Free node coordinate prediction method and system of cable net structure

Abstract

The invention provides a method and a system for predicting coordinates of free nodes of a cable net structure. The method comprises the following steps: acquiring prestress distribution of the cable net structure, coordinates of each boundary node and an optimized direction value of each free node; inputting each boundary node coordinate into a deep learning shape finding model, and outputting a coordinate prediction value of each free node; and calculating a loss function value of the deep learning form-finding model, adjusting parameters of the deep learning form-finding model to update the deep learning form-finding model until the loss function value is smaller than a set error, and obtaining a coordinate predicted value of each free node of the cable net structure. According to the method, the boundary node coordinates of the cable net structure serve as the input of the deep learning shape finding model, the optimized direction values of the free nodes are given, the deep learning shape finding model is used for predicting the free node coordinates of the cable net structure, the purpose of cable net structure shape finding is achieved, and compared with existing manual force finding, the calculation efficiency is high, and rapidness and accuracy are achieved.

Description

Technical Field

[0001] The present invention relates to the field of architectural design, and more specifically, to a method and system for predicting free node coordinates of a cable net structure. Background Art

[0002] Cable net structures have been widely used in large public buildings, showing novel, beautiful and lightweight effects, and realizing the reasonable unity of force and shape. Due to the existence of prestress in the steel cables, the geometric shape of the cable net structure will change greatly from the zero stress state to the initial prestress state or load state, which has a great influence on its structural force performance. Therefore, it is very important to determine the initial prestress state of the cable net structure, which is the basis for the subsequent design and calculation analysis of the structure.

[0003] At present, the form-finding methods for cable-net structures include force density method, dynamic relaxation method, nonlinear finite element method, analytical method, etc. These methods have achieved good results and have been applied in many practical projects.

[0004] However, in the above methods, either the mathematical theory is complicated and difficult for structural engineers to master, or the stiffness matrix is ​​prone to singular convergence difficulties or requires repeated debugging to determine the appropriate initial value. The debugging process is cumbersome and time-consuming, and the cable net structure form-finding efficiency is low. Therefore, it is necessary to propose a fast and easy-to-apply form-finding method for cable net structures. Summary of the invention

[0005] The present invention aims at the technical problems existing in the prior art and provides a free node coordinate prediction method and system for a cable net structure, which overcomes the problems of cumbersome and time-consuming debugging process and low form-finding efficiency of the cable net structure.

[0006] According to a first aspect of the present invention, a method for predicting coordinates of free nodes of a cable net structure is provided, comprising:

[0007] Step 1, obtaining the prestress distribution of the cable net structure and the coordinates of each boundary node, and according to the force principle of the cable net structure, giving the optimized direction value of each free node of the cable net structure;

[0008] Step 2, setting the initial parameters of the deep learning form-finding model;

[0009] Step 3, inputting the coordinates of each boundary node into the deep learning form-finding model, and outputting the predicted value of each free node coordinate;

[0010] Step 4, calculating the loss function value of the deep learning form-finding model according to the coordinates of all boundary nodes, the optimized direction values ​​of all free nodes and the predicted values ​​of all free node coordinates;

[0011] Step 5: According to the loss function value, adjust the parameters of the deep learning form-finding model to update the deep learning form-finding model, repeat steps 2 to 5 until the loss function value is less than the set error, and obtain the predicted value of the coordinates of each free node of the cable net structure.

[0012] According to a second aspect of the present invention, there is provided a free node coordinate prediction system for a cable net structure, comprising:

[0013] An acquisition module is used to obtain the prestress distribution of the cable net structure and the coordinates of each boundary node, and to provide the optimized direction value of each free node of the cable net structure according to the force principle of the cable net structure;

[0014] The setup module is used to set the initial parameters of the deep learning form-finding model;

[0015] A prediction module, used for inputting the coordinates of each boundary node into the deep learning form-finding model and outputting a predicted value of each free node coordinate;

[0016] A calculation module, used to calculate the loss function value of the deep learning form-finding model according to all boundary node coordinates, all free node optimization direction values ​​and all free node coordinate prediction values;

[0017] The adjustment module is used to adjust the parameters of the deep learning form-finding model according to the loss function value to update the deep learning form-finding model, repeatedly call the prediction module and the calculation module until the loss function value is less than the set error, and obtain the predicted value of the coordinates of each free node of the cable net structure.

