Method and system for determining arrangement distance of steel diaphragm plates in UHPC box girder

Abstract

The invention discloses a method and a system for determining the arrangement spacing of steel diaphragm plates in a UHPC box girder. The method comprises the following steps: 1, parameterized analog simulation; 2, optimizing the data; 3, data evaluation and prediction; and 4, outputting data. A corresponding system comprises a parameterized analog simulation module, a data optimization module, a data evaluation and prediction module and a data interface module. The parameterized analog simulation module is used for executing the parameterized analog simulation step; the data optimization module is used for executing a data optimization step; the data evaluation and prediction module is used for executing a data evaluation and prediction step; and the data interface module is used for executing the step of data output. According to the method, on the premise of comprehensively considering the material characteristics and geometric characteristics of the UHPC box girder, the diaphragm plate spacing of the newly-built UHPC box girder is predicted and solved by using big data, the problem of applicability in the prior art is solved, the diaphragm plate spacing solving speed and precision are effectively improved, and the safety and stability of the UHPC box girder are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of computer-aided design, and particularly to a method and system for determining the layout spacing of steel diaphragms in UHPC box girders. Background Art

[0002] Ultra-high performance concrete (UHPC) has been widely used in bridge construction due to its material properties such as high strength, high toughness and good durability, making the bridge structure tend to be thinner and lighter. However, this advantage places more stringent requirements on the layout spacing of the internal diaphragms of UHPC box girders to ensure the stability of the structure.

[0003] The design concept of steel diaphragms for UHPC box girders has been proposed in existing materials, but there is a lack of a clear method for determining the diaphragm spacing. Currently, the specifications of countries such as Japan and the United States mainly provide calculation formula methods for determining the spacing for steel structure design, and such methods have limited applicability to UHPC box girders.

[0004] Due to the significant differences in the characteristics of UHPC and traditional steel structures, existing specifications only have a reference role. In addition, existing calculation methods consider fewer factors and have multiple assumptions, resulting in deviations in universality among different structural types, thus affecting the reliability and safety of projects.

[0005] Therefore, how to effectively solve the problem of the layout of steel diaphragms in UHPC box girders has become an urgent technical problem for those skilled in the art. Summary of the Invention

[0006] In view of the above-mentioned defects of the prior art, the present invention provides a method and system for determining the layout spacing of steel diaphragms in UHPC box girders, and the achieved purpose is to be able to apply to various different environments on the premise of comprehensively considering the material characteristics and structural characteristics of UHPC, and to solve the problem of the layout of steel diaphragms in UHPC box girders with higher accuracy.

[0007] To achieve the above purpose, the present invention discloses a method for determining the layout spacing of steel diaphragms in UHPC box girders, including the following steps:

[0008] Step 1, parametric simulation;

[0009] Step 2, data optimization;

[0010] Step 3, data evaluation and prediction;

[0011] Step 4, data output.

[0012] Preferably, step 1 is specifically as follows:

[0013] Step 1.1: Collect the bridge structure parameters of historical bridge structures similar to the bridge structure corresponding to the layout spacing of the steel diaphragms in the UHPC box girder to be designed.

[0014] Step 1.2: Randomly extract data from the bridge structure parameters to form a set of random bridge structure model information.

[0015] The application of the design load for each set of structural model information in the set of random bridge structure model information is by default divided into the following two cases:

[0016] The first case is that the corresponding design load is a non-eccentric load, which is applied to the bridge structure model in the form of a surface load.

[0017] The second case is that the corresponding design load is an eccentric load, and the distortional load is obtained by the load decomposition method, and the distortional load is applied to the bridge structure model in the form of a surface load.

[0018] Step 1.3: Use a script written in the Python language and automatically generate the bridge structure model corresponding to each structural model information through the script interface in the finite element software ABAQUS, that is, the UHPC box girder structure with internal steel diaphragms, namely the bridge structure model.

[0019] Then import the bridge structure model into the simulation software.

[0020] Step 1.4: Use the simulation software to analyze all the structural models, and organize the corresponding each structural model information and the corresponding analysis results into a single data packet.

[0021] Collect all the single data packets to form an initial result sample set.

[0022] More preferably, the bridge structure parameters include the main beam span, the set of beam section information, the set of diaphragm section characteristics, the diaphragm spacing, the set of boundary conditions, and the design load.

