Retaining wall three-dimensional design decision support system and method based on cloud computing

Abstract

The invention belongs to the technical field of engineering design, and discloses a retaining wall three-dimensional design decision support system and method based on cloud computing. Comprising the steps that a cloud platform receives design condition data sent by a user side; according to the design condition data, the structure type of the retaining wall is selected, and the arrangement form of the retaining wall is determined; fusing the design condition data, the structure type and the arrangement form, and designing a groups of retaining wall size data; constructing a corresponding three-dimensional model for each group of retaining wall data, performing structural analysis, and screening out b groups of excellent size data according to a structural analysis result; receiving design size data sent by the user side, matching the design size data with the excellent size data, and screening out optimal size data; based on the optimal size data, optimal steel bar configuration is generated; the optimal design scheme can be automatically generated, so that the design process of the retaining wall is more flexible and quick in response, and the safety and stability of the retaining wall are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of engineering design, and more specifically, to a three-dimensional design decision support system and method for retaining walls based on cloud computing. Background Art

[0002] In the field of modern civil engineering, retaining walls, as one of the important infrastructure facilities, are widely used in highways, railways, urban infrastructure, and water conservancy projects; the main function of retaining walls is to resist the action of soil or other external loads, prevent soil landslides, settlements or collapses, thus ensuring the safety of surrounding facilities and personnel; with the acceleration of the urbanization process, the design and construction of retaining walls are facing increasingly complex environmental conditions, design requirements and engineering challenges; traditional retaining wall design methods usually rely on manual calculations, two-dimensional modeling and empirical decision-making, and engineers complete the design through static two-dimensional drawings and models based on existing design experience and design specifications; there are problems such as low efficiency, limited design perspective, and insufficient decision support; therefore, there is an urgent need for an intelligent three-dimensional design method to improve the overall efficiency and quality of retaining wall design.

[0003] The patent with publication number CN110704895B discloses a parametric modeling method for retaining walls based on a three-dimensional analysis platform; it includes: establishing a project information overview and inputting project exploration data, establishing a three-dimensional geological model, and analyzing the exploration geological report to construct a rapid overall framework and mode of the retaining wall; establishing a three-dimensional geological model through Civil3D, setting three-dimensional parameters of the retaining wall, and through the analysis of the retaining wall structure, constructing an analysis platform that organically combines three-dimensional design technology, parametric design technology, calculation and analysis functions based on specifications and finite element methods for the integrated design and construction of retaining walls, so that the simulation analysis of the actual site conditions and construction plan is consistent, and the relevant on-site data can be timely fed back to the integrated platform, and data sharing among functional modules reduces data transfer losses and improves work efficiency, achieving the four aspects of parameterization, visualization, dynamicization, and seamlessization.

[0004] However, although the above technology can realize the three-dimensional design of retaining walls and perform analysis and simulation by establishing a three-dimensional geological model and a parametric design platform, it mainly focuses on model construction and structural analysis and lacks flexible design decision-making capabilities; in the face of complex and variable design conditions, it cannot perform real-time optimization and intelligent adjustment according to the specific design needs of users; therefore, its adaptability and decision support capabilities in a dynamic design environment are limited, especially in the selection, adjustment, and optimization of design schemes, lacking a sufficient intelligent decision-making mechanism, resulting in insufficient design efficiency and accuracy to meet complex design requirements.

[0005] In view of this, the present invention provides a three-dimensional design decision support system and method for retaining walls based on cloud computing to solve the above problems. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art and achieve the above object, the present invention provides the following technical solutions: A three-dimensional design decision support method for retaining walls based on cloud computing, which is applied to a cloud platform and includes:

[0007] Receiving design condition data sent by a user terminal;

[0008] According to the design condition data, selecting the structural type of the retaining wall, and determining the layout form of the retaining wall based on the structural type;

[0009] Integrating the design condition data, the structural type and the layout form, designing a set of retaining wall dimension data, where a is an integer greater than 1; constructing a corresponding three-dimensional model for each set of retaining wall data, and performing structural analysis, screening out b sets of retaining wall dimension data according to the structural analysis results, and marking them as excellent dimension data, where 1 < b < a;

[0010] Receiving design dimension data sent by the user terminal, matching the design dimension data with the excellent dimension data, screening out the excellent dimension data with the highest matching degree, and marking it as the best dimension data;

[0011] Based on the three-dimensional model corresponding to the best dimension data, using a multi-objective optimization algorithm to automatically generate the optimal steel bar configuration in the three-dimensional model, and sending the three-dimensional model to the corresponding user terminal.

[0012] Furthermore, the design condition data includes material data and boundary data; the material data includes retaining wall material data and geotechnical material data;

[0013] The steps of selecting the structural type of the retaining wall include:

[0014] Step S101: Constructing multiple fuzzy sets for each data in the design condition data;

[0015] Step S102: Converting each data in the design condition data into the membership degree of the corresponding fuzzy set through a fuzzification technique respectively;

[0016] Step S103: Defining fuzzy rules;

[0017] Step S104: Matching the fuzzified design condition data with the fuzzy rules, performing fuzzy inference, and obtaining a fuzzy inference result, where the fuzzy inference result is the membership degree corresponding to each structural type;

[0018] Step S105: Compare each membership degree in the fuzzy inference result, and select the structural type corresponding to the membership degree with the largest value as the structural type of the retaining wall;

[0019] The method for determining the layout form of the retaining wall includes:

[0020] Set different digital tags for the text data in the design condition data and mark them as condition tags. The text data is the data in the design condition data that is not a numerical value; set different digital tags for different structural types and mark them as type tags; replace all the text data in the design condition data with the corresponding condition tags, and mark the design condition data after replacement as design replacement data; according to the selected structural type of the retaining wall, obtain the corresponding type tag and mark it as the selected tag; use the design replacement data and the selected tag as analysis data, input the analysis data into the trained form determination model, and obtain the corresponding form tag. The form tag is the digital tag corresponding to the layout form, and the digital tags corresponding to different layout forms are all different; obtain the corresponding layout form according to the form tag and determine it as the layout form of the retaining wall; the form determination model is a deep neural network model.

[0021] Further, the method for designing a set of retaining wall dimension data includes:

[0022] Use the analysis data and the form tag as design data, input the design data into the trained range prediction model, and predict the corresponding range tag. The range tag is the digital tag corresponding to the dimension range, and the digital tags corresponding to different dimension ranges are all different; the range prediction model is a deep neural network model; according to the predicted range tag, obtain the corresponding dimension range. The dimension range includes the range corresponding to each data in the retaining wall dimension data; randomly select a value from each range within the dimension range to construct a set of retaining wall dimension data, and a total of a sets of retaining wall dimension data are constructed, and the a sets of retaining wall dimension data are all different;

[0023] Use a 3D modeling tool to construct a corresponding 3D model for each set of retaining wall data; the structural analysis includes structural stability calculation and internal force calculation;

[0024] The method for screening out b sets of retaining wall dimension data according to the structural analysis results includes:

[0025] A preset weight set, where the weight set includes the weight coefficients corresponding to each type of data in the structural analysis results; multiply each data in the a groups of retaining wall size data by the corresponding weight coefficient, and then add them in sequence to obtain the structural factor corresponding to each group of retaining wall size data; sort each structural factor from largest to smallest to generate a factor ranking table; screen out the top b structural factors from the factor ranking table and mark them as excellent factors, and obtain the retaining wall size data corresponding to the excellent factors.

