methods to obtain a design condition classification result;

[0044] Classifying the civil engineering test data based on the design condition classification result to obtain multiple sub-data sets;

[0045] Inputting each of the sub-data sets into a formula generation intelligent model taking into account physical constraints to obtain a sub-calculation formula;

[0046] The sub-calculation formulas are combined to generate a civil engineering multi-condition design formula.

[0047] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0048] Obtain civil engineering test data and design conditions;

[0049] The design conditions are divided by using spectral clustering and decision tree methods to obtain a design condition classification result;

[0050] Classifying the civil engineering test data based on the design condition classification result to obtain multiple sub-data sets;

[0051] Inputting each of the sub-data sets into a formula generation intelligent model taking into account physical constraints to obtain a sub-calculation formula;

[0052] The sub-calculation formulas are combined to generate a civil engineering multi-condition design formula.

[0053] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0054] Obtain civil engineering test data and design conditions;

[0055] The design conditions are divided by using spectral clustering and decision tree methods to obtain a design condition classification result;

[0056] Classifying the civil engineering test data based on the design condition classification result to obtain multiple sub-data sets;

[0057] Inputting each of the sub-data sets into a formula generation intelligent model taking into account physical constraints to obtain a sub-calculation formula;

[0058] The sub-calculation formulas are combined to generate a civil engineering multi-condition design formula.

[0059] The above-mentioned civil engineering multi-condition design formula intelligent generation method, device, computer equipment, storage medium and computer program product obtain civil engineering test data and design conditions; use spectral clustering and decision tree methods to divide the design conditions to obtain design condition classification results, classify the civil engineering test data based on the design condition classification results to obtain multiple sub-data sets, use a formula considering physical constraints to generate an intelligent model for each sub-data set to obtain a sub-calculation formula, and then merge these sub-calculation formulas to generate a civil engineering multi-condition design formula. In the whole process, the civil engineering test data and design conditions are automatically processed, corresponding sub-data sets are obtained for different design conditions, sub-calculation formulas are generated in a targeted manner, and an intelligent model is used to generate formulas considering physical constraints to automatically generate formulas, which can achieve efficient and accurate intelligent generation of civil engineering multi-condition design formulas. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] FIG1 is an application environment diagram of a method for intelligently generating multi-conditional design formulas for civil engineering in one embodiment;

[0061] FIG2 is a schematic diagram of a flow chart of a method for intelligently generating multi-conditional design formulas for civil engineering in one embodiment;

[0062] FIG3 is a schematic diagram of a technical flow of a method for intelligently generating multi-conditional design formulas for civil engineering in one embodiment;

[0063] FIG4 is a schematic diagram of a sub-process of S200 in one embodiment;

[0064] FIG5 is a schematic diagram of the process and application of a design condition intelligent partitioning algorithm in one embodiment;

[0065] FIG6 is a schematic diagram of a sub-process of S400 in one embodiment;

[0066] FIG7 is a structural block diagram of a device for intelligently generating multi-conditional design formulas for civil engineering in one embodiment;

[0067] FIG. 8 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0068] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application 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 application and are not used to limit the present application.

[0069] The intelligent generation method of civil engineering multi-condition design formula provided by the embodiment of the present application can be applied to the application environment shown in Figure 1. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The user operates on the terminal 102 side, inputs data such as civil engineering test data and design conditions, and clicks the formula generation button. The terminal 102 responds to the user operation, generates a formula generation request, sends the formula generation request to the server 104, and the server 104 parses the formula generation request to obtain the civil engineering test data and design conditions; the design conditions are divided by spectral clustering and decision tree methods to obtain the design condition classification results, and the civil engineering test data is classified based on the design condition classification results to obtain multiple sub-data sets. For each sub-data set, a formula considering physical constraints is used to generate an intelligent model to obtain a sub-calculation formula, and then these sub-calculation formulas are merged to generate a civil engineering multi-condition design formula. Further, the server 104 can push the generated civil engineering multi-condition design formula to the terminal 102. The terminal 102 may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. The portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers.

[0070] In one embodiment, as shown in FIG. 2 and FIG. 3 , the present application provides a method for intelligently generating a multi-condition design formula for civil engineering, which is described by taking the method applied to the server 104 in FIG. 1 as an example, and includes the following steps:

[0071] S100: Obtain civil engineering test data and design conditions.