[0018] According to a third aspect of the present invention, there is provided an electronic device comprising a memory and a processor, wherein the processor is used to implement the steps of a free node coordinate prediction method of a cable net structure when executing a computer management program stored in the memory.

[0019] According to a fourth aspect of the present invention, there is provided a computer-readable storage medium having a computer management program stored thereon, which, when executed by a processor, implements the steps of a method for predicting free node coordinates of a cable-net structure.

[0020] The present invention provides a method and system for predicting the coordinates of free nodes of a cable net structure, which obtains the prestress distribution of the cable net structure and the coordinates of each boundary node, and gives the optimized direction value of each free node according to the force principle of the cable net structure; inputs each boundary node coordinate into a deep learning form-finding model, and outputs the predicted value of each free node coordinate; calculates the loss function value of the deep learning form-finding model, adjusts the parameters of the deep learning form-finding model, and updates the deep learning form-finding model until the loss function value is less than the set error, and obtains the predicted value of each free node coordinate of the cable net structure. The present invention uses the boundary node coordinates of the cable net structure as the input of the deep learning form-finding model, and gives the optimized direction value of the free node, and uses the deep learning form-finding model to predict the free node coordinates of the cable net structure, so as to achieve the purpose of cable net structure form-finding. Compared with the existing manual force-finding, the calculation efficiency is high, and it is fast and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A flow chart of a method for predicting free node coordinates of a cable net structure provided by the present invention;

[0022] Figure 2 Schematic diagram of the working principle of the deep learning shape-finding model;

[0023] Figure 3 This is a schematic diagram of the chain rule;

[0024] Figure 4 It is a schematic diagram of the orthogonal cable net structure;

[0025] Figure 5 This is a schematic diagram of the form-finding results of the orthogonal cable net structure;

[0026] Figure 6 It is a schematic diagram of the oblique cable net structure;

[0027] Figure 7 This is a schematic diagram of the form-finding results of the oblique cable net structure;

[0028] Figure 8 It is a schematic diagram of the irregular cable net structure;

[0029] Fig. 9 This is a schematic diagram of the form-finding results of the irregular cable net structure;

[0030] Figure 10-1 It is a schematic diagram of the finite element calculation results of the cable net stress of the irregular cable net structure;

[0031] Figure 10-2 It is a schematic diagram of the finite element calculation results of the cable net displacement of the irregular cable net structure;

[0032] Fig.11A schematic diagram of the structure of a free node coordinate prediction system for a cable net structure provided by the present invention;

[0033] Fig.12 A schematic diagram of the hardware structure of a possible electronic device provided by the present invention;

[0034] Fig.13 A schematic diagram of the hardware structure of a possible computer-readable storage medium provided by the present invention. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in 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 embodiments in 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. In addition, the technical features in the various embodiments or single embodiments provided by the present invention can be arbitrarily combined with each other to form a feasible technical solution. This combination is not subject to the constraints of the sequence of steps and / or the structural composition mode, but must be based on the ability of ordinary technicians in this field to achieve. When the combination of technical solutions is contradictory or cannot be achieved, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0036] Based on the background technology, in view of this, the present invention proposes a deep learning form-finding model for cable net structures driven by physical laws. The model solves the initial state of the cable net structure through the known prestress distribution of the cable net structure: the boundary node coordinates are used as the input layer, and the free node coordinates are used as the output layer. The known physical equilibrium equations of the cable prestress and the physical force-deformation principle are added to the loss function in the deep learning model, driving the deep learning model to reduce the loss function value through unsupervised training, and gradually approaching zero, thereby achieving the purpose of cable net structure form-finding.