[0026] Further, the designed size data is the retaining wall size data designed by the user;

[0027] The method for matching the designed size data with the excellent size data includes:

[0028] Take the designed size data and the excellent size data as calculation data, and take the data of the same type in the calculation data as a group of data sets, and the data sets correspond one by one to the data in the retaining wall size data; obtain the largest value in each group of data sets and mark it as the maximum value, obtain the smallest value in each group of data sets and mark it as the minimum value; subtract the corresponding minimum value from the maximum value of each group of data sets to obtain the value range corresponding to each group of data sets; subtract the minimum value of the corresponding data set from each data in the calculation data, and then divide by the data range of the corresponding data set to obtain the standard value corresponding to each data in the calculation data; subtract the standard value corresponding to the corresponding data in the designed size data from the standard value corresponding to each data in each group of excellent size data to obtain the data difference corresponding to each data in the excellent size data; add up the data differences corresponding to each group of excellent size data in sequence to obtain the total data difference corresponding to each group of excellent size data, and take the reciprocal of the total data difference as the matching degree of the corresponding excellent size data.

[0029] Further, the steps for automatically generating the optimal steel bar configuration in the 3D model include:

[0030] Step S201: Construct d groups of configuration sets, set different digital labels for each group of configuration sets, and mark them as set labels, and the range of the set labels is [1, d];

[0031] Step S202: Initialize the population. The population includes n individuals, and the n individuals are all in one-to-one correspondence with the set labels. Initialize the iteration times t corresponding to the population, where t = 0, and 1 < n < d;

[0032] Step S203: Define the iteration threshold T;

[0033] Step S204: Calculate the ideal value set corresponding to each individual;

[0034] Step S205: Divide all individuals into groups;

[0035] Step S206: Calculate the sparsity corresponding to each individual;

[0036] Step S207: Construct a dynamic boundary, calculate the reverse individual corresponding to each individual, and regard the reverse individual as an individual;

[0037] Step S208: Calculate the screening probability of each individual;

[0038] Step S209: Screen out n individuals from all individuals;

[0039] Step S210: Compare the iteration number with the iteration threshold. If t≥T, go to Step S211. If t<T, set t = t + 1 and return to Step S204;

[0040] Step S211: Screen out the best individual from all individuals, and obtain the set label corresponding to the best individual; according to the configuration set corresponding to the obtained set label, automatically generate the optimal steel bar configuration in the 3D model.

[0041] Further, in the said Step S201, the method for constructing d groups of configuration sets is as follows: Obtain the steel bar configuration range, where the steel bar configuration range includes the range values corresponding to each type of data in the steel bar configuration data; randomly select a value from each range in the steel bar configuration range to construct a group of configuration sets, and a total of d groups of configuration sets are constructed. The d groups of configuration sets are all different, and d>1;

[0042] In the said Step S202, the method for generating n individuals in the initialized population is as follows: Subtract the minimum value from the maximum value in the range corresponding to the set label to obtain the set interval; randomly generate n values from the interval [0,1] and mark them as random coefficients, and the random coefficients correspond to the individuals one by one; multiply the set interval by each random coefficient and add the minimum value of the set label to generate n individuals;

[0043] In the said Step S204, the method for calculating the ideal value set corresponding to each individual is as follows: Obtain the configuration set corresponding to the set label corresponding to each individual, and regard the design data, the best size data, and a group of configuration sets as a group of analysis data; input each group of analysis data into the trained performance analysis model respectively to predict the corresponding performance parameters; where, there are l performance prediction models in the performance analysis model, l is the number of types of parameters in the performance parameters, the performance prediction models in the performance analysis model correspond to the parameters in the performance parameters one by one, and each performance prediction model is a deep neural network model; regard the performance parameters corresponding to each group of analysis data as the ideal value set corresponding to the corresponding individual; where, each parameter in the performance parameters is the ideal value in the corresponding ideal value set.

[0044] Further, in step S205, the steps of dividing all individuals into groups include:

[0045] Step S301: Take every two individuals as a set of individuals;

[0046] Step S302: According to the ideal value sets corresponding to each individual, sequentially determine whether there is an individual dominating another individual in each set of individuals, and obtain the judgment result;

[0047] Step S303: According to the judgment result, obtain the dominated set corresponding to each individual;

[0048] Step S304: Divide the individuals for which there are no individuals in the dominated set into one group, mark it as the current group, mark the individuals in the current group as the current individuals, and delete all the current individuals in the dominated sets;

[0049] Step S305: Divide the individuals for which there is only the current individual in the dominated set into one group, mark it as the updated group, update the current group to the updated group, update the current individuals to the individuals in the updated group, and delete all the current individuals in the dominated sets;

[0050] Step S306: Loop step S305 until all individuals have completed group division. When the loop ends, sequentially increment and set digital labels from the earliest to the latest according to the order in which each group is marked as the current group, and mark them as group labels. The range of the group labels is [1, g], where g is the number of groups;

[0051] In step S302, the method for obtaining the judgment result includes:

[0052] Mark the two ideal value sets corresponding to each set of individuals as the first set and the second set respectively, and compare the ideal values of the same type in the first set and the second set; if all the ideal values in the first set are less than or equal to the corresponding ideal values in the second set, and there is an ideal value in the second set that is greater than the corresponding ideal value in the first set, then the individual corresponding to the second set dominates the individual corresponding to the first set; if there is an ideal value in the first set that is less than or equal to the corresponding ideal value in the second set, and there is an ideal value in the first set that is greater than the corresponding ideal value in the second set, then the individual corresponding to the first set and the individual corresponding to the second set do not dominate each other.

[0053] Further, in step S206, the expression for the sparsity is:

[0054]

[0055] In the formula, is the sparsity of the i-th individual in the t-th iteration process, is the k-th ideal value in the ideal value set corresponding to the (i + 1)-th individual in the t-th iteration process, P k (max) is the k-th ideal value with the largest value in all ideal value sets, P k (min) is the k-th ideal value with the smallest value in all ideal value sets, k ∈ [1, l];

[0056] In the step S207, the method for constructing the dynamic boundary is as follows: compare the set labels corresponding to all individuals, mark the set label with the largest value as the maximum label, and use the set label with the smallest value as the minimum label; construct the dynamic boundary according to the maximum label and the minimum label, that is, the maximum value of the dynamic boundary is the maximum label, and the minimum value is the minimum label;

[0057] The method for calculating the reverse individual corresponding to each individual is as follows: randomly generate n values from the interval [0, 1] and mark them as random factors; add the maximum label and the minimum label to obtain the reverse factor, and the reverse factor corresponds to each individual one by one; multiply each reverse factor by the corresponding random factor, and then subtract the set label of the corresponding individual to obtain the reverse individual corresponding to each individual; among them, if the reverse individual corresponding to an individual is not within the dynamic boundary, mark the corresponding individual as an out-of-bound individual, and randomly generate a value within the dynamic boundary using a random function and use it as the reverse individual of the out-of-bound individual;

[0058] In the step S208, the method for calculating the screening probability of each individual includes:

[0059] Preset a ratio set, where the ratio set includes the ratio coefficient corresponding to each ideal value in the ideal value set; according to the ratio set, obtain the ratio coefficient corresponding to each ideal value in the ideal value set, multiply each ideal value in the ideal value set corresponding to each individual by the corresponding ratio coefficient, and add them up in sequence to obtain the total ideal value corresponding to each individual; add up the total ideal values of each individual in sequence to obtain the population ideal value; divide the total ideal value of each individual by the population ideal value to obtain the screening probability of each individual.