[0072] Civil engineering test data usually include various physical quantities, chemical quantities, mechanical quantities, etc., such as material strength, elastic modulus, Poisson's ratio, density, etc., as well as the response of the structure under different loads, such as displacement, strain, stress, etc. These data usually need to be obtained through experiments, such as concrete specimen compressive strength test, reinforced concrete simply supported beam test, etc. In the experiment, by controlling different variables, such as material type, material dosage, load size and mode of action, various responses are observed and recorded, and analyzed and processed to obtain useful information and conclusions. Civil engineering test data is of great significance to both engineering practice and scientific research. By analyzing the data, we can understand the performance of the material and the response of the structure, and provide a basis and support for design and construction. At the same time, test data can also be used to verify and improve existing theories and models, and promote the development of civil engineering disciplines. Design conditions refer to the conditions set corresponding to the civil engineering design plan, which can be analyzed from the entire civil engineering design plan.

[0073] S200: The design conditions are divided by using spectral clustering and decision tree methods to obtain the classification results of the design conditions.

[0074] Spectral clustering is a classic unsupervised clustering algorithm that uses the correlation between data to group. Unlike traditional K-means and hierarchical clustering methods, spectral clustering treats data samples as nodes in a graph, constructs a graph using the correlation between nodes, and then obtains clustering results through spectral decomposition of the graph. Spectral clustering focuses on the "continuity" between data points, that is, each cluster of data points in the classification result constitutes a continuous hypersurface, so spectral clustering can be used to divide the continuous branches of the data set. Therefore, using spectral clustering here can make the final design condition classification results continuous, and it is also convenient to merge different sub-calculation formulas later. Furthermore, spectral clustering is a black box model like neural network. The classification criteria cannot be expressed explicitly to write the design formula, so the classification results need to be further explained. The decision tree method of binary classification is introduced, and the pre-selected alternative reference quantity is used as the input data, and the classification result of spectral clustering is used as its category identifier to construct a decision tree model. After the model training is completed, a specific decision boundary is output. If the error between the decision boundary and the cluster boundary is within the allowable range, the data can be re-divided according to the decision boundary. After using spectral clustering and decision tree methods, the design conditions are reasonably divided to obtain the design condition classification results.

[0075] S300: Classify the civil engineering test data based on the design condition classification result to obtain multiple sub-data sets.

[0076] There are corresponding data for different design conditions. Here, the entire civil engineering test data is classified according to the classification results of the design conditions to obtain multiple sub-data sets. Further, the mechanical response of civil engineering material components is a complex nonlinear problem, so the corresponding design formula is likely not a continuous function. On the other hand, engineering experience tells us that a slight change in a certain property of a material or component will not cause a huge change in the mechanical response in most cases, that is, the "local" corresponding function f is almost continuous everywhere. According to the Littlewood principle, we can find several "continuous branches" of f, so that it is a continuous function on each branch set, and then we can find the elementary function expression of f on each branch, and finally get a piecewise function. If you want to get an explicit design formula, the piecewise condition of the function, that is, the distinguishing condition of the branch set, also needs to have an explicit expression. In engineering practice, these piecewise conditions correspond to different design conditions. Therefore, the design conditions are classified in S200, and then the civil engineering test data are classified here according to the classification results of the design conditions to obtain multiple sub-data sets.

[0077] S400: Input each sub-data set into a formula generation intelligent model taking physical constraints into consideration to obtain a sub-calculation formula.

[0078] Civil engineering test data is divided into multiple sub-datasets according to different design conditions, and formulas are searched on each sub-dataset. The traditional symbolic regression algorithm regards the formula as a binary tree, and each tree node represents a symbol of the expression. Inheriting this method and combining the Polish Notation to eliminate brackets, any formula can be converted into a sequence of category vectors, and each element in the sequence is a token in a mathematical dictionary. After the above conversion, any complex formula can be represented by a one-dimensional sequence. Based on the above conversion, the formula regression problem can be analogized to a sequence generation problem in the field of natural language processing. Based on this technical principle, a formula generation intelligent model can be generated. The model can determine the symbol sequence corresponding to the formula based on the sub-dataset, and then restore it to a self-calculating formula based on the symbol sequence. Specifically, the formula generation intelligent model considering physical constraints can be established based on the long short-term memory model. Through the self-learning function of the model, the probability of each token at a certain position can be given according to the context information (obtained based on the sub-dataset). On the basis of this probability distribution, the physical constraints are considered, and the token with the highest probability under the design conditions can be obtained, and the corresponding symbol sequence corresponding to the formula can be generated.

[0079] S500: Merge sub-calculation formulas to generate multi-condition design formulas for civil engineering.

[0080] For the generated sub-calculation formulas, a continuous civil engineering multi-condition design formula can be obtained by merging them.

