Protective device optimization design method and computer-readable storage medium

Abstract

The present invention discloses a protective device optimization design method and computer-readable storage medium, which belongs to the technical field of bridge engineering structure optimization, including the following steps: designing the constraints and objectives of the protective device according to the basic information of the project; obtaining the preliminary parameters and similarity ratio coefficient set of the protective device according to the constraints and objectives; establishing a prediction model for the crushing performance of the protective device, obtaining the basic design parameters of the protective device according to the preliminary parameters and the similarity ratio coefficient set, calculating the crushing curve of the protective device under different basic design parameters, and then calculating the protective performance and engineering cost of the protective device under ship collision conditions. If the defense requirements and engineering cost requirements are met, the optimization design result set is entered, and the similarity ratio coefficient set is returned to perform the next round of iterative calculation. The protective device optimization design method and computer-readable storage medium of the present application are based on the extreme learning machine prediction model, meet the constraints and design objectives in a variety of actual engineering projects, and improve the design efficiency of the protective device.

Description

Technical Field

[0001] The present invention relates to the field of bridge engineering structure optimization, and in particular to a protective device optimization design method and a computer-readable storage medium. Background Art

[0002] In the optimization design of bridge engineering structures, improving the impact resistance of protective devices is a key step in ensuring bridge safety. Currently, the industry generally adopts optimization design methods such as enhancing the impact resistance of protective devices with high-strength materials, installing anti-collision blocks around the main supporting structures of bridges to absorb impact energy, and installing vegetation or green belts on both sides of the bridge to beautify the environment and assist in absorbing impact forces. However, these methods face many challenges in practical application: First, due to the significant differences in the structural forms and surrounding environments of different bridges, existing protective devices often show insufficient adaptability when dealing with specific working conditions; second, the impact crushing performance of protective devices is affected by the coupling of multiple structural parameters and complex nonlinear relationships. Traditional modeling methods cannot fully describe this complexity, making it difficult to accurately predict and control the protective effect; in addition, the high maintenance cost also limits the widespread application of existing protective devices. If the protective device fails to effectively resist impact, it may cause serious damage to the bridge structure and even lead to serious consequences such as casualties. Summary of the Invention

[0003] The present invention aims to provide a protective device optimization design method and computer-readable storage medium that meet multiple constraints and design objectives in actual engineering projects. The protective device optimization design method is based on an extreme learning machine prediction model, can adapt to ever-changing engineering needs, and effectively solves the shortcomings of existing protective devices in meeting multiple actual engineering project constraints and design objectives.

[0004] To achieve the above object, the technical solution of the present invention is:

[0005] A protective device optimization design method, comprising:

[0006] Step S1: Clarify basic project information;

[0007] Step S2: designing constraints and objectives of the protection device based on the basic project information;

[0008] Step S3: obtaining preliminary parameters and similarity ratio coefficient sets of the protection device according to the constraints and objectives of the protection device;

[0009] Step S4: establishing a prediction model for the crush performance of the protective device; obtaining multiple sets of basic design parameters of the protective device based on the preselected parameters and similarity ratio coefficient set of the protective device, and using the established prediction model to calculate the crush curves of the protective device under different basic design parameters;

[0010] Step S5: Calculate the protection performance of the protection device under ship collision conditions based on the crush curve; if the protection requirements are met, enter the optimization design result set and return to the similarity ratio coefficient set for the next round of iterative calculation;

[0011] Step S6: Calculate the engineering cost of the protective device according to the basic design parameters and the similarity ratio coefficient; if the engineering cost requirements are met, enter the optimization design result set and return to the similarity ratio coefficient set for the next round of iterative calculation.

[0012] Optionally, the basic project information includes: pier width, water area margin, defense requirements and project cost requirements.

[0013] Optionally, the constraints of the protective device in step S2 include: the width of the protective device is less than or equal to the water area margin, and the axial length of the protective device is greater than the pier width but not more than 1.1 times the pier width.

[0014] Optionally, the objectives of designing the protective device in step S2 include: meeting defense requirements and meeting project cost requirements; meeting the defense requirements includes: the impact force value after being reduced by the protective device is less than or equal to the peak impact force that the pier can withstand; meeting the project cost requirements includes: on the basis of meeting the defense requirements, selecting the solution with the lowest project cost.

