Traffic engineering structure design scheme recommendation method fusing physical information and neural network

Abstract

The application relates to the technical field of traffic engineering investigation and design scheme recommendation, and discloses a traffic engineering structure design scheme recommendation method fusing physical information and a neural network, which comprises the following steps: acquiring characteristic parameters of a design scheme, and determining an input parameter vector group according to the characteristic parameters, wherein the input parameter vector group is obtained by vectorization and perturbation according to the characteristic parameters; inputting the input parameter vector group into a scheme recommendation model to obtain a design scheme group; acquiring physical information and evaluation indexes of the design scheme, performing scheme checking calculation and index evaluation on each design scheme in the design scheme group according to the physical information and the evaluation indexes of the design scheme, and obtaining evaluation scores of the design schemes; and determining a recommended scheme according to the evaluation scores of the design schemes. The application can accurately evaluate traffic engineering investigation and design schemes, and effectively improves the accuracy of the recommended scheme.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of traffic engineering survey and design scheme recommendation, and in particular to a traffic engineering structure design scheme recommendation method fusing physical information and a neural network. BACKGROUND

[0002] With the development of artificial intelligence technology, scientists have proposed various neural networks, and the more classic ones are LeNet, AlexNet, VGG, ResNet and GoogLeNet, etc. With the penetration of artificial intelligence into various industries, more and more industries have realized the digital transformation of the industry through intelligent technology empowerment. At present, the biggest difficulty is to solve the actual industry business through intelligent technology, and the existing neural network cannot completely solve the actual problems in the industry landing process.

[0003] This is because the solution in the field of business is not the only optimal solution, and the pros and cons of each solution often need to be judged according to the specific scene and business target. For example, in the traffic engineering survey and design industry, the specific problem is: (1) Non-unique optimal solution: The problems in the field of traffic engineering survey and design are often diverse and unique, resulting in different solutions for the same project due to design concepts, engineering requirements, and other factors. The design scheme generated by artificial intelligence algorithm may only be one of the many possible schemes, and it is difficult to guarantee that it is the optimal solution. For example, an artificial intelligence algorithm may generate a design scheme with low economic cost in a project, but due to differences in environmental factors and engineering requirements, the same scheme may no longer be applicable in another project. (2) Limitations of traditional evaluation indicators: Traditional evaluation indicators such as cost, cycle, and quality can only evaluate design schemes from one or several dimensions, making it difficult to fully reflect the pros and cons of the scheme. In the traffic engineering survey and design industry, for example, a design scheme may perform well in terms of cost and cycle, but may cause quality problems due to the neglect of certain special engineering requirements. (3) Dependence on professional knowledge: Traffic engineering survey and design work is highly dependent on professional knowledge and experience, and the scheme generated by artificial intelligence algorithm needs to be evaluated by professional engineers. For example, an artificial intelligence algorithm may propose an innovative scheme for structural layout, but whether it can meet the structural safety and functional requirements needs to be judged by a structural engineer with rich experience. (4) Complexity of practical application: The actual application scenario of traffic engineering survey and design work is complex and varied, and the scheme generated by artificial intelligence algorithm needs to be verified in actual projects. For example, a design scheme that performs well in a simulated environment may face unexpected challenges in actual construction, such as construction condition restrictions, material supply problems, etc. (5) Innovation demand of evaluation method: According to the characteristics of the traffic engineering survey and design industry, new evaluation methods need to be explored, such as a hybrid evaluation system combining expert knowledge and artificial intelligence algorithm. For example, an expert system can be used to assist in evaluating the design scheme generated by artificial intelligence, and the expert system can integrate domain knowledge and algorithm-generated data to provide more comprehensive evaluation. (6) Importance of case library: Establishing a case library containing various traffic engineering survey and design cases can help artificial intelligence algorithms learn from and draw lessons from past excellent design schemes. For example, by analyzing successful cases in the case library, artificial intelligence algorithms can learn which design elements have achieved good results in different projects, and thus draw on these elements when generating new schemes.

[0004] In summary, the design scheme in the field of traffic engineering survey and design has characteristics such as complex scheme, non-unique optimal solution, and strong dependence on professionals, and the current intelligent algorithm evaluation indicators cannot adapt to the quality of the traffic engineering survey and design scheme generated by intelligent algorithms, and thus cannot evaluate the pros and cons of intelligent algorithms, resulting in poor accuracy of recommended schemes.

[0005] Therefore, how to accurately evaluate the traffic engineering structure design scheme and then effectively improve the accuracy of the recommended scheme is a problem to be solved at present.

[0006] The above content is only used to assist in understanding the technical solutions of the present application and does not represent the acknowledgement of the above content as prior art. SUMMARY

[0007] The main purpose of the present application is to provide a traffic engineering structure design scheme recommendation method combining physical information and neural network, aiming to solve the technical problem of how to accurately evaluate the traffic engineering survey and design scheme and then effectively improve the accuracy of the recommended scheme.

[0008] To achieve the above purpose, the present application provides a traffic engineering structure design scheme recommendation method combining physical information and neural network, which comprises:

[0009] Obtaining the characteristic parameters of the design scheme, and determining the input parameter vector group according to the characteristic parameters, wherein the input parameter vector group is obtained by vectorization and perturbation of the characteristic parameters;

[0010] Inputting the input parameter vector group into a scheme recommendation model to obtain a design scheme group, wherein the scheme recommendation model comprises an input layer, a fully connected public hidden layer and a plurality of regional fully connected output layers;

[0011] Obtaining the physical information and evaluation indexes of the design scheme, and performing scheme calculation and index evaluation on each design scheme in the design scheme group according to the physical information and evaluation indexes of the design scheme, to obtain the evaluation score of each design scheme;

[0012] Determining the recommended scheme according to the evaluation score of each design scheme.

[0013] In an embodiment, the obtaining of the characteristic parameters of the design scheme and the determination of the input parameter vector group according to the characteristic parameters comprises:

[0014] Obtaining the characteristic parameters of the design scheme, and vectorizing the characteristic parameters to obtain a characteristic parameter vector;

[0015] Perturbing a plurality of characteristic parameters in the characteristic parameter vector to obtain an input parameter vector group.

[0016] In an embodiment, before the inputting of the input parameter vector group into the scheme recommendation model to obtain the design scheme group, it further comprises:

[0017] Obtaining a sample library of scheme recommendation;

[0018] Pre-training a single-tailed neural network by samples in the sample library to obtain an initial recommendation model;

[0019] Training the initial recommendation model by data samples in the sample library in stages to obtain a scheme recommendation model.

[0020] In an embodiment, the pre-training of the single-tailed neural network by samples in the sample library to obtain an initial recommendation model comprises:

[0021] Pre-training a single-tailed neural network by samples in the sample library to obtain target weights;

[0022] Migrating to a multi-tailed neural network according to the target weights to obtain an initial recommendation model.

[0023] In an embodiment, the obtaining of the physical information and evaluation indexes of the design scheme and the scheme calculation and index evaluation of each design scheme in the design scheme group according to the physical information and evaluation indexes of the design scheme to obtain the evaluation score of each design scheme comprises:

[0024] Obtaining the physical information and evaluation indexes of the design scheme;

[0025] Constructing a finite element model according to the physical information of the design scheme and determining a target evaluation function according to the evaluation indexes of the design scheme;

[0026] Performing scheme calculation and index evaluation of each design scheme in the design scheme group according to the finite element model and the target evaluation function to obtain the evaluation score of each design scheme.