[0023] All the elements in the main beam span, all the elements in the set of beam section information, all the elements in the set of diaphragm section characteristics, the diaphragm spacing, and all the elements in the set of boundary conditions are all range values.

[0024] More preferably, step 2 is specifically as follows:

[0025] Step 2.1: Call the corresponding analysis results in each single data packet, and extract and analyze the structural response results at the specified model area in each analysis result.

[0026] The structural response results at the specified model area include the warping stress σ under the action of the eccentric loadDW and the bending stress σ under non-eccentric load b ;

[0027] Step 2.2. Screen the structural response results at the specified model area according to the following rules. The conditions are:

[0028]

[0029] Eliminate all the single data packets in the initial result sample set that do not meet the above conditions to form a complete database Data.

[0030] More preferably, in step 3, use the main girder span in the database Data, all elements in the beam section information set, all elements in the diaphragm section property set, the diaphragm spacing, and all elements in the boundary condition set as learning features, and use the diaphragm spacing as the prediction feature. Through training and learning of a large number of data samples, establish a mapping relationship between the learning features and the prediction features.

[0031] More preferably, the training and learning of the large number of data samples in step 3 is the random forest algorithm, and the expression is as follows:

[0032]

[0033] where y is the prediction probability of each category given by the random forest; t i represents the i-th decision tree; t i ·prob(x) represents the prediction of x by the i-th decision tree, and the output is the prediction probability of each category; k represents the upper limit of the decision tree, that is, the scale number of the forest;

[0034] The training process is as follows:

[0035] Step 3.1. Random sample collection, specifically: Randomly draw multiple sub-sample sets with replacement from the optimized database Data by the Bagging sampling method, and each sub-sample set is used to train the decision tree;

[0036] Step 3.2. Feature sampling, specifically: In the training of each decision tree, randomly select some features as splitting nodes for splitting selection;

[0037] The randomly selected some features include the main girder span, all elements in the beam section information set, all elements in the diaphragm section property set, the diaphragm spacing, and / or all elements in the boundary condition set;

[0038] Step 3.3. Decision tree training, specifically: Based on the training samples, train the prediction model and analyze the influence weight evaluation of the data features;

[0039] Among them, each of the decision trees grows as much as possible to the maximum extent without pruning.

[0040] In step 3.4, the parameters of the bridge structure corresponding to the layout spacing of the steel diaphragm in the to-be-designed UHPC box girder are used as to-be-predicted samples and input into the trained prediction model. Each decision tree obtains a predicted feature value based on the feature mapping of the to-be-predicted samples. Finally, the average value of the predicted feature values of multiple decision trees is taken as the layout spacing of the steel diaphragm in the to-be-designed UHPC box girder.

[0041] The present invention also provides a system for determining the layout spacing of the steel diaphragm in the UHPC box girder, which is characterized in that it is used to run the method for determining the layout spacing of the steel diaphragm in the UHPC box girder as described in any one of claims 1 to 6, and specifically includes: a parametric simulation module, a data optimization module, a data evaluation and prediction module, and a data interface module.

[0042] The parametric simulation module is used to perform the steps of parametric simulation and transmit the result data packet to the data optimization module.

[0043] The data optimization module is used to perform the steps of data optimization and transmit the database Data as the optimized training samples to the data evaluation and prediction module.

[0044] The data evaluation and prediction module is used to perform the steps of data evaluation and prediction and transmit the prediction result to the interface module.

[0045] The data interface module is used to perform the steps of data output and transmit the parameters of the bridge structure corresponding to the layout spacing of the steel diaphragm in the to-be-designed UHPC box girder input by the operator as to-be-predicted samples to the data evaluation and prediction module.

[0046] The beneficial effects of the present invention:

[0047] On the premise of comprehensively considering the material properties and geometric properties of the UHPC box girder, the present invention uses the big data idea to predict and solve the diaphragm spacing of the newly built UHPC box girder, solves the applicability problem of the existing technology, effectively improves the solving speed and accuracy of the diaphragm spacing, and ensures the safety and stability of the UHPC box girder.

[0048] The following will further illustrate the concept, specific structure and technical effects generated by the present invention with reference to the drawings, so as to fully understand the purpose, features and effects of the present invention. Description of the Drawings

[0049] Figure 1 Shows a flowchart of an embodiment of the present invention.

[0050] Figure 2 Shows the flowchart of parametric simulation in an embodiment of the present invention.

[0051] Figure 3 Shows the flowchart of data optimization in an embodiment of the present invention.