[0015] Optionally, in step S3, the preliminarily selected parameters include: the width of the protective device, the axial length of the protective device, the type of the protective device, the shell thickness, the circumferential angle steel plate thickness, the axial angle steel plate thickness, the circumferential angle steel spacing, and the type of filling material; the width of the protective device and the axial length of the protective device are calculated according to the constraints of the protective device, and the widths of the protective devices are arranged in descending order as a width parameter set of the protective device; a similarity ratio coefficient set is calculated according to the ratio of the width parameter set of the protective device to the model size; the types of the protective device include: steel protective device, steel-PUF protective device, and steel-PUF-GFRP protective device; the recommended type of protective device is determined according to the axial length of the protective device.

[0016] Optionally, in step S4, a prediction model for the crushing performance of the protective device is established, including: training based on an extreme learning machine prediction model to obtain the relationship between the basic design parameters and the crushing performance; the input variables of the extreme learning machine prediction model include: deformation, shell thickness, circumferential angle steel plate thickness, axial angle steel plate thickness, circumferential angle steel spacing and filling material type, and the filling material types include: steel, steel-PUF and steel-GFRP-PUF; the output variables of the extreme learning machine prediction model include: impact force value; the crushing curve is a curve in which the impact force value of the protective device changes with the deformation.

[0017] Optionally, in step S4, basic design parameter = preliminary parameter × similarity ratio coefficient in the similarity ratio coefficient set.

[0018] Optionally, calculating the protective performance of the protective device under ship collision conditions in step S5 includes: using a DME method to evaluate the protective performance of the protective device under ship collision conditions under all similarity ratio coefficients.

[0019] Optionally, the similarity ratio coefficient set is returned for the next round of iterative calculation, including: reselecting a similarity ratio coefficient in the similarity ratio coefficient set, and generating a set of corresponding basic design parameters for calculation.

[0020] A computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the computer program implements a protective device optimization design method as described in any one of the above.

[0021] This application uses an extreme learning machine prediction model to iteratively optimize the structure of the protective device, meeting the constraints and design goals in a variety of actual engineering projects, and providing a reliable basis and guidance for practical engineering applications. By constructing an extreme learning machine prediction model, the crushing performance of the protective device is quickly predicted based on known basic design parameters, reducing the time for model testing and numerical simulation, and significantly improving the design efficiency of the protective device. At the same time, the extreme learning machine prediction model provides data-driven technical support for the optimization and evaluation of the performance of the protective device, achieving rapid optimization and performance evaluation of the multi-objective optimization problem of the protective device, helping to achieve a balance between performance and engineering costs, and further promoting the design and optimization process of the protective device.

[0022] In order to make the above features and advantages of the present application more obvious and easy to understand, embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Flowchart of the protective device optimization design method proposed in this application.

[0024] Figure 2Schematic diagram of the neural network structure of the ELM prediction model.

[0025] Figure 3 Schematic diagram of the crush curve.

[0026] Figure 4 Schematic diagram of the relationship between the calculation accuracy and the number of iterations of the ELM prediction model.

[0027] Figure 5 Figure (a) is a schematic diagram showing the comparison between the actual impact force value of the training set and the predicted impact force value.

[0028] Figure 5 Figure (b) is a schematic diagram showing the comparison between the actual impact force value and the predicted impact force value of the test set.

[0029] Figure 6 Figure (a) is the determination coefficient R of the training set 2 Schematic diagram of .

[0030] Figure 6 Figure (b) shows the coefficient of determination R of the test set 2 Schematic diagram of .

[0031] Figure 7 Comparison of the crush curves of the ELM prediction model and the FEA model.

[0032] Figure 8 Figure (a) is a schematic diagram of a Class III waterway (1000DWT) bridge in working condition 1.

[0033] Figure 8 Figure (b) is a schematic diagram of the Class III waterway (1000DWT) bridge in working condition 2.

[0034] Figure 8 Figure (c) is a schematic diagram of the Class III waterway (1000DWT) bridge in working condition 3.