[0060] Furthermore, in the step S209, the method for screening out n individuals from all individuals includes:

[0061] Sort the screening probabilities of each individual from largest to smallest to generate a probability ranking table, and mark the individual corresponding to the screening probability ranked first in the probability ranking table as the first individual; add each screening probability to all the screening probabilities before the corresponding screening probability to obtain the screening ratio corresponding to each screening probability; replace each screening probability in the probability ranking table with the corresponding screening ratio, and according to the positive order of the probability ranking table, take every two adjacent screening ratios as a set of ratio sets in turn; compare the screening ratios in each set of ratio sets, mark the screening ratio with a larger value as the first ratio, and mark the screening ratio with a smaller value as the second ratio; construct the screening range corresponding to each set of ratio sets according to the first ratio and the second ratio in each set of ratio sets; wherein, the first ratio is the maximum value of the screening range, the second ratio is the minimum value of the screening range, and the maximum value of the screening range is a closed interval, and the minimum value is an open interval; each set of ratio sets corresponds to the individual corresponding to the second ratio, and the minimum value of the screening range corresponding to the first individual is 0, and the maximum value is the corresponding screening probability; randomly generate numerical values from the interval [0, 1] and mark them as random factors; from all individuals, screen out the individuals corresponding to the screening range where the random factor is located and mark them as preliminarily screened individuals;

[0062] Sort each preliminarily screened individual from largest to smallest according to the corresponding group label to generate an individual ranking table; wherein, if the group labels of multiple preliminarily screened individuals are the same, sort them from largest to smallest according to the corresponding sparsity; according to the positive order of the individual ranking table, screen out n individuals from all preliminarily screened individuals;

[0063] In the step S211, the method for screening out the best individual from all individuals is: compare the ideal total values of all individuals, screen out the individual with the largest ideal total value from all individuals, and mark it as the best individual.

[0064] A three-dimensional design decision support system for retaining walls based on cloud computing, implementing the three-dimensional design decision support method based on cloud computing, and applied to a cloud platform, including:

[0065] A data receiving module for receiving design condition data sent by the user terminal;

[0066] A structure design module for selecting the structure type of the retaining wall according to the design condition data and determining the layout form of the retaining wall based on the structure type;

[0067] A dimension design module, which is used to integrate design condition data, structural types, and layout forms to design a set of retaining wall dimension data, where a is an integer greater than 1; for each set of retaining wall data, a corresponding 3D model is constructed and structural analysis is carried out. According to the results of the structural analysis, b sets of retaining wall dimension data are selected and marked as excellent dimension data, where 1 < b < a;

[0068] A dimension matching module, which is used to receive the design dimension data sent by the user terminal, match the design dimension data with the excellent dimension data, select the excellent dimension data with the highest matching degree, and mark it as the best dimension data;

[0069] A steel bar configuration module, which is used to automatically generate the optimal steel bar configuration in the 3D model based on the 3D model corresponding to the best dimension data and send the 3D model to the corresponding user terminal.

[0070] The technical effects and advantages of the 3D design decision support system and method for retaining walls based on cloud computing of the present invention:

[0071] By receiving the design condition data of the user terminal and combining advanced fuzzy logic algorithms and deep learning technologies, the intelligent design of the appearance contour of the retaining wall is realized; at the same time, a multi-objective optimization algorithm is used to realize the optimization of the steel bar configuration inside the retaining wall, improving the efficiency, accuracy, and intelligent level of the 3D design of the retaining wall; making full use of the advantages of cloud computing, realizing remote online design, and integrating intelligent decision-making modules such as the selection of structural types and layout forms, dimension optimization, and steel bar configuration optimization, which can automatically generate the optimal design scheme that meets the user's needs. At the same time, 3D models are established to provide users with an intuitive display of the design effect; making the design process of the retaining wall more flexible and responsive, ensuring the safety and stability of the retaining wall, and thus adapting to the complex and changeable construction environment. Description of the Drawings

[0072] Figure 1 It is a schematic diagram of the 3D design decision support system for retaining walls based on cloud computing according to Embodiment 1 of the present invention;

[0073] Figure 2 It is a flowchart of the 3D design decision support method for retaining walls based on cloud computing according to Embodiment 2 of the present invention. Detailed Embodiments

[0074] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0075] Embodiment 1

[0076] Please refer to Figure 1 As shown, the cloud computing-based three-dimensional design decision support system for retaining walls in this embodiment includes a data receiving module, a structure design module, a dimension design module, a dimension matching module, and a steel bar configuration module; each module is connected by wired and / or wireless means to achieve data transmission between modules.

[0077] The data receiving module is used to receive the design condition data sent by the user terminal on the cloud platform.

[0078] The cloud platform is a platform that provides cloud computing services; the user terminal is a user who conducts three-dimensional design of retaining walls; the design condition data includes material data and boundary data;

[0079] The material data includes retaining wall material data and geotechnical material data; the retaining wall material data such as the type of wall material (such as concrete, reinforced concrete, stone, etc.), the physical properties of the wall (such as density, water absorption, thermal conductivity, etc.), the mechanical properties of the wall (such as compressive strength, elastic modulus, etc.), etc.; the geotechnical material data such as the type of geotechnical material (such as clay, sand, gravel, etc.), the physical properties of the geotechnical material (such as soil particle composition, water content, etc.), the mechanical properties of the geotechnical material (internal friction angle, quasi-plasticity, etc.), etc.; the material data is obtained by the user through on-site measurement, literature query and other means.

[0080] The boundary data such as load conditions, soil slope profile, water level, geological conditions, etc.; the load conditions are various loads acting on the retaining wall, including static loads and dynamic loads, such as soil pressure, water pressure, etc.; the soil slope profile is the shape and height of the soil slopes in front of and behind the retaining wall; the water level is the height of groundwater or surface water; the geological conditions are the geological characteristics of the area where the retaining wall is located, such as soil layer distribution (such as soil layer thickness, soil layer type), rock layer characteristics (such as degree of fragmentation, arrangement method), etc.; the boundary data is obtained by the user through on-site measurement, laboratory tests, geological exploration and other means.

[0081] It should be noted that the purpose of obtaining the design condition data is to facilitate the subsequent optimization of the design scheme of the retaining wall through accurate material and boundary data, so as to ensure the safety and stability of the retaining wall and meet the requirements of engineering specifications.

[0082] The structure design module is used to select the structure type of the retaining wall based on the design condition data on the cloud platform, and determine the layout form of the retaining wall based on the structure type.

[0083] The structure types of the retaining wall such as gravity retaining wall, cantilever retaining wall, counterfort retaining wall, open caisson retaining wall, etc.; the layout forms of the retaining wall such as linear layout, curved layout, stepped layout, etc.