[0081] The above-mentioned intelligent generation method of civil engineering multi-condition design formula obtains civil engineering test data and design conditions; uses spectral clustering and decision tree methods to divide the design conditions to obtain the design condition classification results, classifies the civil engineering test data based on the design condition classification results to obtain multiple sub-data sets, and uses a formula considering physical constraints to generate an intelligent model for each sub-data set to obtain a sub-calculation formula, and then merges these sub-calculation formulas to generate a civil engineering multi-condition design formula. In the whole process, the civil engineering test data and design conditions are automatically processed, and corresponding sub-data sets are obtained for different design conditions, sub-calculation formulas are generated in a targeted manner, and an intelligent model is used to generate formulas considering physical constraints to automatically generate formulas, which can realize efficient and accurate intelligent generation of civil engineering multi-condition design formulas.

[0082] As shown in FIG. 4 , in one embodiment, S200 includes:

[0083] S210: Selecting a set of candidate reference quantities for dividing design conditions.

[0084] In order to realize the automatic identification of design conditions, the alternative reference quantities are selected first. The alternative reference quantities are alternative physical quantities for dividing the design conditions. Furthermore, in order to obtain the set of alternative reference quantities, it is necessary to first obtain the dimensional matrix of the mechanical performance prediction problem in the civil engineering formula design and the target quantity of the prediction problem; then, based on the dimensional matrix and the target quantity, select the set of alternative reference quantities for dividing the design conditions. Specifically, in a mechanical performance prediction problem, there are n related static features. , the target quantity to be predicted is y. The dimension of any feature can be expressed by the power product of the dimensions of basic physical quantities. The basic physical quantities commonly used in civil engineering are length, mass and time, and the corresponding dimensions are L, T and M. Therefore, the feature x i The dimension of dim x i It can be expressed as the power product of L, T and M:

[0085]

[0086] Arrange the three exponents in the dimensional expression of each static feature in order, and you will get the dimensional matrix of all the features of the problem:

[0087]

[0088] Then there is a function So that:

[0089]

[0090] In order to obtain a practical design formula, we hope that f has an expression of elementary functions, but elementary functions are all continuous functions, and the mechanical response of civil engineering material components is a complex nonlinear problem, so f as a whole may not be a continuous function. Engineering experience tells us that a slight change in a certain property of a material or component will not cause a huge change in the mechanical response in most cases, that is, f is almost continuous everywhere. According to Littlewood's principle, we can find several "continuous branches" of f, so that it is a continuous function on each branch set, and then we can find the elementary function expression of f on each branch, and finally get a piecewise function. If you want to get an explicit design formula, the piecewise condition of the function, that is, the condition for distinguishing the branch set, also needs to have an explicit expression. In engineering practice, these piecewise conditions correspond to different design conditions.

[0091] In order to realize the automatic identification of design conditions, the alternative reference quantities are selected first. The alternative reference quantities are the alternative physical quantities for dividing the design conditions. It is calculated as follows:

[0092]

[0093] in is the dimensional matrix of the problem, ; r is the dimension imbalance limit, which is usually 0. For simple problems, it can be equal to or greater than 1; s is the complexity limit, which is usually 2 to 4; the equivalence relation “~” will In the equation, vectors with opposite numbers are considered equal. is a finite set, let There are l elements in it, recorded as:

[0094]

[0095] S220: Calculate alternative reference quantities from a set of alternative reference quantities based on civil engineering test data.

[0096] When selecting alternative reference quantities, it is necessary to select from the set of alternative reference quantities in combination with civil engineering test data. Specifically, the design conditions are automatically found in a data-driven manner. Assume that there are m test data, recorded as and ,in Represents the jth data, according to Calculate the alternative reference quantity:

[0097]

[0098] Furthermore, the obtained candidate reference quantity can be normalized to the maximum and minimum, for a fixed k:

[0099]

[0100]

[0101] So we get the l+1-dimensional hypercube m points in To divide the design conditions (continuous branches), we only need to consider For each continuity.

[0102] S230: Perform spectral clustering on the candidate reference quantities to obtain spectral clustering classification results.

[0103] As mentioned above, spectral clustering is a classic unsupervised clustering algorithm that uses the correlation between data to group. Different from traditional K-means and hierarchical clustering methods, spectral clustering treats data samples as nodes in a graph, constructs a graph using the correlation between nodes, and then obtains clustering results through spectral decomposition of the graph. Spectral clustering focuses on the "continuity" between data points, that is, each cluster of data points in the classification result constitutes a continuous hypersurface, so spectral clustering can be used to divide the data set into continuous branches. As input, after spectral clustering, the data is divided into q categories:

[0104]

[0105] where c j Representative data points The category to which it belongs.