[0037] Figure 1 The present invention provides a flow chart of a method for predicting the coordinates of free nodes in a cable network structure, which is written in Python and developed based on the Pytorch library. Figure 1 As shown, the specific steps are as follows:

[0038] Step 1: Obtain the prestress distribution of the cable net structure and the coordinates of each boundary node, and give the optimized direction value of each free node of the cable net structure according to the force principle of the cable net structure.

[0039] It is understandable that the purpose of the present invention is to solve the coordinates of the free nodes of the cable net structure when the prestress distribution of the cable net structure and the coordinates of each boundary node are known, and to perform form finding on the cable net structure. According to the prestress of the cable net structure and the force principle of the cable net structure, the optimized direction value of each free node can be given to control the moving direction of the free node.

[0040] Step 2: Set the initial parameters of the deep learning form-finding model.

[0041] Among them, see Figure 2 , which is a structural diagram of the deep learning form-finding model. The deep learning form-finding model includes an input layer, multiple hidden layers and an output layer. Then, first of all, setting the initial parameters of the deep learning form-finding model mainly includes setting the initial values ​​of the weight coefficient and bias coefficient of each layer of the deep learning form-finding model.

[0042] Step 3: Input the coordinates of each boundary node into the deep learning form-finding model, and output the predicted value of each free node coordinate.

[0043] It is understandable that the working principle of the deep learning form-finding model is: input the boundary node coordinates (x f ,y f ,z f ), passed to the hidden layer (N layers) and the output layer through the input layer, each layer of neurons receives the data transmitted by the neurons in the previous layer, and passes it to the next layer after weighted summation and activation processing until the output end outputs the data, that is, the predicted value of each free node coordinate.

[0044] The weighted summation and activation processing in the above deep learning form-finding model is implemented by the following formula:

[0045] y i =σ(w i .y i+1 +b i ), 1≤i≤n

[0046] z=w n+1 ·y n +b n+1

[0047] In the formula, i is the i-th hidden layer, n+1 is the output layer, and y i is the output value of the i-th hidden layer, z is the output value of the deep learning force model, σ() is the activation function, and w i and b i are the weight coefficients and bias coefficients of the i-th hidden layer, are the parameters of the deep learning shape-finding model, and w n+1 is the weight coefficient of the output layer, b n+1 is the bias coefficient of the output layer.

[0048] Step 4: Calculate the loss function value of the deep learning form-finding model according to all boundary node coordinates, all free node optimization direction values ​​and all free node coordinate prediction values.

[0049] It is understandable that after the deep learning form-finding model performs prediction, the loss function of the deep learning form-finding model is calculated. In the present invention, the loss function of the deep learning form-finding model mainly consists of two parts, one is the mean square residual of the physical equilibrium equation of the free node, and the other is the mean square error of the moving direction of the free node.

[0050] Among them, according to the predicted values ​​of all boundary node coordinates and all free node coordinates, the mean square residual MSE of the physical equilibrium equations in the x, y and z directions of all free nodes is calculated respectively: x , MSE y and MSE z , and calculate the mean square error MSE of the free node moving direction based on the optimized direction values ​​of all free nodes and the predicted coordinate values ​​of all free nodes m ; According to the mean square residual MSE x , MSE y and MSE z and mean square error MSE m , calculate the loss function value of the deep learning form-finding model.