[0027] In an embodiment, the performing of scheme calculation and index evaluation of each design scheme in the design scheme group according to the finite element model and the target evaluation function to obtain the evaluation score of each design scheme comprises:

[0028] Performing scheme calculation of each design scheme in the design scheme group by the finite element model to obtain a calculation result set of each design scheme;

[0029] Performing index evaluation of the calculation result set of each design scheme by the target evaluation function to obtain the evaluation score of each design scheme.

[0030] In an embodiment, the determining of the recommended scheme according to the evaluation score of each design scheme comprises:

[0031] Ranking the design schemes according to the evaluation score of each design scheme from high to low to obtain a design scheme sequence;

[0032] The design scheme sequence with the ranking value being the preset value is taken as a recommended scheme.

[0033] In addition, to achieve the above object, the application further provides a traffic engineering structure design scheme recommendation device fusing physical information and a neural network, which comprises:

[0034] A determination module is configured to acquire feature parameters of design schemes and determine an input parameter vector group according to the feature parameters, wherein the input parameter vector group is obtained by vectorization and perturbation of the feature parameters;

[0035] An input module is configured to input the input parameter vector group into a scheme recommendation model to obtain a design scheme group, wherein the scheme recommendation model comprises an input layer, a fully connected public hidden layer and a plurality of regional fully connected output layers;

[0036] An evaluation module is configured to acquire physical information and evaluation indexes of the design schemes, perform scheme calculation and index evaluation on each design scheme in the design scheme group according to the physical information and the evaluation indexes of the design schemes, and obtain evaluation scores of the design schemes.

[0037] The determination module is further configured to determine a recommended scheme according to the evaluation scores of the design schemes.

[0038] In addition, to achieve the above object, the application further provides a traffic engineering structure design scheme recommendation device fusing physical information and a neural network, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the traffic engineering structure design scheme recommendation method fusing physical information and a neural network.

[0039] In addition, to achieve the above object, the application further provides a storage medium, which is a computer readable storage medium, and a computer program is stored in the storage medium, and the computer program is executed by a processor to implement the steps of the traffic engineering structure design scheme recommendation method fusing physical information and a neural network.

[0040] In addition, to achieve the above object, the application further provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement the steps of the traffic engineering structure design scheme recommendation method fusing physical information and a neural network.

[0041] The application provides a traffic engineering structure design scheme recommendation method fusing physical information and a neural network.

[0042] In conclusion, the input parameter vector group obtained by vectorization and perturbation of the characteristic parameters of the design scheme is input into the scheme recommendation model, the design scheme group output by the model is calculated and evaluated according to the physical information and evaluation indexes of the design scheme, and then the recommended scheme is determined according to the evaluation scores of the design schemes, so that the traffic engineering survey design scheme can be accurately evaluated, and the accuracy of the recommended scheme is effectively improved. BRIEF DESCRIPTION OF DRAWINGS

[0043] The accompanying drawings, which are incorporated into and form part of the specification, illustrate embodiments consistent with the present application and, together with the specification, serve to explain the principles of the application.

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the accompanying drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, those skilled in the art can obtain other drawings according to these drawings without any creative effort.

[0045] Figure 1 The flowchart provided by the traffic engineering structure design scheme recommendation method fusing physical information and a neural network of the first embodiment of the present application;

[0046] Figure 2 The high-speed railway bridge pile foundation layout schematic diagram provided by the traffic engineering structure design scheme recommendation method fusing physical information and a neural network of the first embodiment of the present application;

[0047] Figure 3A pile foundation recommended double-tailed neural network architecture diagram provided by the traffic engineering structure design scheme recommendation method embodiment one of the application fusing physical information and neural networks;

[0048] Figure 4 A pre-training + phased training comprehensive training method schematic diagram provided by the traffic engineering structure design scheme recommendation method embodiment one of the application fusing physical information and neural networks;

[0049] Figure 5 An interface relationship schematic diagram between the pile foundation intelligent selection algorithm and the whole bridge design subsystem provided by the traffic engineering structure design scheme recommendation method embodiment one of the application fusing physical information and neural networks;

[0050] Figure 6 A technical principle diagram provided by the traffic engineering structure design scheme recommendation method embodiment one of the application fusing physical information and neural networks;

[0051] Figure 7 A pile foundation intelligent selection pile foundation design flowchart based on a double-tailed neural network provided by the traffic engineering structure design scheme recommendation method embodiment one of the application fusing physical information and neural networks;

[0052] Figure 8 A flowchart schematic diagram provided by the traffic engineering structure design scheme recommendation method embodiment two of the application fusing physical information and neural networks;

[0053] Figure 9 A module structure schematic diagram of the traffic engineering structure design scheme recommendation device of the embodiment of the application fusing physical information and neural networks;

[0054] Figure 10 A device structure schematic diagram of the hardware running environment involved in the traffic engineering structure design scheme recommendation method embodiment of the application fusing physical information and neural networks.

[0055] The object implementation, functional features and advantages of the application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0056] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the application, and are not used to limit the application.

[0057] In order to better understand the technical solutions of the application, the following will be described in detail in combination with the drawings and specific embodiments of the specification.

[0058] The main solution of the embodiment of the application is: obtaining a characteristic parameter of a design scheme, and determining an input parameter vector group according to the characteristic parameter, wherein the input parameter vector group is obtained by vectorization and perturbation according to the characteristic parameter; inputting the input parameter vector group into a scheme recommendation model to obtain a design scheme group, wherein the scheme recommendation model comprises an input layer, a fully connected public hidden layer and a plurality of regional fully connected output layers; obtaining physical information and evaluation indexes of the design scheme, performing scheme calculation and index evaluation on each design scheme in the design scheme group according to the physical information and the evaluation indexes of the design scheme, to obtain an evaluation score of each design scheme; and determining a recommended scheme according to the evaluation scores of the design schemes.

[0059] The design scheme in the field of traffic engineering investigation and design has the characteristics of complex scheme, non-unique optimal solution and strong professional dependence. The current intelligent algorithm evaluation index cannot adapt to the quality of the traffic engineering investigation and design scheme generated by the intelligent algorithm, and further cannot evaluate the advantages and disadvantages of the intelligent algorithm, resulting in poor accuracy of the recommended scheme. Therefore, how to accurately evaluate the traffic engineering investigation and design scheme and further effectively improve the accuracy of the recommended scheme is a problem to be solved at present.

[0060] The application inputs the input parameter vector based on the characteristic parameter of the design scheme into the scheme recommendation model, evaluates a plurality of design schemes output by the model according to the evaluation indexes, and further determines a recommended scheme according to the evaluation scores of the design schemes, thereby overcoming the technical defects that the current intelligent algorithm evaluation index cannot adapt to the quality of the traffic engineering investigation and design scheme generated by the intelligent algorithm, and further cannot evaluate the advantages and disadvantages of the intelligent algorithm, resulting in poor accuracy of the recommended scheme. The application can accurately evaluate the traffic engineering investigation and design scheme, and further effectively improve the accuracy of the recommended scheme.

[0061] It should be noted that the execution subject of the embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, a traffic engineering structure design scheme recommendation device integrating physical information and neural networks, etc. The traffic engineering structure design scheme recommendation device integrating physical information and neural networks is taken as an example to describe the embodiment and the following embodiments.

[0062] Based on this, the embodiment of the application provides a traffic engineering structure design scheme recommendation method integrating physical information and neural networks, which refers to Figure 1 , Figure 1 The flowchart of the first embodiment of the traffic engineering structure design scheme recommendation method integrating physical information and neural networks of the application is shown in the figure.