[0106] S240: Constructing a decision tree model using the candidate reference amount as input and the spectral clustering classification result as a category identifier.

[0107] Spectral clustering is a black box model like neural network. The classification criteria cannot be expressed explicitly to write into the design formula, so the classification results need to be further explained. The decision tree method for binary classification is introduced, and the pre-selected alternative reference quantity is used as the input data, and the classification result of spectral clustering is used as its category identifier to construct a decision tree model. After the model training is completed, the specific decision boundary is output. If the error between the decision boundary and the cluster boundary is within the allowable range, the data can be re-divided according to the decision boundary. Specifically, train a binary classification decision tree:

[0108]

[0109] Given an explicit classification condition, let Classification, The corresponding categories are , define the consistency coefficient between spectral clustering and decision tree classification:

[0110]

[0111] like , then it is believed that The classification is accurate.

[0112] S250: Divide the design conditions based on the constructed decision tree model to obtain a design condition classification result.

[0113] use The classification conditions are used as the basis for dividing the design conditions, and the civil engineering test data are re-divided according to different design conditions to obtain multiple sub-data sets. Subsequently, the formula is searched separately on the data set of each design condition; at the same time, The alternative reference quantity of the classification condition is the key factor affecting the mechanical properties of the material component, which can assist in the exploration of the mechanical mechanism. Specifically, in practical applications, the process and application of the design condition intelligent division algorithm are shown in Figure 5.

[0114] As shown in FIG. 6 , in one embodiment, S400 includes:

[0115] S420: Generate corresponding symbol sequences according to each sub-data set.

[0116] Specifically, the original civil engineering test data is divided into multiple sub-datasets according to different design conditions, and the formula is searched on each sub-dataset. The traditional symbolic regression algorithm regards the formula as a binary tree, and each tree node represents a symbol of the expression. Inheriting this method and combining it with Polish Notation to eliminate brackets, any formula can be converted into a sequence of category vectors, and each element in the sequence is a token in a mathematical dictionary. Take the conversion formula of the shear modulus and elastic modulus of isotropic elastic materials as an example:

[0117]

[0118] Converted to Polish notation:

[0119]

[0120] After the above transformation, any complex formula can be represented by a one-dimensional sequence. Here, for each sub-data set, the corresponding complex formula is represented by a symbol sequence in a similar manner to generate the corresponding symbol sequence.

[0121] S440: Input the symbol sequence into a formula generation intelligent model that takes physical constraints into consideration to obtain an output result, and extract the calculation results carried in each sub-data set.

[0122] Based on the above transformation, the formula regression problem can be analogized to a sequence generation problem in the field of natural language processing. Here, by introducing an intelligent model for formula generation that considers physical constraints, the output result is obtained, and the calculation results corresponding to different calculations are obtained based on the data carried in the sub-dataset. Specifically, the intelligent model for formula generation that considers physical constraints can be generated by using intelligent models such as long and short memory. In order to ensure that the generated symbol sequence can form a physical formula with reasonable meaning, the symbols generated by the model need to be restricted according to the position and context of the symbols. For each physical constraint, there needs to be a way to encode it into the model. This can be achieved by introducing additional loss function terms that penalize formulas that violate the laws of physics. Specifically, it is necessary to first ensure that the symbol sequence generated by the model conforms to the grammatical rules of the Polish expression, and infer whether an operator or a variable should be generated at a certain position through the partial symbol sequence generated previously, or whether the sequence has formed a complete Polish expression, thereby stopping the generation of the sequence. That is, when constructing the formula generation intelligent model, physical constraints need to be considered. The physical constraints mainly include two parts. One is to ensure that the generated symbol sequence conforms to the grammatical rules of Polish expressions. The other is to limit the choice of symbols through local and overall dimensional constraints to ensure that the model generates dimensionally balanced formulas to improve the interpretability of formula mechanics.

[0123] More specifically, in order to ensure that the generated symbol sequence can form a physical formula with reasonable meaning, the symbols generated by the model need to be restricted according to the position and context of the symbols. First, it is necessary to ensure that the symbol sequence generated by the model conforms to the grammatical rules of the Polish expression, and to infer whether an operator or a variable should be generated at a certain position, or whether the sequence has formed a complete Polish expression, through the partial symbol sequence generated previously, so as to stop the generation of the sequence. On this basis, the selection of symbols is restricted by local and overall dimensional constraints to ensure that the model generates dimensionally balanced formulas to improve its mechanical interpretability. Local dimensional constraints, such as the variables on the left and right sides of the addition and subtraction operators must have the same dimensions, and the operation objects of operators such as log, sin, and exp must be dimensionless quantities; overall dimensional constraints, that is, the dimensions of the generated formula must be the same as the dimensions of the fitting target quantity. The formula generation intelligent model generated based on the long short-term memory intelligent model can give the probability of each word at a certain position according to the context information, and consider the above-mentioned various constraints on the basis of this probability distribution, and output the word with the highest conditional probability.