[0051] Among them, the loss function of the deep learning form-finding model can be expressed as L total =MSE x +MSE y +MSE z +f(n)·MSE m .MSE x , MSE y and MSE z are the mean square residuals of the physical equilibrium equations in the x, y and z directions on the free nodes, respectively, and control the optimization target of the model. After the prestress is applied to the cable, according to the physical force deformation principle, the middle node will gradually move in a straight line toward the cable until it reaches a stable equilibrium state. Based on this, the deep learning form-finding model uses f(n) and MSE m As the exponential decay function of the training step and the mean square error of the free node moving direction, they control the optimization direction of the model. As the training steps increase, the proportion of the optimization direction term in the loss function value gradually decreases, and the optimization target term gradually dominates until it completely dominates, and finally a reasonable form that satisfies the equilibrium equation is obtained. The specific expression is as follows:

[0052]

[0053] L total =MSE x+MSE y +MSE z +f(n)MSE m

[0054] f(n)=e -αn

[0055] Among them, N s is the number of free nodes, m is the number of cable units connected to free node j, L jk and F jk are the cable unit length and cable force between free node j and node k, respectively, (x j ,y j ,z j ) is the coordinate of free node j, (x k ,y k ,z k ) is the coordinate of node k, (x j ,y j ) pred and (x j ,y j ) opdi are the predicted coordinate values ​​and optimized direction values ​​of the free node j in the x and y directions, respectively. f(n) is the exponential decay function of the training step, n is the number of training steps of the deep learning form-finding model, and α is the decay coefficient.

[0056] Step 5: According to the loss function value, adjust the parameters of the deep learning form-finding model to update the deep learning form-finding model, repeat steps 2 to 5 until the loss function value is less than the set error, and obtain the predicted value of the coordinates of each free node of the cable net structure.

[0057] It can be understood that after the loss function value of the deep learning form-finding model is calculated in step 4, if the loss function value is greater than the set error, the gradient of the loss function in the back propagation of the deep learning form-finding model to the weight coefficients and bias coefficients of each layer is calculated by the chain derivation rule. The chain derivation rule can be found in Figure 3 , update the weight coefficient and bias coefficient of the deep learning form-finding model through the optimization algorithm, re-execute steps 3 to 5, and continuously recalculate the loss function value of the deep learning form-finding model until the loss function is less than the set error. The data obtained from the output end is the free node coordinate (x s ,y s ,z s ).

[0058] The free node coordinate prediction method of the cable net structure provided by the present invention is described below with reference to several specific embodiments.

[0059] Example 1: Orthogonal cable net structure, node numbering as follows Figure 4As shown, the nodes numbered 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, and 12 are boundary node coordinates, and the nodes numbered 13, 14, 15, 16, 17, 18, 19, 20, and 21 are free nodes. The ratio of the transverse and longitudinal cable forces is F1:F2 = 1:1. The boundary nodes and free nodes are fixed on the saddle surface, and the equation is Z = (X 2 -Y 2 ) / 50, by Figure 3 Determine the boundary node coordinates (x f ,y f ,z f ) and the free node optimization direction value (x j ,y j ) opdi For this cable net structure, the calculation steps of the deep learning form-finding model are as follows:

[0060] (1) Input the boundary node coordinates (x f ,y f ,z f ), select optimizer Adam, learning rate 0.01, α = 0.01. Pass it to the hidden layer (6 layers) and output layer through the input layer. Each layer of neurons (32) receives the data transmitted by the neurons in the previous layer, and passes it to the next layer after weighted summation and activation (Tanh function) processing until the output data.

[0061] (2) After the data is output, the loss function is calculated. If the loss function value is greater than the set error, the gradient of the loss function to the weights and biases of each layer in the back propagation of the deep learning model is calculated by the chain rule, and the weights and biases in the model are updated and recalculated through the optimization algorithm.

[0062] (3) The deep learning model automatically repeats steps (1) and (2). After n = 1000 training steps, the loss function is less than the set error, the training stops, and the free node coordinates (x s ,y s ,z s ).

[0063] The shape-finding result is as follows Figure 5 As shown, it is compared with the theoretical results, as shown in Table 1. As can be seen from Table 1, the calculation results of the two methods are in good agreement.