[0052] Figure 4 Shows the flowchart of prediction model training in an embodiment of the present invention.

[0053] Figure 5 Shows the flowchart of using the prediction model to process the layout spacing of steel diaphragms in the UHPC box girder to be designed in an embodiment of the present invention. Detailed implementation manners

[0054] Embodiment

[0055] As Figure 1 shown, the method for determining the layout spacing of steel diaphragms in the UHPC box girder includes the following steps:

[0056] Step 1, Parametric simulation;

[0057] Step 2, Data optimization;

[0058] Step 3, Data evaluation and prediction;

[0059] Step 4, Data output.

[0060] In some embodiments, Step 1 is specifically as follows:

[0061] Step 1.1, Collect the historical bridge structure parameters similar to the bridge structure corresponding to the layout spacing of the steel diaphragms in the UHPC box girder to be designed;

[0062] Step 1.2, Randomly extract data from the bridge structure parameters to form a set of random bridge structure model information;

[0063] The application of the design load for each set of structural model information in the set of random bridge structure model information is by default divided into the following two cases:

[0064] The first case is that the corresponding design load is a non-eccentric load and is applied to the bridge structure model in the form of a surface load;

[0065] The second case is that the corresponding design load is an eccentric load, and the distortion load is obtained by the load decomposition method and is applied to the bridge structure model in the form of a surface load;

[0066] Step 1.3: Write a script based on the Python language and automatically generate the bridge structure model corresponding to each structural model information through the script interface in the finite element software ABAQUS, that is, the UHPC box girder structure with internal steel diaphragms, namely the bridge structure model;

[0067] Then import the bridge structure model into the simulation software;

[0068] Step 1.4: Use the simulation software to analyze all structural models, and organize the corresponding each structural model information and the corresponding analysis results into a single data packet;

[0069] Collect all single data packets to form an initial result sample set.

[0070] In practical applications, the above-mentioned analysis of all structural models using the simulation software is implemented using the ABAQUS simulation software. Usually, a script needs to be written based on the Python language, and then the above functions can be achieved by using the script interface of the ABAQUS simulation software.

[0071] In some embodiments, the bridge structure parameters include the main girder span, the set of beam section information, the set of diaphragm section characteristics, the diaphragm spacing, the set of boundary conditions, and the design load;

[0072] All elements in the main girder span, the set of beam section information, all elements in the set of diaphragm section characteristics, the diaphragm spacing, and all elements in the set of boundary conditions are all range values.

[0073] In practical applications, all elements in the main girder span, the set of beam section information, all elements in the set of diaphragm section characteristics, the diaphragm spacing, and all elements in the set of boundary conditions being range values can make it more convenient to extract sample information later.

[0074] As Figure 2 shown, the main girder span is L, the set of beam section information is S(w, h, t1, t2…), the set of diaphragm section characteristics is G(b1, b2, b3…), the diaphragm spacing is d, the set of boundary conditions is B, and the design load is P(p 偏心 , p 非偏心 ), then the set of multiple groups of structural model information included in the random bridge structure model information set is (n1(L1, S1, HG1, B1, P), n2(L2, S2, HG2, B2, P), n3(L3, S3, HG3, B3, P)……);

[0075] The structural model corresponding to the structural model information generated through the secondary development function of the finite element simulation software is X(L,S,G,d,B,P). Then, all structural models are analyzed through the simulation software, and the corresponding structural model information and corresponding analysis results of each one are sorted into a single data packet Pack, and all single data packets Pack are gathered to form an initial result sample set Train.

[0076] As Figure 3 shown, in some embodiments, step 2 is specifically as follows:

[0077] Step 2.1: Call the corresponding analysis results in each single data packet, and extract and analyze the structural response results at the specified model area in each analysis result;

[0078] The structural response results at the specified model area include the warping stress σ DW under eccentric load and the bending stress σ b under non-eccentric load;

[0079] Step 2.2: Screen the structural response results at the specified model area according to the following rules, and the condition is:

[0080]

[0081] Eliminate all single data packets in the initial result sample set that do not meet the above conditions to form a complete database Data.

[0082] In some embodiments, step 3 uses the main girder span, all elements in the beam section information set, all elements in the diaphragm section property set, the diaphragm spacing, and all elements in the boundary condition set in the database Data as learning features, and uses the diaphragm spacing as the prediction feature. Through the training and learning of a large number of data samples, a mapping relationship between the learning features and the prediction features is established.