[0124] S460: Iteratively update the formula generation intelligent model considering physical constraints based on the output results and the calculation results to obtain an updated formula generation intelligent model considering physical constraints.

[0125] When conducting a specific model training process, it is based on the model output results and the actual calculation results. The data results are converted into corresponding calculation formulas, that is, the relevant parameters are substituted into the calculation formula to obtain the formula calculation results; if there is a deviation between the formula calculation results and the calculation results, it indicates that the calculation formula generated in the formula generation intelligent model needs to be adjusted. At this time, the formula generation intelligent model is iteratively updated until the formula calculation results are finally equal to the calculation results, that is, a qualified formula generation intelligent model that considers physical constraints is generated.

[0126] S480: Input each sub-data set into the updated formula generation intelligent model considering physical constraints to obtain a sub-calculation formula.

[0127] By inputting sub-datasets of different design conditions into the formula generation intelligent model considering physical constraints, a response calculation formula with dimensional balance, strong mechanical interpretability and high accuracy can be generated. The calculation formulas of different design conditions are combined to obtain a complete design formula.

[0128] In specific application examples, the intelligent generation method of multi-condition design formulas for civil engineering in this application has the following technically significant effects:

[0129] 1) The data-driven intelligent partitioning algorithm for design conditions can process civil engineering test data with multiple mechanical and geometric variables, identify key factors affecting the mechanical properties of material components and automatically partition design conditions, thus simplifying the data set and assisting in the exploration of mechanical mechanisms.

[0130] 2) The intelligent model for formula generation that takes physical constraints into consideration can generate calculation formulas with dimensional balance, strong mechanical interpretability and high accuracy. It simplifies the process of formula discovery while ensuring the quality of the formula, thus greatly saving manpower and time costs.

[0131] 3) The entire method can intelligently and efficiently discover design calculation formulas from engineering data. The whole process has a high degree of automation, which greatly reduces the workload of manual fitting. The results have strong mechanical significance and can assist in exploring the mechanical mechanisms behind new materials and new structures, accelerating their practical application.

[0132] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0133] Based on the same inventive concept, the embodiment of the present application also provides a civil engineering multi-condition design formula intelligent generation device for realizing the civil engineering multi-condition design formula intelligent generation method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations of one or more civil engineering multi-condition design formula intelligent generation device embodiments provided below can refer to the limitations of the civil engineering multi-condition design formula intelligent generation method above, and will not be repeated here.

[0134] As shown in FIG7 , in one embodiment, the present application also provides a device for intelligently generating multi-condition design formulas for civil engineering. The device includes:

[0135] The data acquisition module 100 is used to acquire civil engineering test data and design conditions;

[0136] A division module 200 is used to divide the design conditions by using spectral clustering and decision tree methods to obtain a classification result of the design conditions;

[0137] A classification module 300 is used to classify the civil engineering test data based on the design condition classification result to obtain multiple sub-data sets;

[0138] A sub-formula generation module 400 is used to input each sub-data set into a formula generation intelligent model considering physical constraints to obtain a sub-calculation formula;

[0139] The merging module 500 is used to merge the sub-calculation formulas to generate a civil engineering multi-condition design formula.

[0140] In one embodiment, the partitioning module 200 is further used to select a set of alternative reference quantities for partitioning design conditions; calculate alternative reference quantities from the set of alternative reference quantities based on civil engineering test data; perform spectral clustering on the alternative reference quantities to obtain spectral clustering classification results; construct a decision tree model using the alternative reference quantities as input and the spectral clustering classification results as category identifiers; and partition the design conditions based on the constructed decision tree model to obtain design condition classification results.

[0141] In one embodiment, the partitioning module 200 is also used to obtain the dimensional matrix of the mechanical performance prediction problem in the civil engineering formula design, and the target quantity of the prediction problem; based on the dimensional matrix and the target quantity, a set of alternative reference quantities for partitioning the design conditions is selected.

[0142] In one embodiment, the partitioning module 200 is also used to obtain the mechanical property prediction problem in the civil engineering formula design; identify the characteristics of the mechanical property prediction problem and the target quantity of the prediction problem; and construct a dimensional matrix of the characteristics.

[0143] In one embodiment, the partitioning module 200 is further configured to perform maximum-minimum normalization on the candidate reference quantities.