[0064] Table 1 Comparison of form-finding results of free node Z coordinates of orthogonal cable net structures

[0065] serial number 13 14 15 16 17 18 19 20 21 Theoretical value (m) 0.000 -0.428 0.000 0.428 0.000 0.428 0.000 -0.428 0.000 Predicted value (m) 0.000 -0.438 0.000 0.438 0.000 0.438 0.000 -0.438 0.000 error(%) 0.0 2.3 0.0 2.3 0.0 2.3 0.0 2.3 0.0

[0066] Embodiment 2: oblique cable net structure, node numbering as follows Figure 6As shown, nodes numbered 1, 2, 3, 4 and 5 are boundary nodes, nodes numbered 6, 7, 8, 9 and 10 are free nodes, and the cable force ratio F1:F2:F3:F4:F5 = 1:1:1:1:1. The boundary nodes are fixed on the saddle surface, and their equation is Z = (X 2 -Y 2 ) / 50, by Figure 6 Determine the boundary node coordinates (x f ,y f ,z f ) and the free node optimization direction value (x j ,y j ) opdi The calculation steps are the same as those in the first embodiment.

[0067] The shape-finding result is as follows Figure 7 As shown in Table 2, the form-finding results obtained by the force density method are compared with those obtained by the force density method. As shown in Table 2, the calculation results of the two methods are in good agreement.

[0068] Table 2 Comparison of form-finding results of oblique cable net structure

[0069]

[0070]

[0071] Example 3: Irregular cable network structure, node numbering such as Figure 8 As shown, the nodes numbered 1, 2, 3, 4, 5, 6, 7, 8, 9 and 10 are boundary nodes, the nodes numbered 11, 12, 13, 14, 15 and 16 are free nodes, and the cable force ratio F1:F2:F3:F4:F5 = 1:1:1:1:1. Figure 8 Determine the boundary node coordinates (x f ,y f ,z f ) and the free node optimization direction value (x j ,y j ) opdi The calculation steps are the same as those in the first embodiment.

[0072] The shape-finding result is as follows Fig. 9 As shown, the preset elastic modulus of the cable net is 200GPa and the cross section is 1000mm 2 The cable prestressing force is 200 MPa. The finite element calculation results can be found in Figure 10-1 and Figure 10-2 ,in, Figure 10-1 is the finite element calculation result of cable net stress, Figure 10-2is the finite element calculation result of the cable net displacement. It is compared with the form-finding result of the model of the present invention, as shown in Table 3. It can be seen from Table 3 that the calculation results of the two methods are in good agreement.

[0073] Table 3 Comparison of form-finding results of irregular cable net structures

[0074]

[0075]

[0076] The above three cases all show that the form-finding results of the model of the present invention are in good agreement with the calculation results of the theoretical solution, the line density method and the finite element method, with high accuracy, and the time used does not exceed 1 minute, with high calculation efficiency.

[0077] See also Fig.11 , a free node coordinate prediction system of a cable net structure of the present invention is provided, the system comprising:

[0078] The acquisition module 1101 is used to obtain the prestress distribution of the cable net structure and the coordinates of each boundary node, and to provide the optimized direction value of each free node of the cable net structure according to the force principle of the cable net structure;

[0079] A setting module 1102 is used to set initial parameters of the deep learning form-finding model;

[0080] A prediction module, used for inputting the coordinates of each boundary node into the deep learning form-finding model and outputting a predicted value of each free node coordinate;

[0081] A calculation module 1103 is used to calculate the loss function value of the deep learning form-finding model according to all boundary node coordinates, all free node optimization direction values ​​and all free node coordinate prediction values;

[0082] The adjustment module 1104 is used to adjust the parameters of the deep learning form-finding model according to the loss function value to update the deep learning form-finding model, repeatedly call the prediction module and the calculation module until the loss function value is less than the set error, and obtain the predicted value of the coordinates of each free node of the cable net structure.

[0083] It can be understood that the free node coordinate prediction system of the cable net structure provided by the present invention corresponds to the free node coordinate prediction method of the cable net structure provided by the aforementioned embodiments, and the relevant technical features of the free node coordinate prediction system of the cable net structure can refer to the relevant technical features of the free node coordinate prediction method of the cable net structure, which will not be repeated here.