[0063] In this embodiment, the traffic engineering structure design scheme recommendation method fusing physical information and neural network comprises steps S10-S40.

[0064] Step S10, obtain the characteristic parameters of the design scheme, and determine an input parameter vector group according to the characteristic parameters, wherein the input parameter vector group is obtained by vectorization and perturbation according to the characteristic parameters.

[0065] It should be noted that the design scheme can be a design scheme of a structure in the field of traffic engineering survey and design, for example, a bridge engineering group pile foundation scheme design, a tunnel engineering support scheme design, etc. This embodiment does not make specific limitations thereto, and the bridge engineering group pile foundation scheme design is taken as an example for illustration.

[0066] It can be understood that the characteristic parameters of the design scheme refer to a data set capable of comprehensively describing the characteristics of the design scheme, including but not limited to geological conditions, load requirements, environmental restrictions, material properties, construction conditions, etc. In the bridge engineering group pile foundation scheme design, the characteristic parameters can be main beam types, main span lengths, main beam lengths, pier heights, pier positions, topographic features, etc. This embodiment does not make specific limitations thereto. In the bridge engineering group pile foundation scheme design, the design scheme is a pile foundation scheme, which is formed by combination and arrangement of pile diameters and pile numbers to form different types of pile foundation configuration types, such as 8-1.0m and 10-1.25m, which respectively represent a pile foundation of 8 piles with a diameter of 1m and a group pile foundation of 10 piles with a diameter of 1.25m. There are more than 10 commonly used pile diameters, such as 1.0m, 1.25m, 1.5m, 1.8m, 2.0m, 2.2m, 2.5m, 3.0m, etc. There are more than 10 pile numbers for each diameter, such as 8, 9, 10, 11, 12, 15, 16, 18, etc. This embodiment does not make specific limitations thereto.

[0067] As shown in FIG. 1, Figure 2 Figure 2 is a schematic diagram of pile foundation arrangement of high-speed railway bridge, which includes 9-1.0m, 10-1.0m, 11-1.0m, and 12-1.0m pile foundation configuration types, wherein the 9-1.0m pile foundation configuration type is a 9-pile pile foundation configuration type arranged in a row-column mode with a diameter of 1m, the 10-1.0m pile foundation configuration type is a 10-pile pile foundation configuration type arranged in a plum blossom mode with a diameter of 1m, the 11-1.0m pile foundation configuration type is an 11-pile pile foundation configuration type arranged in a plum blossom mode with a diameter of 1m, and the 12-1.0m pile foundation configuration type is a 12-pile pile foundation configuration type arranged in a row-column mode with a diameter of 1m.

[0068] ​​It is worth noting that the input parameter vector is obtained by organizing the characteristic parameters of the design scheme into a vector in a numerical form and perturbing it, and is used as the input of the scheme recommendation model. In the scenario of bridge engineering group pile foundation scheme design, the input parameter vector can include but is not limited to the encoding value of the main beam type, the specific numerical value of the main span length, the specific numerical value of the main beam length, the specific numerical value of the pier height, the coordinate value of the pier position, and the quantitative index of the terrain characteristics. These numerical characteristic parameters can comprehensively and accurately reflect the characteristics of the design scheme, and provide a basis for subsequent scheme recommendation and evaluation through the scheme recommendation model.

[0069] In a feasible implementation, step S10 can include: obtaining characteristic parameters of a design scheme, and vectorizing the characteristic parameters to obtain a characteristic parameter vector; perturbing several characteristic parameters in the characteristic parameter vector to obtain an input parameter vector group.

[0070] It should be noted that the complex information in the design scheme can be converted into a numerical form recognizable by the model by vectorizing and perturbing the characteristic parameters, and a certain amount of change is introduced to explore the influence of different design parameters on the overall scheme performance, thereby obtaining an input parameter vector group of multiple schemes.

[0071] It can be understood that vectorization refers to representing the characteristic parameters in the form of a vector. In the bridge engineering group pile foundation scheme design, the characteristic parameter vector = [main beam type, main span length, main beam length, pier height, pier position, terrain characteristics].

[0072] It is worth noting that perturbation refers to introducing a small amount of change to the original characteristic parameter vector to simulate the influence of different design parameters on the scheme performance. This amount of change can be randomly generated or based on certain rules. The specific perturbation method can be: increasing / decreasing the quantized value of any one or more parameters in the characteristic parameter vector by a certain step, i.e., new parameter value = original parameter value ± Δ, then the generated input parameter vector = [main beam type, main span length, main beam length Δ1, pier height ± Δ2, pier position, terrain characteristics]. Through the perturbation process, multiple input parameter vectors corresponding to different schemes can be obtained, representing the possible situations of the design scheme under different parameter combinations. These vectors will be used as the input of the scheme recommendation model to evaluate the performance and advantages and disadvantages of different design schemes, thereby providing a scientific basis for decision-making.

[0073] Step S20 inputs the input parameter vector group into the scheme recommendation model to obtain a design scheme group, wherein the scheme recommendation model includes an input layer, a fully connected public hidden layer, and multiple regional fully connected output layers.

[0074] It should be noted that the scheme recommendation model, i.e. the intelligent scheme recommendation algorithm based on neural network or deep learning constructed according to the field business demand, includes an input layer, a fully connected public hidden layer and a plurality of regional fully connected output layers. The input layer is used to receive and process the input parameter vector after vectorization and perturbation. The fully connected public hidden layer is responsible for capturing the complex relationship and potential mode contained in the parameter vector, extracting the feature representation which has a decisive influence on the performance of the design scheme through multi-level nonlinear transformation. These feature representations not only cover the direct correlation between design parameters, but also imply deep information interwoven through bridge structural mechanics, environmental adaptability, cost effectiveness and other factors. The plurality of regional fully connected output layers correspond to different design parameters respectively. Each output layer is based on the feature representation extracted by the public hidden layer, and outputs the design parameters in the corresponding dimension through a specific weight matrix and an activation function.

[0075] It can be understood that in the design of the pile foundation scheme of the bridge engineering group, the scheme recommendation model is a pile foundation intelligent selection algorithm of double-tailed neural network. The network model is composed of an input layer, a fully connected public hidden layer, a pile foundation diameter decision tail and a pile foundation number decision tail two parallel regional fully connected softmax output layers. The model input layer is composed of characteristic parameters such as main beam type, main span length, main beam length, pier height, pier position and terrain characteristics. The model output layer is composed of two parallel output layers of pile foundation number and pile foundation diameter, as shown in Figure 3 Figure 3 is a double-tailed neural network architecture diagram for pile foundation recommendation. The sum of the loss functions of the double-tailed neural network is the total loss function. The specific formula is as follows:

[0076]

[0077] Wherein, L conf is the loss value, A and B represent the double-tailed part, and the label, i.e. the pile foundation diameter label and the pile foundation number label, 0 label represents the public hidden layer.

[0078] ​In a specific implementation, in the bridge engineering group pile foundation scheme design, in order to evaluate the quality of the pile foundation scheme recommended by the pile foundation intelligent selection algorithm, a plurality of input parameters are formed by perturbing the characteristic parameters such as the main beam type, the main span length, the main beam length, the pier height, the pier position and the terrain characteristics, which are input into the pile foundation intelligent selection algorithm module, so as to generate a plurality of design schemes of low, standard and high levels, i.e. a design scheme group. Specifically, the input parameter vector is sequentially input into the double-tailed neural network pile foundation intelligent selection algorithm to recommend a plurality of recommended schemes of low, standard and high levels. For example, according to physical common sense, the higher the pier, the larger the pile foundation directly required to bear, and the more the number of pile foundations, so the parameter value of the pier height dimension in the input parameter vector [{pier height: pier height value-△2}, {pier height: pier height value}, {pier height: pier height value+△2}] corresponds to the pier height value input into the low, standard and high level pile foundation scheme respectively.