[0084] The steps for selecting the structural type of the retaining wall include:

[0085] Step S101: Construct multiple fuzzy sets for each piece of data in the design condition data; for example, the fuzzy sets corresponding to water absorption are low water absorption, medium water absorption, high water absorption, etc., the fuzzy sets corresponding to soil layer thickness are thick soil layer thickness, medium soil layer thickness, thin soil layer thickness, etc., and the fuzzy sets corresponding to the type of geotechnical material are clay, sand, gravel, etc.;

[0086] Step S102: Convert each piece of data in the design condition data into the membership degree of the corresponding fuzzy set through a fuzzification technique respectively; Fuzzification is the process of converting an exact numerical value into the membership degree corresponding to a fuzzy set, and fuzzification techniques include, for example, triangular membership functions, trapezoidal membership functions, etc.; for example, if the numerical value of water absorption is low, it is inferred that the membership degree of low water absorption is 0.9, the membership degree of medium water absorption is 0.1, and the membership degree of high water absorption is 0;

[0087] Step S103: Define fuzzy rules, and the fuzzy rules are defined based on expert knowledge or relevant literature; for example, if the water absorption is medium, the soil layer thickness is thick, and the type of geotechnical material is gravel, it is inferred that the probability of the structural type of the retaining wall being an empty box retaining wall is high; the empty box retaining wall is suitable for relatively thick soil layers, and the empty box structure can reduce the self-weight and provide good drainage effects; if the water absorption is medium, the soil layer thickness is thin, and the type of geotechnical material is sand, it is inferred that the probability of the structural type of the retaining wall being a cantilever retaining wall is high; the cantilever retaining wall is suitable for areas with thin soil layers, and the cantilever design can effectively support the soil;

[0088] Step S104: Match the fuzzified design condition data with the fuzzy rules, perform fuzzy inference, and obtain the fuzzy inference result. The fuzzy inference result is the membership degree corresponding to each structural type; fuzzy inference methods include, for example, Mamdani or Sugeno fuzzy inference methods; the fuzzy inference result is, for example, the membership degree of the gravity retaining wall is 0.1, the membership degree of the counterfort retaining wall is 0.3, and the membership degree of the cantilever retaining wall is 0.6;

[0089] Step S105: Compare each membership degree in the fuzzy inference result, and select the structural type corresponding to the membership degree with the largest numerical value as the structural type of the retaining wall.

[0090] The method for determining the layout form of the retaining wall includes:

[0091] Different digital tags are set for the text data in the design condition data and marked as condition tags. The text data are the data in the design condition data that are not numerical values, such as wall material types, arrangement methods, etc.; different digital tags are set for different structural types and marked as type tags; the text data in the design condition data are all replaced with the corresponding condition tags, and the design condition data after replacement are marked as design replacement data; according to the selected structural type of the retaining wall, the corresponding type tag is obtained and marked as the selected tag; the design replacement data and the selected tag are used as analysis data, and the analysis data are input into the trained form determination model to obtain the corresponding form tag. The form tag is the digital tag corresponding to the layout form, and the digital tags corresponding to different layout forms are all different; the corresponding layout form is obtained according to the form tag and determined as the layout form of the retaining wall.

[0092] The training process of the form determination model includes:

[0093] Pre-collect c sets of analysis data, and set corresponding form tags for the c sets of analysis data. c is an integer greater than 1. The analysis data and the corresponding form tags are converted into a corresponding set of feature vectors; the form tags corresponding to the analysis data are determined by those skilled in the art in the process of historically determining the layout form of the retaining wall. c sets of analysis data are collected, and each set of analysis data is analyzed in turn in combination with the actual situation to determine the corresponding layout form, and corresponding form tags are set for the c sets of analysis data in turn;

[0094] Each set of feature vectors is used as the input of the form determination model. The form determination model outputs a set of predicted form tags corresponding to each set of analysis data, and uses the actual form tag corresponding to each set of analysis data as the prediction target. The actual form tag is the form tag preset corresponding to the analysis data; minimizing the sum of the prediction errors of all analysis data is used as the training target; among them, the calculation formula of the prediction error is η w =(θ w -ε w ) 2 , where η w is the prediction error, w is the group number of the feature vector corresponding to the analysis data, θ w is the predicted form tag corresponding to the wth set of analysis data, and ε w is the actual form tag corresponding to the wth set of analysis data; the form determination model is trained until the sum of the prediction errors reaches convergence and then the training stops.

[0095] The above-mentioned form determination model is specifically a deep neural network model, which includes an input layer, a hidden layer, and an output layer. Each hidden layer contains multiple neurons, and there are connections between each neuron and the neurons in the next layer. These connections include weights, which determine the importance and influence of data transmission in the neural network. An activation function is applied to each neuron between the hidden layer and the output layer. The activation function introduces non-linearity and allows the network to learn more complex patterns and features.

[0096] The dimension design module is used to integrate design condition data, structure types, and layout forms on the cloud platform to design a set of retaining wall dimension data, where a is an integer greater than 1. For each set of retaining wall data, a corresponding 3D model is constructed and structural analysis is performed. According to the results of the structural analysis, b sets of retaining wall dimension data are selected and marked as excellent dimension data, where 1 < b < a.

[0097] The retaining wall dimension data includes, for example, vertical height, bottom width of the wall, top width of the wall, inclination angle, wall thickness, etc. The method for designing a set of retaining wall dimension data includes:

[0098] Taking the analysis data and form labels as design data, inputting the design data into the trained range prediction model to predict the corresponding range label. The range label is a digital label corresponding to the dimension range, and different dimension ranges correspond to different digital labels. The training process of the range prediction model is the same as that of the form determination model, and both are deep neural network models. According to the predicted range label, the corresponding dimension range is obtained. The dimension range includes the range corresponding to each type of data in the retaining wall dimension data. A random value is selected from each range within the dimension range to construct a set of retaining wall dimension data, and a total of a sets of retaining wall dimension data are constructed, and the a sets of retaining wall dimension data are all different.

[0099] Using 3D modeling tools (such as AutoCAD, Revit, SketchUp) to construct corresponding 3D models for each set of retaining wall data respectively. The structural analysis includes structural stability calculation and internal force calculation. The structural stability calculation includes, for example, anti-sliding stability, foundation bearing capacity, overall anti-sliding stability calculation, etc. The internal force calculation includes, for example, the calculation of bending moment and shear force of structures such as the retaining wall bottom slab, vertical slab, and buttress. The results of the structural analysis include anti-sliding stability, foundation bearing capacity, overall anti-sliding stability, bottom slab bending moment value, vertical slab bending moment value, vertical slab shear force value, buttress shear force value, etc. The structural analysis is automatically implemented through integrated software such as ANSYS, PLAXIS, SAP2000, and ETABS.

[0100] The method for screening out b sets of retaining wall dimension data according to the results of the structural analysis includes:

[0101] A preset weight set, which includes the weight coefficients corresponding to each type of data in the structural analysis results. The weight set is preset by those skilled in the art according to the actual situation; multiply each data in the a groups of retaining wall size data by the corresponding weight coefficient, and then add them up in sequence to obtain the structural factor corresponding to each group of retaining wall size data; sort each structural factor from large to small to generate a factor sorting table; screen out the top b structural factors from the factor sorting table and mark them as excellent factors, and obtain the retaining wall size data corresponding to the excellent factors.

[0102] A size matching module, which is used for the cloud platform to receive the design size data sent by the user terminal, match the design size data with the excellent size data, screen out the excellent size data with the highest matching degree, and mark it as the best size data.

