Inverse prediction method and device for terminal deformation of positioning structure and electronic equipment

Abstract

The application provides an inversion prediction method for terminal deformation of a positioning structure, comprising: using a boundary condition sample data set to perform finite element analysis on a positioning structure simulation model to obtain a boundary load-target point displacement sample data set; using the boundary load-target point displacement sample data set to train and verify a constructed neural network model to obtain a prediction model; inputting a set boundary load condition into the prediction model in a forward direction to obtain a target point displacement prediction value; using the prediction model to perform reverse prediction according to the target point displacement prediction value and taking target point actual displacement data as a constraint condition to obtain a predicted displacement value of the terminal of the positioning structure. For an actual measurement model of the positioning structure and a target point position, a finite element analysis and a neural network model are used to construct a target point displacement data set; based on the target point displacement data set, reverse operation is performed through the trained neural network model, so that inversion prediction of terminal displacement of the positioning structure is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tooling structure deformation measurement, in particular to a positioning structure terminal deformation inversion prediction method and device and electronic equipment. BACKGROUND

[0002] In the aircraft assembly process, the positioning structure is directly connected with the aircraft and bears the load as the tooling, which is a key positioning element in the aircraft assembly process. Most of the positioning structures used in aircraft assembly are in the form of cantilever, including the constraint end and the force end. At the constraint end of the positioning structure, constraints are formed by using holes, pins, lugs, etc.; at the force end of the positioning structure, it is connected with the aircraft and bears the load. In the aircraft assembly process, the positioning structure inevitably deforms under force, and the deformation result directly determines the appearance and position of the aircraft parts connected therewith. Therefore, the deformation of the positioning structure has a great influence on the assembly accuracy of the aircraft. SUMMARY

[0003] In order to solve the positioning structure deformation measurement problem under the condition that the local positioning structure is not visible, the present application provides a positioning structure terminal deformation inversion prediction method, which comprises the following steps:

[0004] Performing finite element analysis on the positioning structure simulation model by using the boundary condition sample data set to obtain a boundary load-target point displacement sample data set;

[0005] Training and verifying the constructed neural network model by using the boundary load-target point displacement sample data set to obtain a prediction model;

[0006] Forwardly inputting the set boundary load condition into the prediction model to obtain a target point displacement prediction value;

[0007] According to the target point displacement prediction value, performing reverse prediction by using the prediction model under the constraint condition of the target point actual displacement data to obtain a predicted displacement value of the positioning structure terminal.

[0008] In some embodiments of the present application, the method further comprises: obtaining the boundary condition sample data set by using a test design method based on Latin hypercube sampling.

[0009] In some embodiments of the present application, the method further comprises: obtaining the positioning structure actual measurement model and the target point actual displacement data by using laser measurement.

[0010] In some embodiments of the present application, the method further comprises: constructing a positioning structure simulation model based on the positioning structure actual measurement model.

[0011] In some embodiments of the present application, the method further comprises: using tetrahedral mesh units to mesh the positioning structure actual model.

[0012] In some embodiments of the present application, the neural network model comprises a multi-layer perceptron neural network model.

[0013] In some embodiments of the present application, the boundary condition sample data set comprises 6 boundary condition values F x , F y , F z , M x , M y , M z , wherein F x , F y , F z respectively correspond to forces on the x, y, z axes, and M x , M y , M z respectively correspond to bending moment loads on the x, y, z axes.

[0014] According to another aspect of the present application, there is also provided an inversion prediction device for terminal deformation of a positioning structure, the device comprising:

[0015] A sample data set acquisition module is configured to perform finite element analysis on a positioning structure simulation model using a boundary condition sample data set, and to acquire a boundary load-target point displacement sample data set.

[0016] A prediction model construction module is configured to train and verify a constructed neural network model using the boundary load-target point displacement data set, and to obtain a prediction model.

[0017] A target point displacement prediction module is configured to input a set boundary load condition forward into the prediction model, and to obtain a target point displacement prediction value.

[0018] A terminal displacement inverse prediction module is configured to use the target point actual displacement data as a constraint condition, to perform inverse prediction using the prediction model according to the target point displacement prediction value, and to obtain a predicted displacement value of the terminal of the positioning structure.