[0079] In a feasible implementation, before step S20, further comprising: obtaining a sample library of scheme recommendation; pre-training the single-tailed neural network through the samples in the sample library to obtain an initial recommendation model; and training the initial recommendation model through the data samples in the sample library in stages to obtain a scheme recommendation model.

[0080] It should be noted that the sample library refers to a database composed of a plurality of actual design schemes and corresponding performance evaluation results. In the bridge engineering group pile foundation scheme design, the sample library of scheme recommendation, i.e. the pile foundation intelligent selection sample library, refers to a collection of bridge engineering group pile foundation design schemes and performance evaluation results collected through actual engineering cases, expert experience, numerical simulation or experimental data, etc. These samples contain rich design parameter information (such as main beam type, main span length, pier height, etc.) and corresponding performance evaluation indexes (such as bearing capacity, stability, economy, etc.), which are basic data resources for building and training the scheme recommendation model.

[0081] As shown in Table 1, Table 1 is an example of a pile foundation intelligent selection sample library, which includes sample labels and corresponding input data and sample labels. The input data includes the main beam type, the main span length, the main beam length, the pier height, the pier position and the terrain label, and the sample label includes the pile foundation type. For example, the input data of sample 1 includes the main beam type as a simply supported beam, the main span length as 24m, the main beam length as 24m, the pier height as 1, the pier position as side, and the terrain label as A, and the pile foundation type as 8-1.0m.

[0082] Table 1

[0083]

[0084] It can be understood that in the embodiment, the neural network model is trained by using a "pre-training + phased training" comprehensive training method. The overall technical principle is: first, train the public hidden layer according to the conventional full-connection single-tail neural network, then migrate the weight to the multi-tail neural network, and then use the phased training method. Pre-training refers to preliminary learning using a large amount of sample data in the sample library in the single-tail neural network framework, aiming to enable the network model to capture the basic features and preliminary rules in the input data. Phased training refers to expanding the model into a multi-tail neural network according to specific design requirements and network architecture on the basis of pre-training, and performing fine-grained training in stages, which can improve the understanding and response ability of the model to complex design problems, and ensure that the recommended design scheme not only meets the actual requirements, but also has high performance and economic benefits.

[0085] In a feasible implementation manner, the pre-training of the single-tail neural network by the samples in the sample library to obtain the initial recommendation model comprises: pre-training of the single-tail neural network by the samples in the sample library to obtain target weights; and migrating to the multi-tail neural network according to the target weights to obtain the initial recommendation model.

[0086] It should be noted that pre-training is to enable the single-tail neural network to learn the basic mapping relationship between the design parameters and the pile type by repeated iterative learning. In this process, the single-tail neural network adjusts the weight and bias to minimize the prediction error as the goal, and gradually optimizes its internal parameters. When the single-tail neural network reaches a certain performance index on the sample library, the weight obtained by training is the target weight.

[0087] It can be understood that the target weight is migrated to the double-tail neural network as the starting point of the initial recommendation model, the weight migration accelerates the training process of the double-tail neural network, and ensures that the model can inherit the basic features and rules learned by the single-tail neural network, so as to adapt to complex design problems more quickly under the new network architecture.

[0088] In a specific implementation, in the design of the pile foundation scheme of a bridge engineering, the phased training process is independently trained for each output layer of the double-tailed neural network (i.e., the tail of the pile diameter decision and the tail of the pile number decision), which helps the model to more accurately capture the specific rules of each design parameter (pile diameter and pile number). Each output layer will gradually optimize its prediction performance based on the common features extracted by the public hidden layer, combined with its own specific weight matrix and activation function. After the independent training phase, joint training is performed. The purpose of joint training is to establish a more coordinated relationship between the two output layers and between them and the public hidden layer, ensuring that the entire network model can consider all relevant factors when recommending a design scheme, achieving the optimal overall performance. During the joint training process, network parameters such as weights, biases, learning rates, etc. are constantly adjusted and optimized based on feedback from actual engineering cases to reduce the gap between predicted values and actual values.

[0089] It is worth noting that to further improve the generalization ability and robustness of the model, regularization, dropout, early stopping, etc. training methods can be used. Regularization techniques introduce additional penalty terms to limit model complexity and prevent overfitting; dropout techniques enhance the generalization ability of the model by randomly dropping some neurons during training; early stopping techniques terminate training early when performance starts to decline to avoid overfitting. The combination of these techniques can significantly improve the adaptability and accuracy of the model when dealing with new engineering cases.

[0090] As shown in FIG. 8, Figure 4 Figure 4 is a schematic diagram of the comprehensive training method of “pre-training + phased training”, the training process includes (a) pre-training the model, (b) weight migration, (c) freezing the public + tail A weight, (d) freezing the public + tail B weight, the pre-training model refers to repeatedly iterating learning to enable the single-tailed neural network to learn the basic mapping relationship between the design parameters and the pile type (including the pile diameter and the pile number), thereby obtaining the target weight, the weight migration refers to migrating the target weight to the double-tailed neural network, the freezing the public + tail A weight refers to freezing the weights of the public hidden layer and the tail A of the pile diameter decision, and the freezing the public + tail B weight refers to freezing the weights of the public hidden layer and the tail B of the pile number decision.

[0091] In step S30, the physical information and evaluation indicators of the design scheme are obtained, and the scheme calculation and index evaluation of each design scheme in the design scheme group are performed based on the physical information and evaluation indicators of the design scheme, to obtain the evaluation scores of each design scheme.

[0092] ​It should be noted that the physical information of the design scheme refers to various technical parameters specified in the business rules and design specifications of the design scheme, and the evaluation index refers to a series of quantitative standards for measuring the advantages and disadvantages of the design scheme. In the design of the pile foundation scheme of the bridge engineering group, the evaluation index can include the bearing capacity of the pile foundation, stability analysis, settlement control, etc., and the embodiment is not limited specifically.

[0093] It can be understood that the scheme recommended according to the intelligent algorithm needs to be optimized through the target evaluation function in combination with the physical information required by the business rules, that is, the scheme comparison and selection are performed by setting the target evaluation function and the constraint condition. Taking the high-speed railway as an example, the frequently set target optimization function is the engineering cost and the construction difficulty, and the embodiment is not limited specifically.

[0094] Step S40, determining a recommended scheme according to the evaluation scores of the respective design schemes.

[0095] It should be noted that the design scheme with the highest score can be automatically selected as the final recommended scheme according to the evaluation scores of the respective design schemes, and the final recommended scheme can also be determined by other means, and the embodiment is not limited specifically.

[0096] It can be understood that the process of determining the recommended scheme not only depends on the absolute value of the evaluation score, but also needs to consider the actual demand of the project, the technical feasibility, the economic cost and the environmental impact and multiple aspects. Although the evaluation score provides a basis for quantitative evaluation, it can also be combined with expert opinions, historical case analysis and the latest engineering technology progress for comprehensive judgment, so as to further improve the accuracy of scheme recommendation.

[0097] In a feasible implementation manner, step S40 can include: sorting the respective design schemes according to the evaluation scores of the respective design schemes from high to low to obtain a design scheme sequence; and taking the design scheme with a preset sorting value in the design scheme sequence as the recommended scheme.