[0103] The design size data is the retaining wall size data designed by the user;

[0104] The method of matching the design size data with the excellent size data includes:

[0105] Take the design size data and the excellent size data as calculation data, and take the data of the same type in the calculation data as a group of data sets, and the data sets correspond one by one to the data in the retaining wall size data; obtain the largest value in each group of data sets and mark it as the maximum value, obtain the smallest value in each group of data sets and mark it as the minimum value; subtract the corresponding minimum value from the maximum value of each group of data sets to obtain the value range corresponding to each group of data sets; subtract each data in the calculation data from the minimum value of the corresponding data set, and then divide it by the data range of the corresponding data set to obtain the standard value corresponding to each data in the calculation data; subtract the standard value corresponding to the corresponding data in the design size data from the standard value corresponding to each data in each group of excellent size data to obtain the data difference corresponding to each data in the excellent size data; add up the data differences corresponding to each group of excellent size data in sequence to obtain the total data difference corresponding to each group of excellent size data, and take the reciprocal of the total data difference as the matching degree of the corresponding excellent size data.

[0106] A steel bar configuration module, which is used for the cloud platform to automatically generate the optimal steel bar configuration in the 3D model based on the 3D model corresponding to the best size data, and send the 3D model to the corresponding user terminal.

[0107] The steps of automatically generating the optimal steel bar configuration in the 3D model include:

[0108] Step S201: Construct d groups of configuration sets, set different digital tags for each group of configuration sets respectively, and mark them as set tags. The range of the set tags is [1, d];

[0109] Step S202: Initialize the population. The population includes n individuals, and each of the n individuals corresponds one-to-one with the set labels. Initialize the iteration count t corresponding to the population to 0, where 1 < n < d;

[0110] Step S203: Define the iteration threshold T, which is set by those skilled in the art according to the algorithm accuracy requirements;

[0111] Step S204: Calculate the ideal value set corresponding to each individual;

[0112] Step S205: Classify all individuals into groups;

[0113] Step S206: Calculate the sparsity corresponding to each individual;

[0114] Step S207: Construct the dynamic boundary, calculate the reverse individual corresponding to each individual, and regard the reverse individual as an individual as well;

[0115] Step S208: Calculate the screening probability of each individual;

[0116] Step S209: Select n individuals from all individuals;

[0117] Step S210: Compare the iteration count with the iteration threshold. If t ≥ T, go to Step S211. If t < T, set t = t + 1 and return to Step S204;

[0118] Step S211: Select the best individual from all individuals, and obtain the set label corresponding to the best individual; automatically generate the optimal steel bar configuration in the 3D model according to the configuration set corresponding to the obtained set label.

[0119] In the above Step S201, the method for constructing d groups of configuration sets is as follows: Obtain the steel bar configuration range, which includes the range values corresponding to each type of data in the steel bar configuration data; the steel bar configuration data includes, for example, the material label, steel bar diameter, steel bar quantity, steel bar spacing, etc. Among them, the material label is the digital label corresponding to the steel bar material, and the digital labels corresponding to different steel bar materials are all different. The steel bar materials include, for example, main steel bars, distribution steel bars, stirrups, etc.; randomly select a value from each range in the steel bar configuration range to construct a group of configuration sets, and a total of d groups of configuration sets are constructed. The d groups of configuration sets are all different, where d > 1.

[0120] In the above Step S202, the method for generating the n individuals in the initialized population is as follows: Subtract the minimum value from the maximum value in the corresponding range of the set label to obtain the set interval; randomly generate n values from the interval [0, 1] and mark them as random coefficients, and the random coefficients correspond one-to-one with the individuals; multiply the set interval by each random coefficient and add the minimum value of the set label to generate n individuals.

[0121] In the above step S204, the method for calculating the ideal value set corresponding to each individual is as follows: Obtain the configuration set corresponding to the set label of each individual, and use the design data, the optimal dimension data, and a set of configuration sets as a set of analysis data; Input each set of analysis data into the trained performance analysis model respectively to predict the corresponding performance parameters; Among them, the performance parameters such as structural strength, crack resistance performance, shear resistance ability, etc., there are l performance prediction models in the performance analysis model, l is the number of types of parameters in the performance parameters, the performance prediction models in the performance analysis model correspond one by one to the parameters in the performance parameters, the training process of each performance prediction model is the same as that of the form determination model, and both are deep neural network models; Use the performance parameters corresponding to each set of analysis data as the ideal value set corresponding to the corresponding individual; Among them, each parameter in the performance parameters is the ideal value in the corresponding ideal value set.

[0122] In the above step S205, the steps for grouping all individuals include:

[0123] Step S301: Take every two individuals as a set of individual sets;

[0124] Step S302: According to the ideal value set corresponding to each individual, sequentially determine whether there is an individual dominating another individual in each set of individual sets, and obtain the judgment result;

[0125] Step S303: According to the judgment result, obtain the dominated set corresponding to each individual; Among them, the dominated set of an individual includes all individuals that dominate it.

[0126] Step S304: Divide the individuals whose dominated sets do not have individuals into one group, and mark it as the current group, mark the individuals in the current group as the current individuals, and delete all the current individuals in the dominated sets;

[0127] Step S305: Divide the individuals whose dominated sets only have the current individuals into one group, and mark it as the updated group, update the current group to the updated group, update the current individuals to the individuals in the updated group, and delete all the current individuals in the dominated sets;

[0128] Step S306: Loop step S305 until all individuals have completed grouping. After the loop ends, sequentially increase and set digital labels in the order of each group being marked as the current group, and mark them as group labels. The range of the group labels is [1, g], where g is the number of groups.

[0129] In the above step S302, the method for obtaining the judgment result includes:

[0130] Mark the two sets of ideal values corresponding to each set of individual collections as the first set and the second set respectively, and compare the ideal values of the same type in the first set and the second set; if all the ideal values in the first set are less than or equal to the corresponding ideal values in the second set, and there are ideal values in the second set that are greater than the corresponding ideal values in the first set, then the individuals corresponding to the second set dominate the individuals corresponding to the first set; if there are ideal values in the first set that are less than or equal to the corresponding ideal values in the second set, and there are ideal values in the first set that are greater than the corresponding ideal values in the second set, then the individuals corresponding to the first set and the individuals corresponding to the second set do not dominate each other.

[0131] In the above step S206, the expression for the sparsity is:

[0132] In the formula, is the sparsity of the i-th individual in the t-th iteration process, is the k-th ideal value in the ideal value set corresponding to the (i + 1)-th individual in the t-th iteration process, P k (max) is the k-th ideal value with the largest value in all ideal value sets, P k (min) is the k-th ideal value with the smallest value in all ideal value sets, k ∈ [1, l].

[0133] In the above step S207, the method for constructing the dynamic boundary is: compare the set labels corresponding to all individuals, mark the set label with the largest value as the largest label, and use the set label with the smallest value as the smallest label; construct the dynamic boundary according to the largest label and the smallest label, that is, the maximum value of the dynamic boundary is the largest label, and the minimum value is the smallest label;

[0134] The method for calculating the reverse individual corresponding to each individual is: randomly generate n values from the interval [0, 1] and mark them as random factors; add the largest label and the smallest label to obtain the reverse factor, and the reverse factor corresponds to the individual one by one; multiply each reverse factor by the random factor and then subtract the set label of the corresponding individual to obtain the reverse individual corresponding to each individual; among them, if the reverse individual corresponding to the individual is not within the dynamic boundary, then mark the corresponding individual as an out-of-bounds individual, and use a random function to randomly generate a value within the dynamic boundary and use it as the reverse individual of the out-of-bounds individual.