[0019] In some embodiments of the present application, the device further comprises:

[0020] An actual data acquisition module is configured to acquire a positioning structure actual model and the target point actual displacement data using laser measurement.

[0021] A boundary condition acquisition module is configured to acquire the boundary condition sample data set using a DOE method.

[0022] The simulation model construction module is configured to construct the simulation model of the positioning structure based on the measured model of the positioning structure.

[0023] According to another aspect of the present application, an electronic device for inversion prediction of terminal deformation of a positioning structure is also provided, comprising:

[0024] one or more processors;

[0025] a storage device configured to store one or more programs;

[0026] When the one or more programs are executed by the one or more processors, the one or more processors implement the above method.

[0027] The inversion prediction method for terminal deformation of a positioning structure provided by the present application is used for a measured model of a positioning structure, and is used for a target point position marked on the positioning structure. Finite element analysis and a neural network model are used to construct a target point displacement data set. Based on the constructed target point displacement data set, a trained neural network model is used for reverse operation, so as to realize inversion prediction of terminal displacement of the positioning structure, and solve the problem of deformation measurement of invisible parts. It is verified that the predicted terminal displacement value is within the allowable error range compared with the actual error value. BRIEF DESCRIPTION OF DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained according to these drawings without departing from the scope of the present application.

[0029] Figure 1 A positioning structure schematic diagram according to an example embodiment of the present application is shown;

[0030] Figure 2 A positioning structure terminal deformation prediction process schematic diagram according to an example embodiment of the present application is shown;

[0031] Figure 3 A coordinate system fusion process schematic diagram according to an example embodiment of the present application is shown;

[0032] Figure 4 An inversion prediction method flow chart of terminal deformation of a positioning structure according to a first example embodiment of the present application is shown;

[0033] Figure 5 A target point displacement forward prediction model schematic diagram based on an example embodiment of the present application is shown;

[0034] Figure 6A flow chart of an inverse prediction method of positioning structure terminal deformation according to a second example embodiment of the present application is shown.

[0035] Figure 7 A block diagram of an inverse prediction device of positioning structure terminal deformation according to a first example embodiment of the present application is shown.

[0036] Figure 8 A block diagram of an inverse prediction device of positioning structure terminal deformation according to a second example embodiment of the present application is shown.

[0037] Figure 9 A block diagram of an electronic device according to an example embodiment of the present application is shown. DETAILED DESCRIPTION

[0038] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0039] The terms "first", "second", and the like in the present application are used to distinguish different objects, rather than to describe a predetermined order. In addition, the terms "include" and "have" in the present application and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but can optionally include other steps or units not listed or can optionally include other steps or units inherent to the process, method, product or device.

[0040] Reference to "an embodiment" herein means that a predetermined feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the present application. The phrase appears at various places in the specification does not necessarily all refer to the same embodiment, nor is it necessarily independent or alternative embodiments to other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0041] At present, the measurement system based on visual measurement is suitable for three-dimensional space point or size detection, especially for large workpiece profile detection, due to its simple structure, easy to move, fast data acquisition, easy to operate, low measurement cost, and potential for online real-time three-dimensional measurement, etc. For example, laser scanning measurement, close-range photogrammetry, etc. are used for visual-based assembly deformation measurement.

[0042] However, in the aircraft assembly site, due to the structural characteristics of the assembly tool and the influence of the assembly environment, the key feature parts of the positioning structure cannot be directly visually measured (for example, blocked). The blocked key features (such as the terminal connection joint) determine the assembly accuracy of the overall structure. Therefore, the existing visual measurement method for measuring the deformation of the positioning structure cannot be applied to the case where the positioning structure is partially invisible.

[0043] In order to solve the problem of measuring the deformation of the positioning structure in the case of partial invisibility of the positioning structure, the present application provides a method for predicting the deformation of the terminal of the positioning structure. For the measured model of the positioning structure, for the target point position marked on the positioning structure, a target point displacement data set is constructed by using finite element analysis and a neural network model; based on the constructed target point displacement data set, the trained neural network model is used for reverse operation, so as to realize the inversion prediction of the terminal displacement of the positioning structure.

[0044] The technical solutions of the present application will be described in detail below with reference to the drawings.

[0045] Figure 1 A positioning structure according to an example embodiment of the present application is shown. In the following, the positioning structure in Figure 1 will be taken as an example to introduce in detail the method for predicting the deformation of the terminal of the positioning structure provided by the present application.