[0083] In some embodiments, the training and learning of the large number of data samples in step 3 is the random forest algorithm, and the expression is as follows:

[0084]

[0085] Among them, y is the prediction probability of each category given by the random forest; t i represents the i-th decision tree; t i ·prob(x) represents the prediction of x by the i-th decision tree, and the output is the prediction probability of each category; k represents the upper limit of the decision tree, that is, the number of forest scales;

[0086] As Figure 4 shown, the training process is as follows:

[0087] Step 3.1: Random sample collection, specifically: Randomly and with replacement draw multiple sub-sample sets from the optimized database Data by the Bagging sampling method, and each sub-sample set is used to train a decision tree;

[0088] Step 3.2: Feature sampling, specifically: During the training of each decision tree, randomly select some features as splitting nodes for splitting selection;

[0089] Randomly selecting some features includes all elements in the main girder span, the beam section information set, all elements in the diaphragm section property set, the diaphragm spacing, and / or all elements in the boundary condition set;

[0090] Step 3.3: Decision tree training, specifically: Based on the training samples, train the prediction model and analyze the influence weight evaluation of data features;

[0091] Among them, each decision tree grows as much as possible to the maximum extent without pruning;

[0092] As Figure 5 shown, after completing Step 3.4: Take the parameters of the bridge structure corresponding to the layout spacing of the internal steel diaphragms of the UHPC box girder to be designed as the samples to be predicted and input them into the trained prediction model. Each decision tree obtains a predicted feature value based on the feature mapping of the samples to be predicted. Finally, take the average value of the predicted feature values of multiple decision trees as the layout spacing of the internal steel diaphragms of the UHPC box girder to be designed.

[0093] The present invention also provides a system for determining the layout spacing of internal steel diaphragms in UHPC box girders, which is used to run the above method for determining the layout spacing of internal steel diaphragms in UHPC box girders, and specifically includes: a parametric simulation module, a data optimization module, a data evaluation and prediction module, and a data interface module;

[0094] The parametric simulation module is used to execute the steps of parametric simulation and transmit the result data packet to the data optimization module;

[0095] The data optimization module is used to execute the steps of data optimization and transmit the database Data as the optimized training samples to the data evaluation and prediction module;

[0096] The data evaluation and prediction module is used to execute the steps of data evaluation and prediction and transmit the prediction result to the interface module;

[0097] The data interface module is used to execute the steps of data output and transmit the parameters of the bridge structure corresponding to the layout spacing of the internal steel diaphragms of the UHPC box girder to be designed input by the operator as the samples to be predicted to the data evaluation and prediction module.

[0098] In practical applications, the present invention is composed of four major modules: parametric simulation, result data optimization, data evaluation and prediction, and data interface. The parametric simulation module constructs a large number of initial data samples for the data optimization module, and then the result data optimization module screens the initial data samples to provide trainable samples as the prediction basis for practical problems. The data evaluation and prediction module further trains the data training samples provided by the data optimization module to construct a prediction model, so as to predict the samples to be predicted input by the data interface module and return the results.

[0099] The preferred specific embodiments of the present invention have been described in detail above. It should be understood that those of ordinary skill in the art can make many modifications and variations according to the concept of the present invention without creative labor. Therefore, all technical solutions that can be obtained by those skilled in the art in the technical field of the present invention based on the concept of the present invention through logical analysis, reasoning or limited experiments on the basis of the prior art should fall within the protection scope determined by the claims.

Claims

1. Method for determining the layout spacing of steel diaphragms in UHPC box girders; characterized in that, It includes the following steps: Step 1, parametric simulation; Step 2, data optimization; Step 3, data evaluation and prediction; Step 4, data output.

2. The method for determining the layout spacing of steel diaphragms in a UHPC box girder according to claim 1, wherein Step 1 is specifically as follows: Step 1.1, collect the bridge structure parameters of historical bridges similar to the bridge structure corresponding to the layout spacing of the steel diaphragms in the UHPC box girder to be designed; Step 1.2, randomly extract data from the bridge structure parameters to form a set of random bridge structure model information; The application of the design load for each set of structural model information in the set of random bridge structure model information is by default divided into the following two cases: The first case is that the corresponding design load is a non-eccentric load, which is applied to the bridge structure model in the form of a surface load; The second case is that the corresponding design load is an eccentric load, and the distorted load is obtained by the load decomposition method and applied to the bridge structure model in the form of a surface load; Step 1.3, use a script written in the Python language and automatically generate the bridge structure model corresponding to each structural model information through the script interface in the finite element software ABAQUS, that is, the UHPC box girder structure with internal steel diaphragms, namely the bridge structure model; Then import the bridge structure model into the simulation software; Step 1.4, use the simulation software to analyze all the structural models, and organize the corresponding each structural model information and the corresponding analysis results into a single data packet; Collect all the single data packets to form an initial result sample set.