[0135] In the above step S208, the method for calculating the screening probability of each individual includes:

[0136] A preset ratio set, where the ratio set includes the ratio coefficients corresponding to each ideal value in the ideal value set, and the ratio set is preset by those skilled in the art according to the actual situation; according to the ratio set, obtain the ratio coefficient corresponding to each ideal value in the ideal value set, multiply each ideal value in the ideal value set corresponding to each individual by the corresponding ratio coefficient, and add them in sequence to obtain the ideal total value corresponding to each individual; add the ideal total values of each individual in sequence to obtain the population ideal value; divide the ideal total value of each individual by the population ideal value to obtain the screening probability of each individual.

[0137] In the above step S209, the method for screening n individuals from all individuals includes:

[0138] Sort the screening probabilities of each individual from large to small to generate a probability ranking table, and mark the individual corresponding to the screening probability ranked first in the probability ranking table as the starting individual; add each screening probability to all the screening probabilities before the corresponding screening probability to obtain the screening ratio corresponding to each screening probability; replace each screening probability in the probability ranking table with the corresponding screening ratio, and according to the forward order of the probability ranking table, sequentially take every two adjacent screening ratios as a group of ratio sets; compare the screening ratios in each group of ratio sets, mark the screening ratio with a larger value as the first ratio, and mark the screening ratio with a smaller value as the second ratio; construct the screening range corresponding to each group of ratio sets according to the first ratio and the second ratio in each group of ratio sets; where the first ratio is the maximum value of the screening range, the second ratio is the minimum value of the screening range, and the maximum value of the screening range is a closed interval and the minimum value is an open interval; each group of ratio sets corresponds to the individual corresponding to the second ratio, and the minimum value of the screening range corresponding to the starting individual is 0 and the maximum value is the corresponding screening probability; randomly generate a numerical value within the interval [0,1], and mark it as a random factor; screen out the individuals corresponding to the screening range where the random factor is located from all individuals, and mark them as preliminarily screened individuals.

[0139] Exemplarily, the selection probabilities of three individuals are 0.6, 0.3, and 0.1 in sequence. Since there is no value before 0.6, 0.6 remains 0.6 after replacement. Since there is 0.6 before 0.3, 0.3 is replaced with 0.3 + 0.6 = 0.9. Since there are 0.3 and 0.6 before 0.1, 0.1 is replaced with 0.1 + 0.3 + 0.6 = 1. Therefore, the screening ratios are 0.6, 0.9, and 1 in sequence. Among them, 0.6 and 0.9 form a set of ratios, and 0.9 and 1 form a set of ratios. Therefore, the screening range corresponding to the second individual is (0.6, 0.9], the screening range corresponding to the third individual is (0.9, 1], and the screening range corresponding to the first individual is (0, 0.6]. If the randomly generated random factor is 0.7, since 0.7 is within (0.6, 0.9], the second individual is screened out.

[0140] Sort each preliminarily screened individual in descending order according to the corresponding group label to generate an individual ranking list. Among them, if the group labels of multiple preliminarily screened individuals are the same, they are sorted in descending order according to the corresponding sparsity. According to the positive order of the individual ranking list, n individuals are screened out from all preliminarily screened individuals.

[0141] In the above step S211, the method for screening out the best individual from all individuals is: compare the ideal total values of all individuals, screen out the individual with the largest ideal total value from all individuals, and mark it as the best individual.

[0142] In this embodiment, by receiving the design condition data of the user terminal, combining advanced fuzzy logic algorithms and deep learning technologies, the intelligent design of the appearance contour of the retaining wall is realized. At the same time, a multi-objective optimization algorithm is adopted to realize the optimization of the steel bar configuration inside the retaining wall, improving the efficiency, accuracy, and intelligent level of the 3D design of the retaining wall. Making full use of the advantages of cloud computing, remote online design is realized, and intelligent decision-making modules such as the selection of structural types and layout forms, dimension optimization, and steel bar configuration optimization are integrated, which can automatically generate the optimal design scheme that meets the user's needs. At the same time, 3D model establishment is adopted to provide an intuitive design effect display for the user. Making the design process of the retaining wall more flexible and responsive, ensuring the safety and stability of the retaining wall, so as to adapt to the complex and changeable construction environment.

[0143] Embodiment 2

[0144] Please refer to Figure 2 As shown, the parts not described in detail in this embodiment can be seen in the description of Embodiment 1. A 3D design decision support method for retaining walls based on cloud computing is provided and applied to the cloud platform. The method includes:

[0145] Receiving the design condition data sent by the user terminal;

[0146] According to the design condition data, select the structural type of the retaining wall, and based on the structural type, determine the layout form of the retaining wall;

[0147] Integrate the design condition data, structural type and layout form, and design a set of retaining wall dimension data, where a is an integer greater than 1; construct corresponding 3D models for each set of retaining wall data respectively, and conduct structural analysis. According to the structural analysis results, screen out b sets of retaining wall dimension data and mark them as excellent dimension data, where 1 < b < a;

[0148] Receive the design dimension data sent by the client, match the design dimension data with the excellent dimension data, screen out the excellent dimension data with the highest matching degree, and mark it as the best dimension data;

[0149] Based on the 3D model corresponding to the best dimension data, use the multi-objective optimization algorithm to automatically generate the optimal steel bar configuration in the 3D model, and send the 3D model to the corresponding client.

[0150] Embodiment 3

[0151] This application also provides an electronic device. The electronic device may include one or more processors and one or more memories. Among them, computer-readable code is stored in the memory, and when the computer-readable code is run by one or more processors, it can execute the above-mentioned 3D design decision support method for retaining walls based on cloud computing.

[0152] The method or system according to the embodiment of the present application can also be implemented by means of the architecture of the electronic device shown in the present application. The electronic device may include a bus, one or more CPUs, ROM, RAM, a communication port connected to the network, input / output, a hard disk, etc. The storage device in the electronic device, such as ROM or hard disk, can store the 3D design decision support method for retaining walls based on cloud computing provided by the present application. Further, the electronic device may also include a user interface. Of course, the architecture shown in the present application is only exemplary, and when implementing different devices, one or more components shown in the electronic device of the present application can be omitted according to actual needs.

[0153] Embodiment 4

[0154] One embodiment of the present application discloses a computer-readable storage medium. Computer-readable instructions are stored on the computer-readable storage medium. When the computer-readable instructions are run by a processor, the 3D design decision support method for retaining walls based on cloud computing according to the embodiment of the present application described with reference to the above drawings can be executed. The storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory, etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0155] In addition, according to the embodiments of the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the present application provides a non-transitory machine-readable storage medium storing machine-readable instructions that can be run by a processor to execute instructions corresponding to the method steps provided by the present application, such as: a three-dimensional design decision support method for retaining walls based on cloud computing. When the computer program is executed by a central processing unit (CPU), the above functions defined in the method of the present application are executed.