[0046] Referring to Figure 1 , the positioning structure in the figure is a typical positioning structure in the process of assembling the aircraft component tool, and the outer size is about 900mmx700mmx170mm. In the assembly process, the aircraft component and its assembly jig are connected to the support base through the positioning structure. When the deformation of the positioning structure under the load is too large, it will inevitably cause the attitude of the aircraft assembly jig to change, further affecting the assembly of the aircraft component. However, due to the blocking of the connecting terminal of the positioning structure by the aircraft wing and other parts, it cannot be directly visually measured.

[0047] In order to determine the deformation of the connecting terminal of the positioning structure, the present application provides a method for predicting the displacement of the terminal of the positioning structure, which predicts the displacement of the terminal to determine the deformation, thereby solving the problem of being unable to directly visually measure.

[0048] Figure 2 A schematic diagram of the terminal deformation prediction process of the positioning structure according to an example embodiment of the present application is shown. Referring to Figure 2 , the terminal deformation prediction process of the positioning structure provided by the present application includes the following contents.

[0049] First, the measured model of the positioning structure and the measured displacement data of the identified target point are obtained.

[0050] For the positioning structure to be measured, before loading, initial measurement data of the positioning structure can be obtained by measurement, including the three-dimensional model (e.g., point cloud data) of the positioning structure before loading and the three-dimensional coordinates of the target points. For example, the three-dimensional data of the positioning structure to be measured can be obtained by scanning the positioning structure by a laser scanner; and after reconstruction, the measured model of the positioning structure can be obtained.

[0051] After loading, the positioning structure is deformed; and after measurement again, the measurement data after loading can be obtained, including the three-dimensional model of the deformed positioning structure and the three-dimensional coordinates of the target points. The three-dimensional model of the positioning structure and the three-dimensional coordinates of the target points obtained by measurement before and after loading can be fused and calculated to obtain the measured deformation data of the target points on the positioning structure, including the relative positions of the target points, the three-dimensional displacement deformation, and the like.

[0052] In actual operation, the target points can be marked (e.g., special marks are pasted) on the surface of the positioning structure according to the requirements of the laser scanner measurement, and the target points can cover as many key areas as possible. In the example embodiments of the present application, the key areas can be the terminal, free end, and the like of the positioning structure. In the case of using a handheld laser scanner, after matching the handheld laser scanner with the computer software, the operator can hold the instrument to scan all the target points that can be scanned on the positioning structure to be measured; the obtained data can be transmitted into the software system to match the three-dimensional coordinates, and then the approximate coordinate range of the positioning structure can be obtained. In order to obtain more complete scanning data, the operator can hold the instrument to scan the positioning structure and the target points in all directions, and different areas can be scanned by switching multiple scanning modes; for example, the line scanning mode is switched to scan the narrow gap area of the positioning structure, so that complete point cloud data can be obtained. Based on the obtained point cloud data, after selecting the reference coordinate system, the measured three-dimensional model of the positioning structure before loading and deformation can be obtained by three-dimensional reconstruction. In this process, for the part of the model that lacks details, the point cloud data can be supplemented by rescan and three-dimensional reconstruction. After the loading and deformation test of the positioning structure, the above operation steps can be repeated to obtain the measured three-dimensional model after loading and deformation. According to the reference coordinate system, the coordinate systems of the two models can be fused and matched, so that the spatial positions and loading deformation of the target points can be obtained.

[0053] Figure 3 The coordinate system fusion process according to the example embodiments of the present application is shown. As shown in FIG. 1, the coordinate system fusion process includes the following steps. Figure 3As shown, the positioning structure three-dimensional data obtained before loading can be reconstructed in a first laser scanning coordinate system; the positioning structure three-dimensional data obtained after loading can be reconstructed in a second laser scanning coordinate system; the first laser scanning coordinate system and the second laser scanning coordinate system are changed to an intermediate coordinate system; the scanning data before loading is transformed into a first laser scanning coordinate in the intermediate coordinate system, the scanning data after loading is transformed into a second laser scanning coordinate in the intermediate coordinate system, and coordinate system matching and fusion are performed, so that the spatial position of the target point and the loading deformation can be obtained.