[0144] In one of the embodiments, the sub-formula generation module 400 is also used to generate a corresponding symbol sequence according to each sub-data set; input the symbol sequence into the formula generation intelligent model considering physical constraints to obtain an output result, and extract the calculation results carried in each sub-data set; iteratively update the formula generation intelligent model considering physical constraints based on the output result and the calculation result to obtain an updated formula generation intelligent model considering physical constraints; input each sub-data set into the updated formula generation intelligent model considering physical constraints to obtain a sub-calculation formula.

[0145] In one of the embodiments, the sub-formula generation module 400 is also used to obtain the grammatical rules of the Polish expression and the local and global dimensional constraints, and the local and global dimensional constraints are used to limit the symbol selection so that the model generates dimensionally balanced formulas; based on the grammatical rules of the Polish expression and the local and global dimensional constraints, a long short-term memory model is used to generate an intelligent model for formula generation that takes physical constraints into account.

[0146] Each module in the above-mentioned civil engineering multi-condition design formula intelligent generation device can be fully or partially implemented by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.

[0147] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be shown in FIG8. The computer device includes a processor, a memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store preset data or historical data, etc. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for intelligently generating multi-conditional design formulas for civil engineering is implemented.

[0148] Those skilled in the art will understand that the structure shown in FIG. 8 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0149] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0150] Obtain civil engineering test data and design conditions;

[0151] The design conditions are divided by using spectral clustering and decision tree methods to obtain the classification results of design conditions;

[0152] Based on the design condition classification results, the civil engineering test data is classified to obtain multiple sub-datasets;

[0153] Input each sub-data set into a formula generation intelligent model considering physical constraints to obtain a sub-calculation formula;

[0154] Combine sub-calculation formulas to generate multi-condition design formulas for civil engineering.

[0155] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0156] A set of alternative reference quantities for dividing design conditions is selected; alternative reference quantities are calculated from the set of alternative reference quantities according to civil engineering test data; spectral clustering is performed on the alternative reference quantities to obtain spectral clustering classification results; a decision tree model is constructed using the alternative reference quantities as input and the spectral clustering classification results as category identifiers; design conditions are divided based on the constructed decision tree model to obtain design condition classification results.

[0157] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0158] Obtain the dimensional matrix of the mechanical performance prediction problem in civil engineering formula design and the target quantity of the prediction problem; based on the dimensional matrix and the target quantity, select a set of alternative reference quantities for dividing the design conditions.

[0159] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0160] Obtain the mechanical properties prediction problem in civil engineering formula design; identify the characteristics of the mechanical properties prediction problem and the target quantity of the prediction problem; construct the dimensional matrix of the characteristics.

[0161] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0162] Perform maximum-minimum normalization on the alternative reference amounts.

[0163] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0164] Generate a corresponding symbol sequence according to each sub-data set; input the symbol sequence into the formula generation intelligent model considering physical constraints to obtain the output result, and extract the calculation result carried in each sub-data set; iteratively update the formula generation intelligent model considering physical constraints based on the output result and the calculation result to obtain an updated formula generation intelligent model considering physical constraints; input each sub-data set into the updated formula generation intelligent model considering physical constraints to obtain a sub-calculation formula.

[0165] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0166] The grammatical rules of Polish expressions and local and global dimensional constraints are obtained. The local and global dimensional constraints are used to limit the symbol selection so that the model can generate dimensionally balanced formulas. Based on the grammatical rules of Polish expressions and local and global dimensional constraints, a long short-term memory model is used to generate an intelligent model for formula generation that takes physical constraints into account.

[0167] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0168] Obtain civil engineering test data and design conditions;

[0169] The design conditions are divided by using spectral clustering and decision tree methods to obtain the classification results of design conditions;

[0170] Based on the design condition classification results, the civil engineering test data is classified to obtain multiple sub-datasets;

[0171] Input each sub-data set into a formula generation intelligent model considering physical constraints to obtain a sub-calculation formula;

[0172] Combine sub-calculation formulas to generate multi-condition design formulas for civil engineering.

[0173] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0174] A set of alternative reference quantities for dividing design conditions is selected; alternative reference quantities are calculated from the set of alternative reference quantities according to civil engineering test data; spectral clustering is performed on the alternative reference quantities to obtain spectral clustering classification results; a decision tree model is constructed using the alternative reference quantities as input and the spectral clustering classification results as category identifiers; design conditions are divided based on the constructed decision tree model to obtain design condition classification results.

[0175] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0176] Obtain the dimensional matrix of the mechanical performance prediction problem in civil engineering formula design and the target quantity of the prediction problem; based on the dimensional matrix and the target quantity, select a set of alternative reference quantities for dividing the design conditions.