3. The method for determining the layout spacing of steel diaphragms in a UHPC box girder according to claim 2, characterized in that, The bridge structure parameters include the main beam span, the set of beam section information, the set of diaphragm section characteristics, the diaphragm spacing, the set of boundary conditions, and the design load; All elements in the main beam span, all elements in the set of beam section information, all elements in the set of diaphragm section characteristics, the diaphragm spacing, and all elements in the set of boundary conditions are all range values.

4. The method for determining the layout spacing of steel diaphragms in a UHPC box girder according to claim 3, wherein Step 2 is specifically as follows: Step 2.1, call the corresponding analysis results in each single data packet, and extract and analyze the structural response results at the specified model area in each analysis result; The structural response results at the specified model area include the warping stress σ under eccentric load DW and the bending stress σ under non-eccentric load b ; Step 2.2, screen the structural response results at the specified model area according to the following rules, and the conditions are: Delete all single data packets in the initial result sample set that do not meet the above conditions to form a complete database Data.

5. The method for determining the layout spacing of steel diaphragms in a UHPC box girder according to claim 4, characterized in that, In Step 3, use the main beam span, all elements in the set of beam section information, all elements in the set of diaphragm section characteristics, the diaphragm spacing, and all elements in the set of boundary conditions in the database Data as learning features, and use the diaphragm spacing as the prediction feature. Through the training and learning of a large number of data samples, establish the mapping relationship between the learning features and the prediction features.

6. The method for determining the layout spacing of steel diaphragms in a UHPC box girder according to claim 5, characterized in that, The training and learning of the large number of data samples in Step 3 is the random forest algorithm, and the expression is as follows: where y is the predicted probability of each category given by the random forest; t i represents the i-th decision tree; t i ·prob(x) represents the prediction of the i-th decision tree for x, and the output is the predicted probability of each category; k represents the upper limit of the decision trees, that is, the number of trees in the forest; The training process is as follows: Step 3.1, random sample collection, specifically: randomly and with replacement extract multiple sub-sample sets from the optimized database Data by the Bagging sampling method, and each sub-sample set is used to train a decision tree; Step 3.2, Feature Sampling, specifically: during the training of each of the decision trees, randomly select some features as splitting nodes for splitting selection; The randomly selected some features include the main girder span, all elements in the beam cross-section information set, all elements in the diaphragm cross-section property set, the diaphragm spacing, and / or all elements in the boundary condition set; Step 3.3, Decision Tree Training, specifically: based on the training samples, train the prediction model and analyze the influence weight evaluation of data features; Among them, each of the decision trees grows as much as possible to the maximum extent without pruning; After completing Step 3.4, use the parameters of the bridge structure corresponding to the arrangement spacing of the steel diaphragms in the to-be-designed UHPC box girder as the to-be-predicted samples and input them into the trained prediction model. Each of the decision trees obtains a predicted feature value based on the feature mapping of the to-be-predicted samples. Finally, take the average value of the predicted feature values of multiple decision trees as the arrangement spacing of the steel diaphragms in the to-be-designed UHPC box girder.

7. A system for determining the spacing of steel diaphragms within a UHPC box girder, characterized in that, The method for determining the arrangement spacing of the steel diaphragms in the UHPC box girder as claimed in any one of claims 1 to 6 when running, specifically includes: a parametric simulation module, a data optimization module, a data evaluation and prediction module, and a data interface module; The parametric simulation module is used to execute the steps of parametric simulation and transmit the result data packet to the data optimization module; The data optimization module is used to execute the steps of data optimization and transmit the database Data as the optimized training samples to the data evaluation and prediction module; The data evaluation and prediction module is used to execute the steps of data evaluation and prediction and transmit the prediction results to the interface module; The data interface module is used to execute the steps of data output and transmit the parameters of the bridge structure corresponding to the arrangement spacing of the steel diaphragms in the to-be-designed UHPC box girder input by the operator as the to-be-predicted samples to the data evaluation and prediction module.

Images