[0098] It should be noted that all the generated schemes are sorted according to the scores, and the top-K schemes with the highest scores are selected as the recommended schemes for detailed design of the schemes.

[0099] It can be understood that in the scheme design of bridge group pile foundation, the scheme detailed design is carried out according to the recommended scheme, that is, the intelligent selection algorithm of double-tailed neural network pile foundation is carried out in the whole bridge design subsystem of the bridge intelligent design system. The specific process includes: (1) defining the sample format, according to the bridge scheme elements affecting the group pile foundation scheme, abstracting and refining the sample format; (2) constructing a sample set, collecting the parameters required by the sample format and the corresponding group pile foundation type from the existing bridge engineering projects; (3) designing the network model, using deep learning frameworks such as pytorch and tensorflow to define the neural network model of the pile foundation intelligent selection algorithm, and requiring the number of parameters in the input layer to be unified with the number of parameters required by the sample format; (4) neural network training, inputting the data sample into the neural network to start training, and stopping training when the evaluation indicators such as accuracy, recall rate and F1 value reach the evaluation depth requirement; (5) model migration deployment, after the neural network model is successfully trained, it is deployed to the server and connected to the business system; (6) model reasoning application, carrying out scheme design in the bridge intelligent design system, extracting characteristic parameters such as main girder type, main span length, main girder length, pier height, pier position and terrain characteristics according to the bridge method collected by the system, inputting the group pile foundation recommendation algorithm, and using the neural network model to recommend the specific pile foundation design scheme; (7) scheme expert review, according to the expert experience, the foundation of the bridge scheme is audited to determine the group pile scheme that meets the engineering requirements; (8) scheme optimization and improvement, according to the expert review opinion, the original scheme is modified to a reasonable new scheme; (9) scheme sample storage, the input characteristic parameters of the algorithm such as main girder type, main span length, main girder length, pier height, pier position and terrain characteristics, and the new scheme optimized by the expert are stored in the database to form a new sample record; (10) network model upgrade, with the design of new bridge scheme, a large number of new sample data records are added to the sample set, and the new data set is called again for model training to generate a new neural network model; (11) repeat (5) to (10) to realize the cycle iteration of the neural network upgrade; (12) algorithm popularization and application, when the accuracy of the algorithm meets the needs of engineering design, the popularization and application of the algorithm is carried out.

[0100] As Figure 5 shown, Figure 5This diagram illustrates the interface between the intelligent pile foundation selection algorithm and the full-bridge design subsystem. The intelligent bridge design system includes a pile foundation selection algorithm module, an intelligent span arrangement module, an intelligent optimization module, a parametric modeling module, a finite element calculation module, and a full-bridge design subsystem. The pile foundation selection algorithm module interfaces with the full-bridge design subsystem. Within this module, foundation selection conditions are obtained from the foundation design within the full-bridge design subsystem to determine whether a pile foundation scheme should be adopted. If not, other foundation schemes are returned to the geometry engine within the full-bridge design subsystem. If adopted, intelligent pile foundation selection is performed, and the pile foundation configuration library is queried. Intelligent pile foundation selection is achieved by summarizing rule definitions from an experience case library and combining them with algorithm design. Specifically, pile foundation configurations are defined based on these rules to form the pile foundation configuration library. The algorithm design calls the pile foundation configurations from this library, which then returns the configurations to the bridge foundation library within the full-bridge design subsystem. Based on the information in the bridge foundation library, a pile foundation sample library is generated, which is then used for intelligent algorithm training before deploying the intelligent pile foundation selection. In the full bridge design subsystem, the pile foundation scheme is called from the bridge foundation library and rendered and displayed through the geometry engine, thereby obtaining the foundation design in the bridge span scheme design.

[0101] like Figure 6 As shown, Figure 6 This diagram illustrates the technical principle of a method for recommending design schemes for traffic engineering structures that integrates physical information and neural networks. It generates input parameter vectors through multi-condition perturbations, performs model inference based on these vectors, obtains multiple scheme outputs, conducts multi-scheme analysis, and evaluates the schemes based on the objective function and constraints. The objective function is to minimize cost, and the constraints are that parameters such as deformation, settlement, stiffness, and load-bearing capacity must be less than or equal to limit values. Based on the scheme evaluation results, the optimal scheme is recommended and added to a case library for model training and transfer.

[0102] like Figure 7 As shown, Figure 7 This document outlines a flowchart for intelligent pile foundation selection based on a two-tailed neural network. The process begins with bridge design, followed by bridge foundation design, and then intelligent pile foundation selection. The intelligent pile foundation selection design is reviewed by experts to determine if it meets design requirements. If not, the bridge foundation design is redesigned; if it does, new cases are added to the database based on the design. Finally, the algorithm is promoted and applied. Specifically, the intelligent pile foundation selection service includes: defining the sample format; constructing the sample set; designing the two-tailed neural network; training the neural network model; adding new case samples to the database; upgrading the model training; and deploying the model to the server to implement the intelligent pile foundation selection service.

[0103] The embodiment provides a traffic engineering construction design scheme recommendation method fusing physical information and a neural network, and the embodiment is characterized in that: first, the characteristic parameters of a design scheme are acquired, and an input parameter vector group is determined according to the characteristic parameters, wherein the input parameter vector group is obtained by vectorization and perturbation according to the characteristic parameters; the input parameter vector group is input into a scheme recommendation model to obtain a design scheme group, wherein the scheme recommendation model comprises an input layer, a fully connected public hidden layer and a plurality of regional fully connected output layers; the physical information and evaluation indexes of the design scheme are acquired, the scheme calculation and index evaluation of each design scheme in the design scheme group are respectively performed according to the physical information and evaluation indexes of the design scheme, and the evaluation scores of the design schemes are obtained; and the recommended scheme is determined according to the evaluation scores of the design schemes, so that the traffic engineering survey design scheme can be accurately evaluated, and the accuracy of the recommended scheme is effectively improved.

[0104] In conclusion, the input parameter vector group obtained by vectorization and perturbation of the characteristic parameters of the design scheme is input into the scheme recommendation model, the scheme calculation and index evaluation of the design scheme group output by the model are performed according to the physical information and evaluation indexes of the design scheme, and then the recommended scheme is determined according to the evaluation scores of the design schemes, so that the technical defect that the current intelligent algorithm evaluation index cannot adapt to the quality of the traffic engineering survey design scheme generated by the intelligent algorithm and cannot evaluate the advantages and disadvantages of the intelligent algorithm, thereby the accuracy of the recommended scheme is poor, the traffic engineering survey design scheme can be accurately evaluated, and the accuracy of the recommended scheme is effectively improved.

[0105] Based on the first embodiment, in the second embodiment, the same or similar contents as the above-mentioned first embodiment can be referred to the above description, and the subsequent will not be described in detail. On this basis, please refer to Figure 8 , and the step S30 further comprises steps S301-S303:

[0106] In step S301, the physical information and evaluation indexes of the design scheme are acquired.

[0107] It should be noted that the physical information of the design scheme refers to various technical parameters specified in the business rules and design specifications of the design scheme, including but not limited to the mechanical properties of materials, the geometric dimensions of structures, the load conditions of the environment, etc., and the physical information constitutes the basis of the design scheme. The evaluation indexes are set according to the actual engineering requirements, and are a series of standards for measuring the advantages and disadvantages of the design scheme, such as cost, construction difficulty, safety, durability, environmental impact, etc.