[0054] Next, the DOE method is used to obtain the boundary condition sample data set.

[0055] In the example embodiments of the present application, the design of experiment (DOE) based on Latin hypercube sampling can be used to obtain the boundary condition sample data set. For each individual boundary condition of the n boundary conditions, the original sample space is divided into M parts, and a sample is randomly selected in each part to ensure the uniformity of the original sample; then N groups of different boundary condition combinations are generated by orthogonal experimental design of the n boundary conditions.

[0056] Next, based on the reconstructed positioning structure measured model, a positioning structure simulation model is constructed.

[0057] In the example embodiments of the present application, based on the reconstructed positioning structure measured model, the positioning structure simulation model is constructed by defining material properties, meshing, boundary condition setting, etc. For example, the positioning structure simulation model can be constructed in a finite element analysis software.

[0058] For example, the positioning structure measured model measured by the laser scanner can be imported into the finite element analysis software; if there is a loss of geometric elements, the positioning structure measured model can be repaired in the local area. According to the actual material of the positioning structure, the material properties and the cross section can be defined. In the example embodiments of the present application, tetrahedral mesh elements can be used to mesh the positioning structure to avoid mesh distortion and ensure the convergence of the calculation. During the meshing process, fine meshing can be performed at the chamfer and the opening, and coarse meshing can be performed at other parts to improve the calculation efficiency. In the example embodiments of the present application, for the constructed positioning structure simulation model, an analysis step can be set, the maximum incremental step can be defined as 10000, the initial analysis step is 0.01, the minimum analysis step is 10E-10, and the maximum analysis step is 1. During the analysis step setting process, a node set of the target point region can also be established, such as setting stress (S), strain (E), displacement (U) and other variables for the target point region. According to the actual loading condition of the positioning structure, the measured positioning structure measured model is set with corresponding boundary conditions, and the positioning structure simulation model can be obtained.

[0059] Next, the boundary condition sample data set is used to perform finite element analysis on the positioning structure simulation model to obtain a boundary load-target point displacement sample data set.

[0060] In the example embodiments of the present application, the boundary condition sample data set obtained by the DOE method is used to perform finite element analysis on the constructed positioning structure simulation model, and different boundary condition loads and corresponding target point displacement data sets can be obtained.

[0061] In the example embodiments of the present application, 2000 different boundary load conditions obtained by the DOE method can be loaded onto the simulation model of the positioning structure in the MATLAB simulation analysis software; through loop iteration, 2000 groups of simulation results corresponding to different boundary condition combinations can be obtained.

[0062] Next, the boundary load-target point displacement data is used to train the constructed neural network model to obtain a prediction model.

[0063] In the example embodiments of the present application, a neural network model for predicting displacement can be established according to a multi-layer perception mechanism. For example, the parameters of the neural network model can be set as follows:

[0064] For the first layer of neurons, the number can be set to 128 neurons, and the input value can be a vector with a length of 3n+6, which can be expressed as an input vector X:

[0065] X = (x1, x2, …, xn, x1, x2, …, xn) 3n+6 )

[0066] where 6 represents 6 boundary condition constraints, n represents the number of target points, and x1, x2, … represent input features.

[0067] For each neuron j in the first layer, the output O j can be expressed using the activation function ReLU as:

[0068]

[0069] where w ij is the weight of the i-th input feature to the j-th neuron, and b j is the bias of the j-th neuron.

[0070] For each neuron in the output layer neuron, the output O k can be related to the output of the first layer of neurons and expressed as:

[0071]

[0072] where w ij is the weight of the jth neuron to the kth neuron, b k is the bias of the kth neuron.

[0073] On the basis of the above definitions, the mean square error can be used as the loss function, and the stochastic gradient descent optimization algorithm adam can be selected as the optimizer. In this way, the parameter setting of the neural network model is completed.

[0074] Next, the boundary load-target point displacement dataset is obtained as the training data set and the test data set, and the defined neural network model is trained to obtain a prediction model for predicting the target point displacement.