[0035] Figure 9 Figure (a) is a schematic diagram of the number of iterations of the ELM prediction model and the peak impact force in working condition 1.

[0036] Figure 9 Figure (b) is a schematic diagram of the number of iterations of the ELM prediction model and the project cost in working condition 1.

[0037] Figure 10 Figure (a) is a schematic diagram of the number of iterations of the ELM prediction model and the peak impact force in working condition 2.

[0038] Figure 10 Figure (b) is a schematic diagram of the number of iterations of the ELM prediction model and the project cost in working condition 2.

[0039] Figure 11 Figure (a) is a schematic diagram of the number of iterations of the ELM prediction model and the peak impact force in working condition 3.

[0040] Figure 11 Figure (b) is a schematic diagram of the number of iterations of the ELM prediction model and the project cost in working condition 3. DETAILED DESCRIPTION

[0041] To make the purpose and technical solutions of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] In one embodiment of this application, please refer to Figure 1 , Figure 1 This is a flow chart of the protective device optimization design method proposed in this application, including steps S1 to S6.

[0043] Step S1: Clarify basic project information.

[0044] Step S2: Design the constraints and objectives of the protective device based on the basic project information.

[0045] Step S3: Obtaining a set of preliminary parameters and similarity ratio coefficients of the protective device according to the constraints and objectives of the protective device.

[0046] Step S4: Establish a prediction model for the crush performance of the protective device; obtain multiple sets of basic design parameters of the protective device based on the preliminary parameters and similarity ratio coefficient set of the protective device, and use the established prediction model to calculate the crush curve of the protective device under different basic design parameters.

[0047] Step S5: Calculate the protective performance of the protective device under ship collision conditions based on the crush curve of the protective device; if the protection requirements are met, enter the optimization design result set and return to the similarity ratio coefficient set for the next round of iterative calculation.

[0048] Step S6: Calculate the engineering cost of the protective device based on the basic design parameters and the similarity ratio coefficient; if the engineering cost is less than the engineering cost of the basic project information, enter the optimization design result set and return to the similarity ratio coefficient set for the next round of iterative calculation.

[0049] In step S1, basic project information includes: pier width, water area margin, defense requirements and construction cost requirements.

[0050] In step S2, the pier width, excess water area, defense requirements, and project cost requirements jointly determine the size and performance requirements of the protective device. The pier width and excess water area directly limit the width and axial length of the protective device. The width of the protective device cannot exceed the excess water area. The axial length of the protective device must be coordinated with the pier width. That is, the axial length of the protective device should be greater than the pier width but generally not exceed 1.1 times the pier width. The constraints of the protective device are expressed by the formula:

[0051]

[0052] Among them, W F is the width of the protective device, W B is the water area surplus scale, L B is the width of the pier, L F is the axial length of the guard. This range of axial lengths provides the necessary flexibility in guard layout while allowing for the placement of a certain number of rubber damping elements between the guard and the pier to adjust the gap between them. This design flexibility not only helps mitigate energy transfer during a ship impact but also provides room for further design optimization.

[0053] As an example, the goals of designing the protective device in step S2 include: meeting defense requirements and project cost requirements. Meeting defense requirements and project cost requirements are the core goals of optimizing the protective device. Meeting defense requirements includes: ensuring that the protective performance of the protective device meets the minimum defense requirements, that is, the impact force value after reduction by the protective device should be less than the peak impact force that the bridge pier can withstand. Meeting project cost requirements includes: on the basis of meeting defense requirements, the ultimate goal of the optimized design of the protective device is to reduce the project cost, that is, by comparing different schemes and other means, to maximize economic benefits while meeting defense requirements. The goal of the protective device is expressed by the formula:

[0054]

[0055] Among them, F F F is the impact force reduced by the protective device, min is the peak impact force that the pier can withstand, E c is the project cost, min is the minimum value.