[0177] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0178] Obtain the mechanical properties prediction problem in civil engineering formula design; identify the characteristics of the mechanical properties prediction problem and the target quantity of the prediction problem; construct the dimensional matrix of the characteristics.

[0179] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0180] Perform maximum-minimum normalization on the alternative reference amounts.

[0181] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0182] Generate a corresponding symbol sequence according to each sub-data set; input the symbol sequence into the formula generation intelligent model considering physical constraints to obtain the output result, and extract the calculation result carried in each sub-data set; iteratively update the formula generation intelligent model considering physical constraints based on the output result and the calculation result to obtain an updated formula generation intelligent model considering physical constraints; input each sub-data set into the updated formula generation intelligent model considering physical constraints to obtain a sub-calculation formula.

[0183] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0184] The grammatical rules of Polish expressions and local and global dimensional constraints are obtained. The local and global dimensional constraints are used to limit the symbol selection so that the model can generate dimensionally balanced formulas. Based on the grammatical rules of Polish expressions and local and global dimensional constraints, a long short-term memory model is used to generate an intelligent model for formula generation that takes physical constraints into account.

[0185] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0186] Obtain civil engineering test data and design conditions;

[0187] The design conditions are divided by using spectral clustering and decision tree methods to obtain the classification results of design conditions;

[0188] Based on the design condition classification results, the civil engineering test data is classified to obtain multiple sub-datasets;

[0189] Input each sub-data set into a formula generation intelligent model considering physical constraints to obtain a sub-calculation formula;

[0190] Combine sub-calculation formulas to generate multi-condition design formulas for civil engineering.

[0191] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0192] A set of alternative reference quantities for dividing design conditions is selected; alternative reference quantities are calculated from the set of alternative reference quantities according to civil engineering test data; spectral clustering is performed on the alternative reference quantities to obtain spectral clustering classification results; a decision tree model is constructed using the alternative reference quantities as input and the spectral clustering classification results as category identifiers; design conditions are divided based on the constructed decision tree model to obtain design condition classification results.

[0193] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0194] Obtain the dimensional matrix of the mechanical performance prediction problem in civil engineering formula design and the target quantity of the prediction problem; based on the dimensional matrix and the target quantity, select a set of alternative reference quantities for dividing the design conditions.

[0195] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0196] Obtain the mechanical properties prediction problem in civil engineering formula design; identify the characteristics of the mechanical properties prediction problem and the target quantity of the prediction problem; construct the dimensional matrix of the characteristics.

[0197] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0198] Perform maximum-minimum normalization on the alternative reference amounts.

[0199] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0200] Generate a corresponding symbol sequence according to each sub-data set; input the symbol sequence into the formula generation intelligent model considering physical constraints to obtain the output result, and extract the calculation result carried in each sub-data set; iteratively update the formula generation intelligent model considering physical constraints based on the output result and the calculation result to obtain an updated formula generation intelligent model considering physical constraints; input each sub-data set into the updated formula generation intelligent model considering physical constraints to obtain a sub-calculation formula.

[0201] In one embodiment, when the computer program is executed by a processor, the following steps are also implemented:

[0202] The grammatical rules of Polish expressions and local and global dimensional constraints are obtained. The local and global dimensional constraints are used to limit the symbol selection so that the model can generate dimensionally balanced formulas. Based on the grammatical rules of Polish expressions and local and global dimensional constraints, a long short-term memory model is used to generate an intelligent model for formula generation that takes physical constraints into account.

[0203] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0204] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0205] The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for intelligently generating multi-condition design formulas for civil engineering, characterized in that: The method comprises: Obtain civil engineering test data and design conditions; The design conditions are divided by using spectral clustering and decision tree methods to obtain a design condition classification result; Classifying the civil engineering test data based on the design condition classification result to obtain multiple sub-data sets; Inputting each of the sub-data sets into a formula generation intelligent model taking into account physical constraints to obtain a sub-calculation formula; Combining the sub-calculation formulas to generate a civil engineering multi-condition design formula; The method of using spectral clustering and decision tree method to divide the design conditions to obtain the design condition classification results includes: selecting a set of alternative reference quantities for dividing the design conditions, where the alternative reference quantities are alternative physical quantities for dividing the design conditions; calculating alternative reference quantities from the set of alternative reference quantities according to the civil engineering test data; performing spectral clustering on the alternative reference quantities to obtain spectral clustering classification results; constructing a decision tree model using the alternative reference quantities as input and the spectral clustering classification results as category identifiers; and dividing the design conditions based on the constructed decision tree model to obtain design condition classification results; The step of inputting each of the sub-data sets into the formula generation intelligent model considering physical constraints to obtain the sub-calculation formula comprises: generating a corresponding symbol sequence according to each of the sub-data sets; inputting the symbol sequence into the formula generation intelligent model considering physical constraints to obtain an output result, and extracting the calculation result carried in each of the sub-data sets; iteratively updating the formula generation intelligent model considering physical constraints based on the output result and the calculation result to obtain an updated formula generation intelligent model considering physical constraints; inputting each of the sub-data sets into the updated formula generation intelligent model considering physical constraints to obtain the sub-calculation formula; Before inputting the symbol sequence into the formula generation intelligent model considering physical constraints to obtain the output result, it also includes: obtaining the grammatical rules of the Polish expression and the local and global dimensional constraints, wherein the local and global dimensional constraints are used to limit the symbol selection so that the model generates dimensionally balanced formulas; based on the grammatical rules of the Polish expression and the local and global dimensional constraints, a long short-term memory model is used to generate a formula generation intelligent model considering physical constraints.