[0075] In the example embodiments of the present application, 80% of the boundary load-target point displacement dataset can be passed to the neural network model as training data, and the fit method can be used for training. For example, the number of iterations of training can be set to 50 rounds, and the number of samples in each batch is 16. In the example embodiments of the present application, 20% of the boundary load-target point displacement dataset can be used as a validation set to detect model performance. For example, 20% of the boundary load-target point displacement dataset is used as test data to calculate the loss value of the model to evaluate the performance of the model. After training and testing, a prediction model for predicting the target point displacement under the condition of the set boundary load can be obtained.

[0076] Then, the set boundary load condition is input into the prediction model in the forward direction to obtain the target point displacement prediction value. In the example embodiments of the present application, the set boundary load condition can be input into the prediction model in the forward direction to obtain the target point displacement prediction value.

[0077] Finally, the target point measured displacement data is used as a constraint condition, and the prediction model is used to perform backward prediction according to the output target point displacement prediction value to obtain the predicted displacement value of the positioning structure terminal.

[0078] In the example embodiments of the present application, the prediction model and the output target point displacement prediction value thereof can be used to obtain the displacement prediction value of the positioning structure terminal through inversion difference optimization. For example, the obtained target point measured displacement data can be used as a target function, and the displacement of the positioning structure terminal on the forward input side of the prediction model (or the boundary load) can be continuously adjusted to make the output result of the prediction model (i.e., the target point displacement prediction value) close to the target function. When the output result of the prediction model reaches the condition of the target function, the displacement value of the positioning structure terminal on the input side of the prediction model is the terminal displacement prediction value obtained through inversion prediction. Through the terminal displacement value, the deformation amount after the terminal loading can be determined.

[0079] Figure 4A flowchart of an inversion prediction method of positioning structure terminal deformation according to the first example embodiment of the present application is shown.

[0080] According to the above general process, according to the first example embodiment of the present application, an inversion prediction method of positioning structure terminal deformation is provided. As shown in the figure, the method comprises the following steps. Figure 4

[0081] In step S410, the boundary condition sample data set is used to perform finite element analysis on the positioning structure simulation model to obtain a boundary load-target point displacement sample data set.

[0082] In the example embodiment of the present application, the boundary condition sample data set is used to perform finite element analysis on the constructed positioning structure simulation model, and different boundary condition loads and corresponding target point displacement data sets can be obtained. In the example embodiment of the present application, the 2000 different boundary load conditions obtained can be loaded onto the simulation model of the positioning structure in the MATLAB simulation analysis software; through loop iteration, 2000 groups of simulation results corresponding to different boundary condition combinations are obtained.

[0083] For example, the listdir function in the MATLAB environment can be used to traverse the simulation model file, and then the mdb.JobFromInputFile function command is used to generate a calculation file and set the cpu number, calculation accuracy and other parameters therein; thereafter, the submit command is used to submit the work task. In order to avoid excessive calculation memory caused by simultaneous calculation of multiple files, the db.jobs[subjob].waitForCompletion command is used to interrupt the execution of the program. After the completion of the calculation task, the batch submission and calculation of the 2000 groups of simulation model files corresponding to different boundary condition combinations are completed.

[0084] After batch submission of the simulation and obtaining of the finite element analysis results, the listdir function can be used to traverse all the simulation result files in the simulation result folder, and then the getSubset function is used to obtain the displacement of the target point, and finally the loop is used to batch extract all the simulation result files in the folder to obtain 2000 groups of different boundary loads and corresponding target point displacements. Part of the boundary load-target point (or target point region, such as cross section) displacement data set obtained in the example embodiment of the present application is shown in Table 1.

[0085] Table 1 Boundary load-target point displacement data sample

[0086]

[0087] ​In step S420, the constructed neural network model is trained and verified using the boundary load-target point displacement data to obtain a prediction model.

[0088] In an exemplary embodiment of the present application, a neural network model for predicting displacement can be established based on a multi-layer perception mechanism. For example, the parameters of the neural network model can be set as follows:

[0089] For the first layer of neurons, the number can be set to 128 neurons, and its input value can be a vector of length 3n+6, which can be expressed as input vector X:

[0090] X=(x1,x2,…,x 3n+6 )

[0091] Among them, the number 6 represents 6 boundary condition constraints, n represents the number of target points, and X1, x2,… represent input features.

[0092] For each neuron j in the first layer, its output O j It can be expressed using the activation function ReLU as:

[0093]

[0094] Among them, w ij is the weight of the i-th input feature to the j-th neuron, b j is the bias of the jth neuron.