[0056] In step S3, the preliminarily selected parameters include: the width of the protective device, the axial length of the protective device, the type of protective device, the shell thickness, the thickness of the circumferential angle steel plate, the thickness of the axial angle steel plate, the spacing between the circumferential angle steels, and the type of filler material. The width and axial length of the protective device are calculated based on the constraints of the protective device obtained in step S2; the shell thickness, the thickness of the circumferential angle steel plate, the thickness of the axial angle steel plate, the spacing between the circumferential angle steels, and the type of filler material are set as the initial parameters of the protective device. The calculated widths of the protective device are arranged in descending order as a set of protective device width parameters for subsequent optimization. A set of similarity ratio coefficients is then calculated based on the ratio of the protective device width parameter set to the model size; the similarity ratio coefficient set includes the similarity ratio coefficients calculated based on the ratio of the protective device width to the model size.

[0057] Furthermore, the types of protective devices include steel, steel-PUF, and steel-PUF-GFRP. The recommended type of protective device is determined by the axial length of the protective device. For example, in scenarios where the pier width is narrow, steel-PUF-GFRP protective devices are recommended. This is because the axial length of the protective device is short, and its energy absorption capacity per unit length needs to be maximized to ensure that the protective performance requirements are met. In scenarios where the pier width is wide, steel protective devices are recommended. Although their energy absorption capacity per unit length is relatively low, their structure is more easily conformed to the bow of the ship and has greater adaptability, thereby effectively dispersing the impact force. In scenarios where the pier width is moderate, steel-PUF protective structures are recommended because they can achieve a good balance between protective performance and energy absorption capacity.

[0058] Furthermore, the model size is expressed as the size of the scaled model of the protective device used for experimental testing. Since the protective device is very large, it is difficult to conduct full-scale testing. It is necessary to scale the protective device so that the scaled model of the protective device meets the requirements of laboratory testing, that is, the scaled model of the protective device can be placed in the laboratory.

[0059] In step S4, the prediction model for the crushing performance of the protective device is established, including: training based on the extreme learning machine prediction model (ELM) to obtain the relationship between the basic design parameters of the protective device and the crushing performance. The embodiment of the present application takes the ELM prediction model of the machine learning model as an example, which can efficiently capture the potential correlation and action mechanism between the basic design parameters, thereby providing higher accuracy for the crushing performance prediction. The ELM prediction model includes: an input layer, a hidden layer and an output layer. The input layer includes: N input variables; the hidden layer includes: hidden layer nodes.

[0060] Specifically, define a training set containing N samples, see Figure 2 , Figure 2The schematic diagram of the neural network structure of the ELM prediction model is [x1, x2, x3···x N ] represents a given set of input variables; represents the weight from the Nth input variable in the input layer to the jth neuron in the hidden layer, b j Represents the threshold of the jth neuron, where the weight from the Nth input variable in the input layer to the jth neuron in the hidden layer and the threshold of the neurons in the hidden layer are directly determined by random initialization. No further adjustment is required after setting, thus avoiding the multiple weight and bias updates required by the complex back propagation algorithm in the BP neural network, greatly reducing the computational complexity. The output weight matrix is used to calculate between the hidden layer and the output layer of the hidden layer nodes, which can be expressed as:

[0061]

[0062] in, Indicates containing The output weight matrix between the hidden layer and the output layer of the hidden layer nodes, Indicates the The output weight from the hidden layer node to the jth neuron, T represents the target matrix of the output variable, Indicates the output variables, and m represents the dimension of the target matrix T.

[0063] The jth input variable x j The corresponding input variable x j The weight w of the i-th neuron ji The product of the neuron's threshold is then activated by the activation function f(x j ) to get the output value y of the jth neuron j .contain The activation function of a single hidden layer neural network with hidden layer nodes is:

[0064]

[0065] Among them, x j represents the jth input variable, β i represents the output weight matrix between the i-th hidden layer and the output layer, w ji Represents the jth input variable x j to the weight of the i-th neuron.

[0066] Rewrite formula (5) into matrix form:

[0067] H·β=T

[0068] Formula (6) is further developed to obtain:

[0069] β=H -1 ·T

[0070] Formula (7)

[0071] Where H represents the hidden layer output matrix, which is directly determined by analytical solution without relying on iterative optimization process, making the overall calculation more efficient. β represents the output layer matrix. The output layer matrix β and the hidden layer output matrix H are expressed as follows:

[0072]

[0073] In this example, a database was established using 27 sets of real-world crush curve data for three types of protective devices, yielding a total of 1,420 input variables. These variables were then divided into a training set (80%) and a test set (20%). The 1,136 input variables in the training set were used to train the ELM prediction model, and the reliability of the trained ELM prediction model was tested using the 284 input variables in the test set.