[0108] It can be understood that the analytic hierarchy process is used to construct the evaluation index system, which can decompose complex evaluation indexes into multiple levels, and each level contains several specific evaluation indexes, forming a tree structure. On each level, according to the experience of experts and historical data, the weight of each evaluation index is determined to reflect its contribution to the overall advantages and disadvantages of the design scheme. In this way, multiple aspects of the design scheme can be evaluated more systematically, ensuring the comprehensiveness and accuracy of the evaluation. According to the evaluation index system, the evaluation index corresponding to the current scheme design is determined.

[0109] It can be understood that in the bridge engineering group pile foundation scheme design, the pile foundation scheme evaluation index system is constructed, and the evaluation indexes include pile foundation settlement, bridge pier stiffness, material consumption, construction period, safety risk, engineering cost, etc. This embodiment does not make specific limitations. Taking high-speed railway as an example, the high-speed railway specification requires that the pile foundation settlement index requirement cannot exceed 20mm, and under the premise of meeting the conditions, the smaller the settlement is, the better. The bridge pier stiffness refers to the comprehensive stiffness including pile foundation deformation and bridge pier deformation, and the high-speed railway specification requires that the longitudinal stiffness of the bridge pier cannot be less than 360KN / cm, and under the condition of meeting the stiffness, the smaller the longitudinal stiffness value is, the less the bridge pier material consumption is, and the better the economy is. In the same bridge, the more pile foundation schemes are used, the higher the mechanization degree of the scheme is, the shorter the construction period is, the smaller the safety risk is, and the better the economy is.

[0110] Step S302, constructing a finite element model according to the physical information of the design scheme, and determining a target evaluation function according to the evaluation index of the design scheme.

[0111] It should be noted that the finite element model is a numerical method that discretizes a continuum into a finite number of elements and approximates the overall structure behavior by solving the physical field equations on these elements. In the bridge engineering group pile foundation scheme design, the finite element model can more accurately simulate the interaction between the pile foundation and the soil, as well as the deformation and stress distribution of the pile foundation under load. Through finite element analysis, the settlement, stiffness and other key physical parameters of the pile foundation can be predicted, providing a scientific basis for the optimization of the design scheme. According to the physical information such as business rules and design specifications, the finite element model can ensure the fitting degree of the model to the actual situation, thereby improving the accuracy of the analysis results.

[0112] It can be understood that the target evaluation function refers to a mathematical model for comprehensively evaluating various indicators of a design scheme, and can calculate a value reflecting the overall advantages and disadvantages of the design scheme according to the weights of various evaluation indicators and the actual performance of the design scheme. According to the evaluation indicators of the design scheme, the target evaluation function can be constructed to ensure that the evaluation process has a clear goal orientation, so that the evaluation result is more accurate. The target evaluation function should include all evaluation indicators and reasonably weight each indicator, so as to accurately convert the physical information of the design scheme into quantitative values of the evaluation indicators, thereby realizing comprehensive evaluation of the design scheme.

[0113] It is worth noting that when the target evaluation function is set as a multi-objective optimization problem, a weighted sum method can be used to construct the target evaluation function, in which each evaluation indicator is assigned different weights according to its importance and influence on the overall scheme. In the design of bridge engineering group pile foundation scheme, the comprehensive engineering cost of pile foundation + bridge pier is taken as the target evaluation function.

[0114] In step S303, scheme checking and index evaluation are performed on each design scheme in the design scheme group according to the finite element model and the target evaluation function, respectively, to obtain the evaluation score of each design scheme.

[0115] It should be noted that scheme checking refers to verifying and checking various technical parameters in the design scheme to ensure that the design scheme meets the requirements of business rules and design specifications. Index evaluation refers to comprehensively evaluating each design scheme in the design scheme group according to the evaluation index system to obtain its specific evaluation score, i.e. converting the design scheme from a theoretical concept to a quantifiable evaluation, which directly determines the accuracy and reliability of the recommended scheme.

[0116] It can be understood that through the comprehensive action of scheme checking and index evaluation, the specific evaluation score of each design scheme in the design scheme group can be obtained. These scores not only reflect the performance of the design scheme in various technical indicators, but also consider the economy, construction difficulty, environmental impact and other aspects of the design scheme. Therefore, determining the recommended scheme according to these evaluation scores can effectively improve the accuracy and reliability of the recommended scheme.

[0117] In a feasible implementation, step S303 can include: performing scheme checking on each design scheme in the design scheme group through the finite element model to obtain a set of checking results of each design scheme; and performing index evaluation on the set of checking results of each design scheme through the target evaluation function to obtain the evaluation score of each design scheme.

[0118] It should be noted that the finite element model, i.e. the finite element software fused with business rules and other physical information, i.e. the business index calculation system, sequentially connects the generated multiple schemes to the existing business index calculation system to generate multiple sets of quantitative calculation results. The quantitative calculation results are input into the evaluation objective function to calculate the comprehensive score of each scheme.

[0119] It can be understood that in the group pile foundation scheme design of bridge engineering, based on the multi-grade recommended scheme parameters, the finite element calculation and analysis function in the pile foundation design module is called to perform pile foundation calculation for each scheme respectively to obtain the index results of pile foundation settlement, pier stiffness, material consumption, construction period, safety risk, etc. of each scheme. The design engineering cost index is taken as the total score function, and other indexes are taken as constraint penalty conditions. The scheme with the lowest engineering cost under the condition of meeting the design index is selected as the final pile foundation recommended scheme for detailed design of the scheme.

[0120] In a specific implementation, in the group pile foundation scheme design of bridge engineering, all recommended parameters "quantity-diameter" (such as 8-1.25m scheme) are combined with the pile foundation configuration file to obtain all parameters, generate a finite element model, input the pile foundation calculation module for analysis, and obtain the index calculation results of pile foundation settlement, pier stiffness, material consumption, construction period, safety risk, etc. of each scheme. Taking the comprehensive engineering cost of pile foundation + pier as an example, the engineering cost of all schemes can be calculated according to the engineering economic index and material consumption, and the schemes are scored according to the engineering cost of the schemes.

[0121] In this embodiment, the finite element model and the objective evaluation function are constructed according to the physical information and evaluation indexes of the design scheme, and the scheme calculation and index evaluation of the design scheme are performed through the finite element model and the objective evaluation function, which can effectively improve the accuracy of the evaluation score of the design scheme.

[0122] It should be noted that the above examples are only used for understanding the present application and do not constitute a limitation on the traffic engineering structure design scheme recommendation method of the present application which fuses physical information and neural network. More forms of simple transformation based on this technical concept are within the protection scope of the present application.

[0123] The present application also provides a traffic engineering structure design scheme recommendation device fusing physical information and neural network, which is described in detail in Figure 9 The traffic engineering structure design scheme recommendation device fusing physical information and neural network comprises:

[0124] The determination module 10 is configured to obtain characteristic parameters of a design scheme and determine an input parameter vector group according to the characteristic parameters, wherein the input parameter vector group is obtained by vectorization and perturbation of the characteristic parameters.

[0125] The input module 20 is configured to input the input parameter vector group into the scheme recommendation model to obtain a design scheme group, wherein the scheme recommendation model comprises an input layer, a fully connected common hidden layer, and a plurality of regional fully connected output layers.

[0126] The evaluation module 30 is configured to obtain physical information and evaluation indexes of the design schemes, and perform scheme calculation and index evaluation on each design scheme in the design scheme group according to the physical information and the evaluation indexes of the design schemes, to obtain an evaluation score of each design scheme.