[0095] For each neuron in the output layer, a linear activation function can be used to output it O k It is connected to the output of the first layer of neurons and expressed as:

[0096]

[0097] Among them, w ij is the weight from the jth neuron to the kth neuron, b k is the bias of the kth neuron.

[0098] Based on the above definition, we can use mean square error as the loss function and select the stochastic gradient descent optimization algorithm adam as the optimizer. In this way, the parameter setting of the neural network model is completed.

[0099] Next, the boundary load-target point displacement data set is obtained as the training data set and the test data set to train the defined neural network model and obtain a prediction model for predicting the target point displacement.

[0100] In an example embodiment of the present application, 80% of the boundary load-target point displacement dataset can be passed to the neural network model as training data, and the fit method can be used for training. For example, the number of training iterations can be set to 50 rounds, and the number of samples in each batch is 16. In an example embodiment of the present application, 20% of the boundary load-target point displacement dataset can be used as a validation set to detect model performance. For example, 20% of the boundary load-target point displacement dataset is used as test data to calculate the loss value of the model to evaluate model performance. After training and testing, a prediction model for forward prediction of target point displacement for set boundary load conditions can be obtained.

[0101] In step S430, the set boundary load condition is forwardly input into the prediction model to obtain the target point displacement prediction value.

[0102] In an exemplary embodiment of the present application, the set boundary load conditions can be forward-inputted into the prediction model to obtain the target point displacement prediction value. For example, the set boundary load conditions can be a set of random boundary load conditions, where F x =-756N, F y =-751N, F z =-4411N,M x =1865N·m, M y =-4144N·m, M z =-5997N·m. Through the prediction model (see the prediction process Figure 5 ), the displacement values ​​of 100 target points can be obtained, and some of the results are shown in Table 2 below, where U x 、U y 、U z Represents the displacement values ​​of the corresponding target point in the x-direction, y-direction and z-direction respectively.

[0103] Table 2 Example of target point displacement prediction value

[0104]

[0105]

[0106] In step 440, the measured displacement data of the target point is used as a constraint condition, and according to the predicted displacement value of the target point, a reverse prediction is performed using the prediction model to obtain the predicted displacement value of the positioning structure terminal.

[0107] In the example embodiments of the present application, the target point displacement prediction value of the prediction model and its output can be obtained by inverse difference optimization. For example, the obtained target point measured displacement data can be used as a target function, and the positioning structure terminal displacement (or boundary load) on the input side of the prediction model can be continuously adjusted to make the output result (i.e., the target point displacement prediction value) of the prediction model close to the target function. When the output result of the prediction model reaches the condition of the target function, the positioning structure terminal displacement value on the input side of the prediction model is the terminal displacement prediction value obtained by inverse prediction. The terminal displacement prediction value obtained by the above inverse prediction and the measured value are shown in Table 3, wherein the maximum prediction error is 2.3%. As can be seen, the displacement prediction value obtained by the positioning structure terminal displacement prediction method provided by the present application can meet the accuracy requirements of displacement prediction.

[0108] Table 3 Comparison of terminal displacement inverse prediction value and measured value

[0109]

[0110] Figure 6 An inverse prediction method flow chart of the deformation of the positioning structure terminal according to the second example embodiment of the present application is shown.

[0111] According to the overall process shown in Figure 2 , according to the second example embodiment of the present application, an inverse prediction method of the deformation of the positioning structure terminal is provided. As shown in Figure 6 , in addition to the steps shown in Figure 5 , the method comprises the following steps.

[0112] In step 510, laser measurement is used to obtain the measured model of the positioning structure and the measured displacement data of the target point.

[0113] For the positioning structure to be measured whose target points have been identified, before loading, the initial measurement data of the positioning structure can be obtained by measurement, including the three-dimensional model (e.g., point cloud data) of the positioning structure before loading and the three-dimensional coordinates of the target points. For example, the three-dimensional data of the positioning structure to be measured can be obtained by scanning the positioning structure to be measured by a laser scanner; after reconstruction, the measured model of the positioning structure can be obtained.