[0074] For example, the input variables of the ELM prediction model include deformation, shell thickness, circumferential angle plate thickness, axial angle plate thickness, circumferential angle steel spacing, and filler material type. The filler material types include steel, steel-PUF, and steel-GFRP-PUF. The output variables of the ELM prediction model include impact force value. Figure 3 , Figure 3 This is a schematic diagram of the crush curve, which plots the impact force versus deformation of the protective device. These input variables were selected to comprehensively characterize the basic design parameters of the protective device, ensuring that the established prediction model accurately reflects the relationship between the protective device's crush performance and its basic design parameters.

[0075] Furthermore, since the input variables used have different data magnitudes, data normalization is needed to solve this problem. Scaling the data ensures that all input variables are processed identically and avoids biasing the ELM prediction model. The formula for data normalization is as follows:

[0076]

[0077] Among them, x i represents the i-th input variable, x i,norm represents the data normalization value of the i-th input variable after data normalization, x max Indicates the maximum value of the selected input variable, x min Indicates the minimum value of the selected input variable.

[0078] Furthermore, during the training process, the input variables are randomly permuted to expose the ELM prediction model to more diverse data and exclude the similarity of input variables from the same reference.

[0079] Furthermore, the training data set is trained through the ELM prediction model to obtain a neural network with determined weights for each part, thereby building the relationship between the input variables and the output variables. Figure 4 , Figure 4 This is a diagram showing the relationship between the calculation accuracy and the number of iterations of the ELM prediction model. The maximum number of iterations is set to 100. Figure 4 It can be seen that with the increase of the number of iterations, the relative error of the ELM prediction model gradually decreases. After the number of iterations exceeds 50, the calculation relative error of the model can be stabilized within 0.2%. It can be seen that with the increase of the number of iterations, although the convergence speed decreases, a higher calculation accuracy can be achieved. Therefore, the number of iterations needs to be limited to avoid overfitting the model and control the length of the training process.

[0080] As an example, when performing inversion analysis based on the ELM prediction model, the number of neurons in the hidden layer was determined to be 100 through trial and error. The single-hidden-layer neural network was trained using 1,136 sets of input variables from the training set and their corresponding real-world impact force values from the crushing curves. After training, cross-validation was performed using 284 randomly distributed sets of input variables from the test set. The impact force values from the crushing curves corresponding to these test set input variables were input into the trained ELM prediction model. After processing, the ELM prediction model output the predicted impact force values for the 284 sets of input variables in the test set to evaluate the prediction accuracy of the ELM prediction model.

[0081] See also Figure 5 , Figure 5 Figure (a) is a schematic diagram showing the comparison between the actual impact force value and the predicted impact force value of the training set. Figure 5 Figure (b) is a schematic diagram of the comparison between the actual impact force value and the predicted impact force value of the test set. The actual impact force value is compared with the predicted impact force value to test the performance and accuracy of the ELM prediction model.

[0082] Further, see Figure 6 , Figure 6 Figure (a) is the determination coefficient R of the training set 2 Schematic diagram, Figure 6 Figure (b) shows the coefficient of determination R of the test set 2 Schematic diagram of the coefficient of determination R 2 It is used as a measure to evaluate the ability of the ELM prediction model to explain the variability of the data, so the coefficient of determination R 2The higher the value of , the better the ELM prediction model fits. By comparing the coefficient of determination R of the randomly assigned training set and test set, the 2 To evaluate the performance of the ELM prediction model, the coefficient of determination R 2 It can be expressed as:

[0083]

[0084] Among them, y i represents the true impact force value output by the jth neuron, represents the predicted impact force value of the i-th input variable, N represents the number of samples in the training set or test set, represents y i The average value of . After calculation, the determination coefficient R of the training set of this embodiment is 2 The coefficient of determination R of the test set is 0.9738. 2 It is 0.9392, and there is no obvious overfitting phenomenon.