[0127] The determination module 10 is further configured to determine a recommended scheme according to the evaluation scores of the design schemes.

[0128] The embodiment provides a traffic engineering construction scheme recommendation device fusing physical information and a neural network. The embodiment firstly obtains characteristic parameters of a design scheme, and determines an input parameter vector group according to the characteristic parameters, wherein the input parameter vector group is obtained by vectorization and perturbation of the characteristic parameters; the input parameter vector group is input into a scheme recommendation model to obtain a design scheme group, wherein the scheme recommendation model comprises an input layer, a fully connected common hidden layer, and a plurality of regional fully connected output layers; physical information and evaluation indexes of the design schemes are obtained, and scheme calculation and index evaluation are performed on each design scheme in the design scheme group according to the physical information and the evaluation indexes of the design schemes, to obtain an evaluation score of each design scheme; and a recommended scheme is determined according to the evaluation scores of the design schemes, so that the traffic engineering survey design scheme can be accurately evaluated, and the accuracy of the recommended scheme is effectively improved.

[0129] In conclusion, the input parameter vector group obtained by vectorization and perturbation of the characteristic parameters of the design scheme is input into the scheme recommendation model, the design scheme group output by the model is subjected to scheme calculation and index evaluation according to the physical information and the evaluation indexes of the design schemes, and then a recommended scheme is determined according to the evaluation scores of the design schemes, so that the technical defect that the current intelligent algorithm evaluation index cannot adapt to the quality of the traffic engineering survey design scheme generated by the intelligent algorithm, and the intelligent algorithm cannot be evaluated, and the accuracy of the recommended scheme is poor is overcome, the traffic engineering survey design scheme can be accurately evaluated, and the accuracy of the recommended scheme is effectively improved.

[0130] Optionally, the determination module 10 is further configured to obtain the characteristic parameters of the design scheme, and vectorize the characteristic parameters to obtain a characteristic parameter vector; and perturb a plurality of characteristic parameters in the characteristic parameter vector to obtain the input parameter vector group.

[0131] Optionally, the input module 20 is further configured to obtain a sample library recommended by a scheme; pre-train a one-tailed neural network through samples in the sample library to obtain an initial recommendation model; and train the initial recommendation model through data samples in the sample library in stages to obtain a scheme recommendation model.

[0132] Optionally, the input module 20 is further configured to pre-train a one-tailed neural network through samples in the sample library to obtain a target weight; and migrate to a multi-tailed neural network according to the target weight to obtain an initial recommendation model.

[0133] Optionally, the evaluation module 30 is further configured to obtain physical information and evaluation indexes of a design scheme; construct a finite element model according to the physical information of the design scheme, and determine a target evaluation function according to the evaluation indexes of the design scheme; and perform scheme calculation and index evaluation on each design scheme in the design scheme group according to the finite element model and the target evaluation function, respectively, to obtain an evaluation score of each design scheme.

[0134] Optionally, the evaluation module 30 is further configured to perform scheme calculation on each design scheme in the design scheme group through the finite element model to obtain a calculation result set of each design scheme; and perform index evaluation on the calculation result set of each design scheme through the target evaluation function to obtain an evaluation score of each design scheme.

[0135] Optionally, the determination module 10 is further configured to sort the design schemes from high to low according to the evaluation scores of the design schemes to obtain a design scheme sequence; and take a design scheme with a preset sorting value in the design scheme sequence as a recommended scheme.

[0136] The traffic engineering structure design scheme recommendation device fusing physical information and a neural network provided in the application can solve the technical problem of how to accurately evaluate a traffic engineering survey and design scheme, and effectively improve the accuracy of a recommended scheme. Compared with the prior art, the traffic engineering structure design scheme recommendation device fusing physical information and a neural network provided in the application has the same beneficial effects as the traffic engineering structure design scheme recommendation method fusing physical information and a neural network provided in the above embodiments, and other technical features in the traffic engineering structure design scheme recommendation device fusing physical information and a neural network are the same as the features disclosed in the above method embodiments, which will not be repeated here.

[0137] The application provides a traffic engineering structure design scheme recommendation device fusing physical information and a neural network. The traffic engineering structure design scheme recommendation device fusing physical information and the neural network comprises at least one processor and a memory in communication connection with the at least one processor. The memory stores instructions executable by the at least one processor. The instructions are executed by the at least one processor to enable the at least one processor to perform the traffic engineering structure design scheme recommendation method fusing physical information and the neural network in Embodiment I.

[0138] Reference will be made to the following drawings to Figure 10 which show a structural schematic diagram of the traffic engineering structure design scheme recommendation device fusing physical information and the neural network suitable for being used to implement the embodiments of the application. The traffic engineering structure design scheme recommendation device fusing physical information and the neural network in the embodiments of the application can include, but is not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (Personal Digital Assistant), PADs (Portable Application Description), PMPs (Portable Media Player), vehicle-mounted terminals (for example, vehicle-mounted navigation terminals) and the like, and fixed terminals such as digital TVs, desktop computers and the like. Figure 10 The traffic engineering structure design scheme recommendation device fusing physical information and the neural network shown is only an example and should not bring any limitation to the functions and use ranges of the embodiments of the application.

[0139] As Figure 10As shown, the traffic engineering structure design proposal recommendation device that fuses physical information and neural networks can include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to a program stored in a read only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. In the RAM 1004, various programs and data required for operation of the traffic engineering structure design proposal recommendation device that fuses physical information and neural networks are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. In general, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; the storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the traffic engineering structure design proposal recommendation device that fuses physical information and neural networks to communicate wirelessly or by wire with other devices to exchange data. Although the traffic engineering structure design proposal recommendation device that fuses physical information and neural networks is shown as having various systems, it should be understood that all of the shown systems are not required to be implemented or possessed. More or fewer systems can be alternatively implemented or possessed.

[0140] In particular, according to embodiments of the present disclosure, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing program code for performing the methods illustrated by the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network by a communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above-described functions defined in the methods of the embodiments of the present disclosure are performed.

[0141] The traffic engineering structure design scheme recommendation device fusing physical information and a neural network provided in the application adopts the traffic engineering structure design scheme recommendation method fusing physical information and a neural network in the above embodiment, and can solve the technical problem of how to accurately evaluate a traffic engineering survey and design scheme, and further effectively improve the accuracy of a recommended scheme. Compared with the prior art, the traffic engineering structure design scheme recommendation device fusing physical information and a neural network provided in the application has the same beneficial effects as the traffic engineering structure design scheme recommendation method fusing physical information and a neural network provided in the above embodiment, and other technical features in the traffic engineering structure design scheme recommendation device fusing physical information and a neural network are the same as the features disclosed in the previous embodiment method, which will not be repeated here.

[0142] It should be understood that various parts of the present application can be realized by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0143] The above is merely specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0144] The present application provides a computer readable storage medium having computer readable program instructions (i.e. computer programs) stored thereon, the computer readable program instructions being used to execute the traffic engineering structure design scheme recommendation method fusing physical information and a neural network in the above embodiment.

[0145] The computer readable storage medium provided in the application may be, for example, a U disk, but is not limited to an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, system, or device, or any combination thereof. More specific examples of the computer readable storage medium may include, but are not limited to, an electrical connection with one or more conductive wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the embodiment, the computer readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer readable storage medium can be transmitted by any suitable medium, including but not limited to an electrical wire, an optical cable, an RF (Radio Frequency), and the like, or any suitable combination thereof.