[0114] After loading is completed, the positioning structure deforms; by measuring again, the measurement data after loading can be obtained, including the three-dimensional model of the deformed positioning structure and the three-dimensional coordinates of the target points. The three-dimensional model of the positioning structure and the three-dimensional coordinates of the target points obtained by measurement before and after loading can be fused and calculated to obtain the measured deformation data of the target points on the positioning structure, including the relative positions of the target points, the three-dimensional displacement deformation, etc.

[0115] In step S520, a boundary condition sample data set is obtained by using a DOE method.

[0116] In the example embodiments of the present application, a DOE test design method based on Latin hypercube sampling can be used to obtain the boundary condition sample data set. For each individual boundary condition of the n boundary conditions, the original sample space is divided into M parts, and a sample is randomly selected in each part to ensure the uniformity of the original sample; then, the n boundary conditions are designed by orthogonal experiment to generate N groups of different boundary condition combinations.

[0117] In the example embodiments of the present application, 6 boundary condition values F x , F y , F z , M x , M y , M z are used, where F x , F y , F z correspond to the forces on the x, y, and z axes, respectively, and M x , M y , M z correspond to the bending moment loads on the x, y, and z axes, respectively. For each individual boundary condition of the 6 boundary condition values, the original sample space can be divided into 1000 parts, and a sample is randomly selected in each part, and then the 6 boundary condition values are designed by orthogonal experiment to generate 2000 groups of different boundary condition combinations. Part of the obtained boundary condition combinations are shown in Table 4 below.

[0118] Table 4 Part of the boundary conditions obtained based on the DOE method

[0119]

[0120]

[0121] In step S530, a positioning structure simulation model is constructed based on the positioning structure measured model.

[0122] In the example embodiments of the present application, the positioning structure simulation model is constructed by defining material properties, meshing, setting boundary conditions, etc. based on the reconstructed positioning structure measured model. For example, the positioning structure simulation model can be constructed in a finite element analysis software.

[0123] For example, the positioning structure measured model measured by the laser scanner can be imported into the finite element analysis software; if there is a loss of geometric elements, the positioning structure measured model can be repaired in a local area. According to the actual material of the positioning structure, the material properties and the cross section can be defined. In the example embodiment of the present application, the positioning structure can use 6061T6 aluminum alloy material, and the material parameters are shown in Table 5.

[0124] Table 5 Positioning structure material properties

[0125]

[0126] In the example embodiment of the present application, tetrahedral mesh elements can be used to mesh the positioning structure to avoid mesh distortion and ensure the convergence of the calculation. In the process of meshing, fine meshing can be performed at the chamfer and the opening, and coarse meshing can be performed at other parts to improve the calculation efficiency.

[0127] In the example embodiment of the present application, for the constructed positioning structure simulation model, an analysis step can be set, the maximum incremental step can be defined as 10000, the initial analysis step is 0.01, the minimum analysis step is 10E-10, and the maximum analysis step is 1. In the analysis step setting process, a node set of the target point region can also be established, such as setting stress (S), strain (E), displacement (U) and other variables for the target point region. According to the actual loading condition of the positioning structure, the measured positioning structure simulation model is set with corresponding boundary conditions, and the positioning structure simulation model can be obtained.

[0128] Figure 7 A block diagram of an inversion prediction device for terminal deformation of a positioning structure according to the first example embodiment of the present application is shown.

[0129] According to the above overall process, according to the first example embodiment of the present application, an inversion prediction device 100 for terminal deformation of a positioning structure is provided. As shown in Figure 7 The device includes a sample data set acquisition module 110, a prediction model construction module 120, a target point displacement prediction module 130, and a terminal displacement reverse prediction module 140.

[0130] In the example embodiment of the present application, the sample data set acquisition module 110 is configured to perform finite element analysis on the positioning structure simulation model using the boundary condition sample data set, and obtain a boundary load-target point displacement sample data set.

[0131] In the example embodiment of the present application, the prediction model construction module 120 is configured to train and verify the constructed neural network model using the boundary load-target point displacement data set, and obtain a prediction model.

[0132] In the example embodiment of the present application, the target point displacement prediction module 130 is configured to input the set boundary load condition into the forward prediction model to obtain the target point displacement prediction value.

[0133] In the example embodiment of the present application, the terminal displacement inverse prediction module 140 is configured to use the target point actual displacement data as a constraint condition, and inversely predict the prediction model based on the target point displacement prediction value to obtain the predicted displacement value of the positioning structure terminal.