[0085] As an example, after establishing a prediction model for the crushing performance of a protective device, multiple sets of basic design parameters for the protective device are obtained based on the preliminary parameters of the protective device and the similarity ratio coefficient set in step S3. Specifically, basic design parameters = preliminary parameters × similarity ratio coefficients in the similarity ratio coefficient set. The basic design parameters of the experimental ELM prediction model are input into the trained ELM prediction model. A set of basic design parameters for the experimental ELM prediction model is shown in Table 1. The crushing curve obtained by the ELM prediction model is compared with the crushing curve calculated by the finite element model FEA model. Figure 7 , Figure 7 This is a comparison chart of the crush curves of the ELM prediction model and the FEA model. The comparison results show that the prediction results of the ELM prediction model are in good agreement with the calculation results of the FEA model. The effective deformations obtained by the ELM prediction model and the FEA model are 1.3m and 1.29m respectively. The peak impact force of the ELM prediction model in the elastic stage is 0.76MN, and that of the FEA model is 0.8MN. The calculation efficiency brought by the ELM prediction model is greatly improved, and it has good application prospects in structural optimization design.

[0086] Table 1 Basic design parameters of the experimental ELM prediction model

[0087]

[0088] For example, a set of basic design parameter data corresponds to a similarity ratio coefficient. After obtaining the similarity ratio coefficient, the basic design parameters of the protective device obtained under the similarity ratio coefficient are often not integers. To meet the integer data requirements of the project, the basic design parameters are corrected using rounding to obtain integer data for the basic design parameters.

[0089] In step S5, calculating the protective performance of the protective device under ship collision conditions includes: using the DME method to quickly evaluate the protective performance of the protective device under ship collision conditions under all similarity ratio coefficients. Returning the similarity ratio coefficient set for the next round of iterative calculation includes: reselecting a similarity ratio coefficient from the similarity ratio coefficient set and generating a set of corresponding basic design parameters for calculation.

[0090] In step S6, the engineering cost of the protective device under all similarity ratio coefficients is calculated; the similarity ratio coefficient set is returned to perform the next round of iterative calculation, including: reselecting the similarity ratio coefficient in the similarity ratio coefficient set, and generating a set of corresponding basic design parameters for calculation.

[0091] In one embodiment of this application, three Class III waterway (1000 DWT) bridges were selected as the objects of optimization design. Figure 8 , Figure 8 Figure (a) is a schematic diagram of a Class III waterway (1000 DWT) bridge in Condition 1. Figure 8 Figure (b) is a schematic diagram of a Class III waterway (1000 DWT) bridge in Condition 2. Figure 8 Figure (c) shows a schematic diagram of a Class III waterway (1000 DWT) bridge for Condition 3. The three bridges differ in terms of bridge type, span, and channel width. Basic project information is shown in Table 2. Based on the aforementioned optimization design process, the protective devices for each bridge were optimized.

[0092] Table 2 Basic information of three Class III waterway (1000 DWT) bridge projects

[0093]

[0094] The protective device designs for the three Class III waterway (1000 DWT) bridges mentioned above strive to achieve optimization within the project cost range, while meeting collision resistance requirements. For ease of comparison, the project cost only includes the material price of the ship impact-side protective devices. In actual projects, the cost of protective devices is also affected by comprehensive factors such as the overall size of the bridge piers and construction and installation. Furthermore, the design process assumes that the representative ship types all have straight-heeled bows, ignoring the possibility that shear and bulbous bow types, etc., could bypass the protective devices and directly impact the bridge piers.