[0084] See also Fig.12 , Fig.12Schematic diagram of an electronic device provided by an embodiment of the present invention. Fig.12 As shown, an embodiment of the present invention provides an electronic device 1200, including a memory 1210, a processor 1220, and a computer program 1211 stored in the memory 1210 and executable on the processor 1220. When the processor 1220 executes the computer program 1211, the steps of the free node coordinate prediction method of the cable net structure are implemented.

[0085] See also Fig.13 , Fig.13 Schematic diagram of an embodiment of a computer-readable storage medium provided by the present invention. Fig.13 As shown, this embodiment provides a computer-readable storage medium 1300 on which a computer program 1311 is stored. When the computer program 1311 is executed by a processor, the steps of the free node coordinate prediction method of the cable net structure are implemented.

[0086] The embodiment of the present invention provides a method and system for predicting the coordinates of free nodes of a cable net structure, which obtains the prestress distribution of the cable net structure and the coordinates of each boundary node, and gives the optimized direction value of each free node according to the force principle of the cable net structure; inputs each boundary node coordinate into a deep learning form-finding model, and outputs the predicted value of each free node coordinate; calculates the loss function value of the deep learning form-finding model, adjusts the parameters of the deep learning form-finding model, and updates the deep learning form-finding model until the loss function value is less than the set error, and obtains the predicted value of each free node coordinate of the cable net structure. The present invention uses the boundary node coordinates of the cable net structure as the input of the deep learning form-finding model, and gives the optimized direction value of the free node, and uses the deep learning form-finding model to predict the free node coordinates of the cable net structure, so as to achieve the purpose of cable net structure form-finding. Compared with the existing manual force-finding, the calculation efficiency is high, and it is fast and accurate.

[0087] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0088] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0089] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0090] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0091] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0092] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0093] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for predicting free node coordinates of a cable net structure, characterized in that: include: Step 1, obtaining the prestress distribution of the cable net structure and the coordinates of each boundary node, and according to the force principle of the cable net structure, giving the optimized direction value of each free node of the cable net structure; Step 2, setting the initial parameters of the deep learning form-finding model; Step 3, inputting the coordinates of each boundary node into the deep learning form-finding model, and outputting the predicted value of each free node coordinate; Step 4, calculating the loss function value of the deep learning form-finding model according to the coordinates of all boundary nodes, the optimized direction values ​​of all free nodes and the predicted values ​​of all free node coordinates; Step 5: According to the loss function value, adjust the parameters of the deep learning form-finding model to update the deep learning form-finding model, repeat steps 2 to 5 until the loss function value is less than the set error, and obtain the predicted value of the coordinates of each free node of the cable net structure.

2. The free node coordinate prediction method of the cable net structure according to claim 1 is characterized in that: The deep learning form-finding model includes an input layer, a plurality of hidden layers and an output layer, including: Each of the input boundary node coordinates is passed to multiple hidden layers through the input layer, and the neurons in each hidden layer receive the data transmitted by the neurons in the previous hidden layer, and pass them to the neurons in the next hidden layer after weighted summation and activation processing. After passing through multiple hidden layers, the predicted value of each free node coordinate is output through the output layer.

3. The free node coordinate prediction method of the cable net structure according to claim 1 is characterized in that: The step 2, setting the initial parameters of the deep learning form-finding model, includes: Set the initial weight coefficients and initial bias coefficients for each layer of the deep learning form-finding model.

4. The free node coordinate prediction method of the cable net structure according to claim 2, characterized in that: The weighted summation and activation processing of each hidden layer includes: y i =σ(w i .y i+1 +b i ),1≤i≤n z=w n+1 ·y n +b n+1 In the formula, i is the i-th hidden layer, n+1 is the output layer, and y i is the output value of the i-th hidden layer, z is the output value of the deep learning force model, σ() is the activation function, and w i and b i are the weight coefficients and bias coefficients of the i-th hidden layer, are the parameters of the deep learning shape-finding model, and w n+1 is the weight coefficient of the output layer, b n+1 is the bias coefficient of the output layer.