[0156] The above is only the specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

[0157] Finally: The above is only the preferred embodiment of the present invention and is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A three-dimensional design decision support method for retaining walls based on cloud computing, characterized in that Applied to the cloud platform, including: Receiving the design condition data sent by the user terminal; According to the design condition data, selecting the structural type of the retaining wall, and determining the layout form of the retaining wall based on the structural type; Integrating the design condition data, structural type and layout form, designing a set of retaining wall dimension data, where a is an integer greater than 1; constructing corresponding 3D models for each set of retaining wall data respectively, and performing structural analysis, screening out b sets of retaining wall dimension data according to the structural analysis results, and marking them as excellent dimension data, 1 < b < a; Receiving the design dimension data sent by the user terminal, matching the design dimension data with the excellent dimension data, screening out the excellent dimension data with the highest matching degree, and marking it as the best dimension data; Based on the 3D model corresponding to the best dimension data, using a multi-objective optimization algorithm to automatically generate the optimal steel bar configuration in the 3D model, and sending the 3D model to the corresponding user terminal.

2. The three-dimensional design decision support method for retaining walls based on cloud computing according to claim 1, characterized in that The design condition data includes material data and boundary data; the material data includes retaining wall material data and geotechnical material data; The steps of selecting the structural type of the retaining wall include: Step S101: Constructing multiple fuzzy sets for each data in the design condition data; Step S102: Converting each data in the design condition data into the membership degree of the corresponding fuzzy set through a fuzzification technique respectively; Step S103: Defining fuzzy rules; Step S104: Matching the fuzzified design condition data with the fuzzy rules, performing fuzzy inference, and obtaining the fuzzy inference result, where the fuzzy inference result is the membership degree corresponding to each structural type; Step S105: Comparing each membership degree in the fuzzy inference result, and selecting the structural type corresponding to the membership degree with the largest value as the structural type of the retaining wall; The method for determining the layout form of the retaining wall includes: Setting different digital labels for the text data in the design condition data and marking them as condition labels, where the text data is the data in the design condition data that is not a numerical value; setting different digital labels for different structural types and marking them as type labels; replacing all the text data in the design condition data with the corresponding condition labels, and marking the replaced design condition data as design replacement data; obtaining the corresponding type label according to the selected structural type of the retaining wall and marking it as the selected label; using the design replacement data and the selected label as analysis data, inputting the analysis data into the trained form determination model to obtain the corresponding form label, where the form label is the digital label corresponding to the layout form, and the digital labels corresponding to different layout forms are all different; obtaining the corresponding layout form according to the form label and determining it as the layout form of the retaining wall; the form determination model is a deep neural network model.

3. The three-dimensional design decision support method for retaining walls based on cloud computing according to claim 2, characterized in that The method for designing a set of retaining wall dimension data includes: Taking the analysis data and form tags as design data, inputting the design data into the trained range prediction model to predict the corresponding range tags. The range tags are digital tags corresponding to dimensional ranges, and the digital tags corresponding to different dimensional ranges are all different. The range prediction model is a deep neural network model. According to the predicted range tags, obtain the corresponding dimensional ranges, and the dimensional ranges include the ranges corresponding to each type of data in the retaining wall dimension data. Randomly select a value from each range within the dimensional range to construct a set of retaining wall dimension data, and a total of a sets of retaining wall dimension data are constructed, and the a sets of retaining wall dimension data are all different. Use a 3D modeling tool to construct corresponding 3D models for each set of retaining wall data respectively. The structural analysis includes structural stability calculation and internal force calculation. The method for screening out b sets of retaining wall dimension data according to the structural analysis results includes: Preset a weight set, and the weight set includes the weight coefficients corresponding to each type of data in the structural analysis results. Multiply each data in the a sets of retaining wall dimension data by the corresponding weight coefficient, and then add them in sequence to obtain the structural factor corresponding to each set of retaining wall dimension data. Sort each structural factor from large to small to generate a factor sorting table. Screen out the top b structural factors from the factor sorting table and mark them as excellent factors, and obtain the retaining wall dimension data corresponding to the excellent factors.

4. The three-dimensional design decision support method for retaining walls based on cloud computing according to claim 3, characterized in that The design dimension data is the retaining wall dimension data designed by the user. The method for matching the design dimension data with the excellent dimension data includes: Taking the design dimension data and the excellent dimension data as calculation data, taking the data of the same type in the calculation data as a set of data sets, and the data sets correspond one by one to the data in the retaining wall dimension data. Obtain the largest value in each set of data sets and mark it as the maximum value, obtain the smallest value in each set of data sets and mark it as the minimum value. Subtract the minimum value of each set of data sets from the maximum value to obtain the value range corresponding to each set of data sets. Subtract the minimum value of the corresponding data set from each data in the calculation data, and then divide by the data range of the corresponding data set to obtain the standard value corresponding to each data in the calculation data. Subtract the standard value corresponding to the corresponding data in the design dimension data from the standard value corresponding to each data in each set of excellent dimension data to obtain the data difference corresponding to each data in the excellent dimension data. Add up the data differences corresponding to each set of excellent dimension data in sequence to obtain the total data difference corresponding to each set of excellent dimension data, and take the reciprocal of the total data difference as the matching degree of the corresponding excellent dimension data.

5. The three-dimensional design decision support method for retaining walls based on cloud computing according to claim 4, characterized in that, The steps for automatically generating the optimal steel bar configuration in the 3D model include: Step S201: Construct d sets of configuration sets, set different digital tags for each set of configuration sets respectively, and mark them as set tags, and the range of the set tags is [1, d]. Step S202: Initialize the population. The population includes n individuals, and the n individuals all correspond one by one to the set tags. Initialize the iteration times t corresponding to the population to 0, where 1 < n < d. Step S203: Define the iteration threshold T. Step S204: Calculate the ideal value set corresponding to each individual; Step S205: Divide all individuals into groups; Step S206: Calculate the sparsity corresponding to each individual; Step S207: Construct a dynamic boundary, calculate the reverse individual corresponding to each individual, and regard the reverse individual as an individual; Step S208: Calculate the screening probability of each individual; Step S209: Select n individuals from all individuals; Step S210: Compare the iteration number with the iteration threshold. If t≥T, go to Step S211. If t<T, let t = t + 1 and return to Step S204; Step S211: Select the best individual from all individuals, and obtain the set label corresponding to the best individual; according to the configuration set corresponding to the obtained set label, automatically generate the optimal steel bar configuration in the three-dimensional model.

6. The three-dimensional design decision support method for retaining walls based on cloud computing according to claim 5, characterized in that In the said Step S201, the method for constructing d groups of configuration sets is as follows: Obtain the steel bar configuration range, where the steel bar configuration range includes the range values corresponding to each type of data in the steel bar configuration data; randomly select a value from each range in the steel bar configuration range to construct a group of configuration sets, and a total of d groups of configuration sets are constructed. The d groups of configuration sets are all different, and d>1; In the said Step S202, the method for generating n individuals in the initialized population is as follows: Subtract the minimum value from the maximum value in the range corresponding to the set label to obtain the set interval; Randomly generate n values from the interval [0,1] and mark them as random coefficients. The random coefficients correspond to the individuals one by one; multiply the set interval by each random coefficient respectively, and add the minimum value of the set label to generate n individuals; In the said Step S204, the method for calculating the ideal value set corresponding to each individual is as follows: Obtain the configuration set corresponding to the set label corresponding to each individual, and regard the design data, the best dimension data, and a group of configuration sets as a group of analysis data; input each group of analysis data into the trained performance analysis model respectively to predict the corresponding performance parameters; where, there are l performance prediction models in the performance analysis model, l is the number of types of parameters in the performance parameters, the performance prediction models in the performance analysis model correspond to the parameters in the performance parameters one by one, and each performance prediction model is a deep neural network model; regard the performance parameters corresponding to each group of analysis data as the ideal value set corresponding to the corresponding individual; where, each parameter in the performance parameters is the ideal value in the corresponding ideal value set.