[0095] As an example, after obtaining the basic information of the project, the constraints and objectives of the protective device are first designed according to the basic information of the project; secondly, the preliminary parameters and similarity ratio coefficient set of the protective device are obtained according to the constraints and objectives of the protective device, wherein the preliminary parameters include: the width of the protective device, the axial length of the protective device and the type of the protective device, and the similarity ratio coefficient set includes: the similarity ratio coefficient set of the ratio of the width parameter set of the protective device to the model size; then, based on the preliminary parameters and the similarity ratio coefficient set of the protective device, the basic design parameters of the protective device are obtained, a prediction model of the crushing performance of the protective device is established, and the crushing curve of the protective device under different basic design parameters is calculated; then, based on the crushing curve of the protective device, the protective performance of the protective device under ship collision conditions is calculated. If the defense requirements are met, the optimization design result set is entered. If not, the similarity ratio coefficient set is returned to for the next round of iterative calculation; finally, the engineering cost of the protective device is calculated based on the basic design parameters and the similarity ratio coefficient. If the engineering cost requirements are met, the optimization design result set is entered. If not, the similarity ratio coefficient set is returned to for the next round of iterative calculation.

[0096] As an example, see Figure 9 , Figure 9 Figure (a) is a schematic diagram of the number of iterations of the ELM prediction model and the peak impact force in working condition 1. In working condition 1, a total of 160 iterative calculations were completed. Among all the ELM prediction model schemes, the maximum impact force was 6.95MN and the minimum impact force was 4.85MN. There were 45 groups of schemes that met the defense requirements. Figure 9 Figure (b) shows the relationship between the number of ELM prediction model iterations and project cost for Condition 1. The project cost for all schemes ranged from 57,000 to 73,000 yuan, with four schemes meeting the cost requirements. After a comprehensive analysis of protection performance and cost, two schemes were ultimately determined to meet all design requirements.

[0097] As an example, see Figure 10 , Figure 10 Figure (a) is a schematic diagram of the number of iterations of the ELM prediction model and the peak impact force in working condition 2. In working condition 2, a total of 95 iterative optimizations were performed, and the calculated maximum impact force was 6.44MN, and the minimum impact force was 5.01MN. There were three groups of schemes that met the defense requirements. Figure 10 Figure (b) shows the relationship between the number of ELM prediction model iterations and project cost for Condition 2. The project cost for all schemes ranged from 57,500 to 79,700 yuan, with 12 schemes meeting the cost requirements. After a comprehensive analysis of protection performance and cost, three schemes were ultimately determined to meet all design requirements.

[0098] As an example, see Figure 11, Figure 11 Figure (a) is a schematic diagram of the number of iterations of the ELM prediction model and the peak impact force in working condition 3. In working condition 3, a total of 300 iterative calculations were performed, the maximum impact force was 7.03MN, and the minimum impact force was 6.95MN. There were three groups of schemes that met the defense requirements. Figure 11 Figure (b) shows the relationship between the number of ELM prediction model iterations and project cost for Condition 3. The project costs for all schemes ranged from 56,500 to 160,000 yuan, with 10 schemes meeting the cost requirements. After a comprehensive analysis of protection performance and cost, two schemes were ultimately determined to meet all design requirements.

[0099] As an example, please refer to Table 3, which shows the optimization design results of the protective device under three working conditions.

[0100] Table 3 Optimization design results of protective devices under three working conditions

[0101]

[0102]

[0103] Specifically, a comparative analysis of the optimized design results of the protective device under three working conditions also shows that the protective material with a filling material type of steel-PUF-GFRP is more suitable for scenarios with narrow pier widths, the protective material with a filling material type of steel is more suitable for scenarios with wide pier widths, and the protective material with a filling material type of steel-PUF is more suitable for scenarios with moderate pier widths.

[0104] As an example, the optimization design results of the protective device under three working conditions are selected, and the protective device with the best protection performance is selected. The similarity ratio coefficient corresponding to the selected protective device is called the optimal similarity ratio coefficient.

[0105] In another embodiment of the present application, a computer-readable storage medium is provided, storing a computer program, which, when executed by a processor, implements the above-mentioned method for optimizing the design of a protective device.

[0106] In summary, the design of bridge protective devices involves a variety of influencing factors, and different types of protective devices have their own scope of application. This application is based on the ELM prediction model, and the iterative optimization of the protective device structure meets the constraints and design goals in a variety of actual engineering projects, providing a reliable basis and guidance for actual engineering applications. By constructing the ELM prediction model, the crushing performance of the protective device can be quickly predicted based on known structural parameters, reducing the time for model testing and numerical simulation, and significantly improving the design efficiency of the protective device. At the same time, the ELM prediction model provides data-driven technical support for the optimization and evaluation of the performance of the protective device, realizing the rapid optimization and performance evaluation of the multi-objective optimization problem of the protective device, helping to achieve a balance between performance and engineering costs, and further promoting the design and optimization process of the protective device.