2. The method according to claim 1, characterized in that The set of candidate reference quantities for selecting the partitioning design conditions includes: Obtain the dimensional matrix of the mechanical performance prediction problem in civil engineering formula design and the target quantity of the prediction problem; Based on the dimensional matrix and the target quantity, a set of candidate reference quantities for dividing design conditions is selected.

3. The method according to claim 2, characterized in that The method of obtaining the dimensional matrix of the mechanical performance prediction problem in the civil engineering formula design and the target quantity of the prediction problem includes: Obtain the problem of mechanical performance prediction in civil engineering formula design; Identifying the characteristics of the mechanical property prediction problem and the target quantity of the prediction problem; Construct a dimensional matrix of the features.

4. The method according to claim 1, characterized in that: Before performing spectral clustering on the candidate reference amount to obtain the spectral clustering classification result, the method further includes: The alternative reference amounts are subjected to maximum-minimum normalization.

5. An intelligent generation device for multi-condition design formulas for civil engineering, characterized in that: The device comprises: Data acquisition module, used to obtain civil engineering test data and design conditions; A partitioning module, used to partition the design conditions using spectral clustering and decision tree methods to obtain a classification result of the design conditions; A classification module, used for classifying the civil engineering test data based on the design condition classification result to obtain a plurality of sub-data sets; A sub-formula generation module, used for inputting each of the sub-data sets into a formula generation intelligent model taking physical constraints into consideration to obtain a sub-calculation formula; A merging module, used for merging the sub-calculation formulas to generate a civil engineering multi-condition design formula; The partitioning module is further used to select a set of alternative reference quantities for partitioning design conditions, where the alternative reference quantities are alternative physical quantities for partitioning design conditions; calculate alternative reference quantities from the set of alternative reference quantities according to the civil engineering test data; perform spectral clustering on the alternative reference quantities to obtain spectral clustering classification results; construct a decision tree model using the alternative reference quantities as input and the spectral clustering classification results as category identifiers; and partition the design conditions based on the constructed decision tree model to obtain design condition classification results; The sub-formula generation module is also used to generate corresponding symbol sequences according to each of the sub-data sets; input the symbol sequences into the formula generation intelligent model considering physical constraints to obtain output results, and extract the calculation results carried in each of the sub-data sets; iteratively update the formula generation intelligent model considering physical constraints based on the output results and the calculation results to obtain an updated formula generation intelligent model considering physical constraints; input each of the sub-data sets into the updated formula generation intelligent model considering physical constraints to obtain a sub-calculation formula; The sub-formula generation module is also used to obtain the grammatical rules of the Polish expression and the local and global dimensional constraints, and the local and global dimensional constraints are used to limit the symbol selection so that the model generates dimensionally balanced formulas; based on the grammatical rules of the Polish expression and the local and global dimensional constraints, a long short-term memory model is used to generate an intelligent model for formula generation that takes physical constraints into account.

6. The device according to claim 5, characterized in that The partitioning module is also used to obtain the dimensional matrix of the mechanical property prediction problem in civil engineering formula design, as well as the target quantity of the prediction problem; Based on the dimensional matrix and the target quantity, a set of alternative reference quantities for dividing the design conditions is selected.

7. The device according to claim 6, characterized in that The partitioning module is also used to obtain mechanical property prediction problems in civil engineering formula design; identify the characteristics of mechanical property prediction problems and the target quantity of the prediction problem; and construct a dimensional matrix of the characteristics.

8. The device according to claim 5, characterized in that The partitioning module is also used to perform maximum-minimum normalization on the candidate reference quantities.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.