5. The free node coordinate prediction method of the cable net structure according to claim 1, characterized in that: The step 4, calculating the loss function value of the deep learning form-finding model according to all the boundary node coordinates, all the free node optimization direction values ​​and all the free node coordinate prediction values, includes: According to the predicted values ​​of all boundary node coordinates and all free node coordinates, the mean square residual MSE of the physical equilibrium equations in the x, y and z directions of all free nodes is calculated respectively. x , MSE y and MSE z , and calculate the mean square error MSE of the free node moving direction based on the optimized direction values ​​of all free nodes and the predicted coordinate values ​​of all free nodes m ; According to the mean square residual MSE x , MSE y and MSE z And mean square error MSE m , calculate the loss function value of the deep learning form-finding model.

6. The free node coordinate prediction method of the cable net structure according to claim 5, characterized in that: According to the predicted values ​​of all boundary node coordinates and all free node coordinates, the mean square residual MSE of the physical equilibrium equations in the x, y and z directions of all free nodes is calculated respectively. x , MSE y and MSE z , include: The mean square error (MSE) of the moving direction of the free nodes is calculated based on the optimized direction values ​​of all free nodes and the predicted coordinate values ​​of all free nodes. m ,include: According to the mean square residual MSE x , MSE y and MSE z and mean square error MSE m , calculating the loss function value of the deep learning form-finding model, including: L total =MSE x +MSE y +MSE z +f(n)MSE m ; f(n)=e -αn Among them, N s is the number of free nodes, m is the number of cable units connected to free node j, L jk and F jk are the cable unit length and cable force between free node j and node k, respectively, (x j ,y j ,z j ) is the coordinate of free node j, (x k ,y k ,z k ) is the coordinate of node k, (x j ,y j ) pred and (x j ,y j ) opdi are the predicted coordinate values ​​and optimized direction values ​​of the free node j in the x and y directions, respectively. f(n) is the exponential decay function of the training step, n is the number of training steps of the deep learning form-finding model, and α is the decay coefficient.

7. The free node coordinate prediction method of the cable net structure according to claim 1, characterized in that: The step 5, adjusting the parameters of the deep learning form-finding model according to the loss function value to update the deep learning form-finding model, includes: If the loss function value is greater than the set error range, the gradient of the loss function to the weight coefficient and bias coefficient of each layer in the back propagation of the deep learning form-finding model is calculated by the chain rule, and the weight coefficient and bias coefficient of the deep learning form-finding model are updated by the optimization algorithm.

8. A free node coordinate prediction system for a cable net structure, characterized in that: include: An acquisition module is used to obtain the prestress distribution of the cable net structure and the coordinates of each boundary node, and to provide the optimized direction value of each free node of the cable net structure according to the force principle of the cable net structure; The setup module is used to set the initial parameters of the deep learning form-finding model; A prediction module, used for inputting the coordinates of each boundary node into the deep learning form-finding model and outputting a predicted value of each free node coordinate; A calculation module, used to calculate the loss function value of the deep learning form-finding model according to all boundary node coordinates, all free node optimization direction values ​​and all free node coordinate prediction values; The adjustment module is used to adjust the parameters of the deep learning form-finding model according to the loss function value to update the deep learning form-finding model, repeatedly call the prediction module and the calculation module until the loss function value is less than the set error, and obtain the predicted value of the coordinates of each free node of the cable net structure.

9. An electronic device, characterized in that: It comprises a memory and a processor, wherein the processor is used to implement the free node coordinate prediction method of the cable net structure as described in any one of claims 1 to 7 when executing the computer management program stored in the memory.

10. A computer-readable storage medium having a computer management program stored thereon, characterized in that: The computer management program is executed by the processor to implement the free node coordinate prediction method of the cable net structure described in any one of claims 1-7.

Images