Claims

1. A protective device optimization design method, characterized in that: include: Step S1: Clarify the basic information of the project; Step S2: designing constraints and objectives of the protection device based on the basic project information; Step S3: obtaining preliminary parameters and similarity ratio coefficient sets of the protection device according to the constraints and objectives of the protection device; Step S4: establishing a prediction model for the crush performance of the protective device; obtaining multiple sets of basic design parameters of the protective device based on the preselected parameters and similarity ratio coefficient set of the protective device, and using the established prediction model to calculate the crush curves of the protective device under different basic design parameters; Step S5: Calculate the protection performance of the protection device under ship collision conditions based on the crush curve; if the protection requirements are met, enter the optimization design result set and return to the similarity ratio coefficient set for the next round of iterative calculation; Step S6: Calculating the engineering cost of the protective device according to the basic design parameters and the similarity ratio coefficient; If the engineering cost requirements are met, the optimization design result set is entered and the similarity ratio coefficient set is returned to perform the next round of iterative calculation; Establishing a prediction model for the crush performance of the protective device in step S4 includes: training based on an extreme learning machine prediction model to obtain a relationship between the basic design parameters and the crush performance; input variables of the extreme learning machine prediction model include deformation, shell thickness, circumferential angle steel plate thickness, axial angle steel plate thickness, circumferential angle steel spacing, and filler material type, wherein the filler material types include steel, steel-PUF, and steel-GFRP-PUF; output variables of the extreme learning machine prediction model include impact force value; and a crush curve is a curve showing the transformation of the impact force value of the protective device with the deformation; The basic design parameter=primary parameter×similarity ratio coefficient in the similarity ratio coefficient set.

2. The protective device optimization design method according to claim 1, characterized in that: The basic information of the project includes: pier width, water area margin, defense requirements and project cost requirements.

3. The protective device optimization design method according to claim 2, characterized in that: The constraints of the protective device in step S2 include: the width of the protective device is less than or equal to the water area margin, and the axial length of the protective device is greater than the pier width but not more than 1.1 times the pier width.

4. The protective device optimization design method according to claim 2, characterized in that: The objectives of designing the protective device in step S2 include: meeting defense requirements and meeting project cost requirements; meeting the defense requirements includes: the impact force value after being reduced by the protective device is less than or equal to the peak impact force that the bridge pier can withstand; meeting the project cost requirements includes: on the basis of meeting the defense requirements, selecting the solution with the lowest project cost.

5. The protective device optimization design method according to claim 3, characterized in that: In step S3, the preliminarily selected parameters include: the width of the protective device, the axial length of the protective device, the type of the protective device, the shell thickness, the thickness of the circumferential angle steel plate, the thickness of the axial angle steel plate, the spacing between the circumferential angle steels, and the type of filler material; the width of the protective device and the axial length of the protective device are calculated based on the constraints of the protective device, and the widths of the protective devices are arranged in descending order to form a width parameter set of the protective device; a similarity ratio coefficient set is calculated based on the ratio of the width parameter set of the protective device to the model size; the types of the protective device include: steel protective device, steel-PUF protective device, and steel-PUF-GFRP protective device; and the recommended type of protective device is determined based on the axial length of the protective device.

6. The protective device optimization design method according to claim 5, characterized in that: Calculating the protective performance of the protective device under the ship collision condition in step S5 includes: using the DME method to evaluate the protective performance of the protective device under the ship collision condition under all similarity ratio coefficients.

7. The protective device optimization design method according to claim 5, characterized in that: Returning the similarity ratio coefficient set to perform the next round of iterative calculation includes: reselecting a similarity ratio coefficient in the similarity ratio coefficient set, and generating a set of corresponding basic design parameters for calculation.

8. A computer-readable storage medium, characterized in that A computer program is stored, and when the computer program is executed by a processor, a protective device optimization design method according to any one of claims 1 to 7 is implemented.

Images