7. The method for three-dimensional design decision support of retaining walls based on cloud computing according to claim 6, characterized in that In the said Step S205, the steps for dividing all individuals into groups include: Step S301: Regard every two individuals as a group of individual sets; Step S302: According to the ideal value set corresponding to each individual, sequentially judge whether there is an individual dominating another individual in each group of individual sets to obtain the judgment result; Step S303: According to the judgment result, obtain the dominated set corresponding to each individual; Step S304: Divide the individuals whose dominated sets do not contain individuals into a group, and mark it as the current group. Mark the individuals in the current group as the current individuals, and delete all the current individuals in the dominated sets; Step S305: Divide all individuals that only have the current individual in the dominated set into one group, mark it as the updated group, update the current group to the updated group, update the current individual to the individual in the updated group, and delete all current individuals in the dominated set; Step S306: Loop step S305 until all individuals have completed group division. When the loop ends, set digital labels incrementally from the earliest to the latest according to the order in which each group is marked as the current group, and mark them as group labels. The range of the group labels is [1, g], where g is the number of groups; In the said step S302, the method for obtaining the judgment result includes: Mark the two ideal value sets corresponding to each set of individual sets as the first set and the second set respectively, and compare the ideal values of the same type in the first set and the second set; if all the ideal values in the first set are less than or equal to the corresponding ideal values in the second set, and there are ideal values in the second set that are greater than the corresponding ideal values in the first set, then the individuals corresponding to the second set dominate the individuals corresponding to the first set; if there are ideal values in the first set that are less than or equal to the corresponding ideal values in the second set, and there are ideal values in the first set that are greater than the corresponding ideal values in the second set, then the individuals corresponding to the first set and the individuals corresponding to the second set do not dominate each other.

8. The three-dimensional design decision support method for retaining walls based on cloud computing according to claim 7, characterized in that, In the said step S206, the expression of the sparsity is: In the formula, is the sparsity of the \(i\)-th individual in the \(t\)-th iteration process, is the \(k\)-th ideal value in the ideal value set corresponding to the \((i + 1)\)-th individual in the \(t\)-th iteration process, and \(P\) k (max) is the \(k\)-th ideal value with the largest value in all ideal value sets, and \(P\) k (min) is the \(k\)-th ideal value with the smallest value in all ideal value sets, where \(k\in[1, l]\); In the said step S207, the method for constructing the dynamic boundary is: compare the set labels corresponding to all individuals, mark the set label with the largest value as the maximum label, and use the set label with the smallest value as the minimum label; construct the dynamic boundary according to the maximum label and the minimum label, that is, the maximum value of the dynamic boundary is the maximum label, and the minimum value is the minimum label; The method for calculating the reverse individual corresponding to each individual is: randomly generate n values from the interval [0, 1] and mark them as random factors; add the maximum label and the minimum label to obtain the reverse factor, and the reverse factor corresponds to the individual one by one; multiply each reverse factor by the random factor respectively, and then subtract the set label of the corresponding individual to obtain the reverse individual corresponding to each individual; among them, if the reverse individual corresponding to the individual is not within the dynamic boundary, mark the corresponding individual as the transcended individual, and randomly generate a value within the dynamic boundary using a random function and use it as the reverse individual of the transcended individual; In the said step S208, the method for calculating the screening probability of each individual includes: Preset a ratio set, and the ratio set includes the ratio coefficient corresponding to each ideal value in the ideal value set; according to the ratio set, obtain the ratio coefficient corresponding to each ideal value in the ideal value set, multiply each ideal value in the ideal value set corresponding to each individual by the corresponding ratio coefficient, and add them up in turn to obtain the total ideal value corresponding to each individual; add up the total ideal values of each individual in turn to obtain the population ideal value; divide the total ideal value of each individual by the population ideal value to obtain the screening probability of each individual.

9. The method for three-dimensional design decision support of retaining walls based on cloud computing according to claim 8, wherein In the said step S209, the method for screening n individuals from all individuals includes: Sort the screening probabilities of each individual from largest to smallest to generate a probability ranking table, and mark the individual corresponding to the screening probability ranked first in the probability ranking table as the first individual; add each screening probability to all the screening probabilities before the corresponding screening probability to obtain the screening ratio corresponding to each screening probability; replace each screening probability in the probability ranking table with the corresponding screening ratio, and according to the forward order of the probability ranking table, take every two adjacent screening ratios as a set of ratio sets in turn; compare the screening ratios in each set of ratio sets, mark the screening ratio with a larger value as the first ratio, and mark the screening ratio with a smaller value as the second ratio; construct the screening range corresponding to each set of ratio sets according to the first ratio and the second ratio in each set of ratio sets; where the first ratio is the maximum value of the screening range, the second ratio is the minimum value of the screening range, and the maximum value of the screening range is a closed interval, and the minimum value is an open interval; each set of ratio sets corresponds to the individual corresponding to the second ratio, the minimum value of the screening range corresponding to the first individual is 0, and the maximum value is the corresponding screening probability; randomly generate numerical values and mark them as random factors; from all individuals, screen out the individuals corresponding to the screening range where the random factor is located and mark them as preliminarily screened individuals. Sort each preliminarily screened individual in descending order according to the corresponding group label to generate an individual ranking list; among them, if the group labels of multiple preliminarily screened individuals are the same, then sort them in descending order according to the corresponding sparsity; according to the forward order of the individual ranking list, screen out n individuals from all preliminarily screened individuals; In the step S211, the method for screening out the best individual from all individuals is: compare the ideal total values of all individuals, screen out the individual with the largest ideal total value from all individuals, and mark it as the best individual.

10. A three-dimensional design decision support system for retaining walls based on cloud computing, which implements the three-dimensional design decision support method for retaining walls based on cloud computing according to any one of claims 1-9, characterized in that, Applied in the cloud platform, including: A data receiving module for receiving the design condition data sent by the user terminal; A structure design module for selecting the structure type of the retaining wall according to the design condition data and determining the layout form of the retaining wall based on the structure type; A dimension design module for integrating the design condition data, the structure type and the layout form to design a set of retaining wall dimension data, where a is an integer greater than 1; construct a corresponding three-dimensional model for each set of retaining wall data and perform structural analysis, and screen out b sets of retaining wall dimension data according to the structural analysis results and mark them as excellent dimension data, where 1 < b < a; A dimension matching module for receiving the design dimension data sent by the user terminal, matching the design dimension data with the excellent dimension data, screening out the excellent dimension data with the highest matching degree, and marking it as the best dimension data; A steel bar configuration module for automatically generating the optimal steel bar configuration in the three-dimensional model based on the three-dimensional model corresponding to the best dimension data by using a multi-objective optimization algorithm and sending the three-dimensional model to the corresponding user terminal.

Images