[0146] The computer readable storage medium described above may be included in the traffic engineering structure design scheme recommendation device fusing physical information and neural networks, or may exist independently without being assembled into the traffic engineering structure design scheme recommendation device fusing physical information and neural networks.

[0147] The computer readable storage medium described above carries one or more programs, which, when executed by the traffic engineering structure design scheme recommendation device fusing physical information and neural networks, cause the traffic engineering structure design scheme recommendation device fusing physical information and neural networks to: obtain characteristic parameters of a design scheme, and determine an input parameter vector group according to the characteristic parameters, wherein the input parameter vector group is obtained by vectorization and perturbation according to the characteristic parameters; input the input parameter vector group into a scheme recommendation model to obtain a design scheme group, wherein the scheme recommendation model includes an input layer, a fully connected public hidden layer, and a plurality of regional fully connected output layers; obtain physical information and evaluation indexes of the design scheme, and perform scheme calculation and index evaluation on each design scheme in the design scheme group according to the physical information and the evaluation indexes of the design scheme, to obtain evaluation scores of the design schemes; and determine a recommended scheme according to the evaluation scores of the design schemes.

[0148] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0149] The computer program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0150] The modules involved in the embodiments of the present application can be implemented in software or hardware. In some cases, the names of the modules do not constitute a limitation on the modules themselves.

[0151] The readable storage medium provided by the application is a computer readable storage medium, and the computer readable storage medium stores computer readable program instructions (i.e. computer programs) for executing the traffic engineering structure design scheme recommendation method of fusing physical information and a neural network, and can solve the technical problem of how to accurately evaluate the traffic engineering survey design scheme and further effectively improve the accuracy of the recommended scheme. Compared with the prior art, the computer readable storage medium provided by the application has the same beneficial effects as the traffic engineering structure design scheme recommendation method of fusing physical information and a neural network provided by the above-mentioned embodiments, and will not be repeated here.

[0152] The application also provides a computer program product comprising a computer program, which, when executed by a processor, implements the steps of the traffic engineering structure design scheme recommendation method of fusing physical information and a neural network as described above.

[0153] The computer program product provided by the application can solve the technical problem of how to accurately evaluate the traffic engineering survey design scheme and further effectively improve the accuracy of the recommended scheme. Compared with the prior art, the computer program product provided by the application has the same beneficial effects as the traffic engineering structure design scheme recommendation method of fusing physical information and a neural network provided by the above-mentioned embodiments, and will not be repeated here.

[0154] The above-mentioned is only part of the embodiments of the application, and does not limit the patent scope of the application, and any equivalent structural transformation made by using the content of the application specification and drawings, or direct / indirect application in other related technical fields is included in the patent protection scope of the application.

Claims

1. A traffic engineering structure design scheme recommendation method that fuses physical information and a neural network, characterized by, The method comprises: obtaining characteristic parameters of a design scheme, and determining an input parameter vector group according to the characteristic parameters, wherein the input parameter vector group is obtained by vectorization and perturbation of the characteristic parameters; inputting the input parameter vector group into a scheme recommendation model to obtain a design scheme group, wherein the scheme recommendation model is an intelligent scheme recommendation algorithm based on a neural network or deep learning, and comprises an input layer, a fully connected public hidden layer, and a plurality of regional fully connected output layers; obtaining physical information and evaluation indexes of the design scheme, and performing scheme calculation and index evaluation on each design scheme in the design scheme group according to the physical information and evaluation indexes of the design scheme to obtain evaluation scores of the design schemes; determining a recommended scheme according to the evaluation scores of the design schemes; the method comprises: obtaining characteristic parameters of a design scheme, and performing vectorization on the characteristic parameters to obtain a characteristic parameter vector; performing perturbation on a plurality of characteristic parameters in the characteristic parameter vector to obtain an input parameter vector group; the method comprises: obtaining physical information and evaluation indexes of the design scheme; constructing a finite element model according to the physical information of the design scheme, and determining a target evaluation function according to the evaluation indexes of the design scheme; performing scheme calculation and index evaluation on each design scheme in the design scheme group according to the finite element model and the target evaluation function to obtain evaluation scores of the design schemes; the method further comprises, before inputting the input parameter vector group into the scheme recommendation model to obtain the design scheme group: obtaining a sample library for scheme recommendation; pre-training a one-tailed neural network through samples in the sample library to obtain an initial recommendation model; training the initial recommendation model through data samples in the sample library in stages to obtain a scheme recommendation model.

2. The method of claim 1, wherein, the method comprises: pre-training a one-tailed neural network through samples in the sample library to obtain target weights; migrating to a multi-tailed neural network according to the target weights to obtain an initial recommendation model.

3. The method of claim 1, wherein, the method comprises: performing scheme calculation on each design scheme in the design scheme group through the finite element model to obtain a calculation result set of each design scheme; performing index evaluation on the calculation result set of each design scheme through the target evaluation function to obtain evaluation scores of the design schemes.

4. The method of claim 1, wherein, the method comprises: The design schemes are ranked in descending order of the evaluation scores of the design schemes, to obtain a design scheme sequence; A design scheme with a ranking value of a preset value in the design scheme sequence is taken as a recommended scheme.

5. A traffic engineering structure design scheme recommendation device that fuses physical information and a neural network, characterized by, The traffic engineering structure design scheme recommendation device fusing physical information and neural networks comprises: A determination module configured to obtain feature parameters of design schemes, and determine an input parameter vector group according to the feature parameters, wherein the input parameter vector group is obtained by vectorization and perturbation of the feature parameters; An input module configured to input the input parameter vector group into a scheme recommendation model to obtain a design scheme group, wherein the scheme recommendation model is an intelligent scheme recommendation algorithm based on neural networks or deep learning, and comprises an input layer, a fully connected public hidden layer, and a plurality of regional fully connected output layers; An evaluation module configured to obtain physical information and evaluation indexes of the design schemes, and perform scheme calculation and index evaluation on each design scheme in the design scheme group according to the physical information and the evaluation indexes of the design schemes, to obtain evaluation scores of the design schemes; The determination module is further configured to determine a recommended scheme according to the evaluation scores of the design schemes, obtain feature parameters of design schemes, and vectorize the feature parameters to obtain feature parameter vectors; and perform perturbation on a plurality of feature parameters in the feature parameter vectors to obtain an input parameter vector group. The evaluation module is further configured to obtain physical information and evaluation indexes of the design schemes, construct a finite element model according to the physical information of the design schemes, determine a target evaluation function according to the evaluation indexes of the design schemes, and perform scheme calculation and index evaluation on each design scheme in the design scheme group according to the finite element model and the target evaluation function, to obtain evaluation scores of the design schemes. 6.A traffic engineering structure design scheme recommendation device that fuses physical information and a neural network, characterized by The traffic engineering structure design scheme recommendation device fusing physical information and neural networks comprises a memory, a processor, and a traffic engineering structure design scheme recommendation program fusing physical information and neural networks stored on the memory and executable on the processor, wherein the traffic engineering structure design scheme recommendation program fusing physical information and neural networks is configured to implement the traffic engineering structure design scheme recommendation method fusing physical information and neural networks according to any one of claims 1 to 4.

7. A storage medium, characterized by The storage medium stores a traffic engineering structure design scheme recommendation program fusing physical information and neural networks, and the traffic engineering structure design scheme recommendation program fusing physical information and neural networks implements the traffic engineering structure design scheme recommendation method fusing physical information and neural networks according to any one of claims 1 to 4 when executed by the processor.

Images