[0134] Figure 8 A block diagram of an inverse prediction device for terminal deformation of a positioning structure according to the second example embodiment of the present application is shown.

[0135] According to the above overall process, the inverse prediction device 200 for terminal deformation of a positioning structure is provided according to the second example embodiment of the present application. As shown in Figure 8 In addition to the modules shown in Figure 7 , the device 200 further includes an actual data acquisition module 210, a boundary condition acquisition module 220, and a simulation model construction module 230.

[0136] In the example embodiment of the present application, the actual data acquisition module 210 is configured to acquire the actual model of the positioning structure and the actual displacement data of the target point by using laser measurement.

[0137] In the example embodiment of the present application, the boundary condition acquisition module 220 is configured to acquire the boundary condition sample data set by using the DOE method.

[0138] In the example embodiment of the present application, the simulation model construction module 230 is configured to construct the simulation model of the positioning structure based on the actual model of the positioning structure

[0139] Figure 9 A block diagram of an electronic device according to the example embodiment of the present application is shown.

[0140] According to the above overall process, the electronic device 700 for inverse prediction of terminal deformation of a positioning structure is provided according to the example embodiment of the present application. Figure 9 The electronic device 700 shown is merely an example, and should not impose any limitation on the functions and use range of the embodiments of the present application.

[0141] As shown in Figure 9 , the electronic device 700 is in the form of a general computing device. The components of the electronic device 700 can include, but are not limited to, at least one processing unit 710, at least one storage unit 720, a bus 730 connecting different system components (including the storage unit 720 and the processing unit 710), etc.

[0142] The storage unit 720 stores program codes which can be executed by the processing unit 710 so that the processing unit 710 performs the method according to the embodiments of the present application described in the specification.

[0143] The storage unit 720 can include a readable medium in the form of a volatile storage unit, such as a random access memory (RAM) 7201 and / or a cache memory 7202, and can further include a read-only memory (ROM) 7203.

[0144] The storage unit 720 can further include a program / utility 7204 having a set of program modules 7205, including but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which can include an implementation of a networking environment, or a combination thereof.

[0145] The bus 730 can represent one or more of several types of bus structures, including a storage unit bus or bus controller, a peripheral bus, a graphics acceleration port, a processing unit bus, or a local bus using any of a variety of bus architectures.

[0146] The electronic device 700 can also communicate with one or more external devices 7001, such as a touch screen, a keyboard, a pointing device, a Bluetooth device, etc.; and can communicate with one or more devices that enable a user to interact with the electronic device 700 and / or one or more devices that enable the electronic device 700 to communicate with one or more other computing devices. Such communication can be via an input / output (I / O) interface 750. Further, the electronic device 700 can communicate with one or more networks, such as a local area network (LAN), a wide area network (WAN), and / or the public network, such as the Internet, via a network adapter 760. The network adapter 760 can communicate with the other modules of the electronic device 700 via the bus 730. It should be appreciated that other hardware and / or software modules can be used in conjunction with the electronic device 700, including but not limited to, microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data archival storage systems, etc.

[0147] In addition, the present application also provides a computer readable medium, which stores a computer program, and the program is executed by a processor to implement the inversion prediction method of the positioning structure terminal deformation.

[0148] The application provides an inversion prediction method for terminal deformation of a positioning structure, and the method comprises the following steps: adopting finite element analysis and a neural network model to construct a target point displacement data set according to a measured model of the positioning structure and a target point position marked on the positioning structure; and performing reverse operation on the trained neural network model based on the constructed target point displacement data set, so as to realize inversion prediction of terminal displacement of the positioning structure and solve the problem of deformation measurement of invisible parts. It is verified that the predicted terminal displacement value is compared with an actual error value, and the error is within an allowable error range.

[0149] The above describes the embodiments of the application in detail, and the principles and implementation manners of the application are described by using specific examples. The above embodiment descriptions are only used to help understand the method and core idea of the application. Meanwhile, changes or deformations made by those skilled in the art according to the idea of the application, based on the specific implementation manners and application range of the application, all belong to the protection range of the application. In summary, the content of the specification should not be understood as a limitation of the application.

Claims

1. A method for inverse prediction of terminal deformation of a positioning structure, characterized in that: The method comprises: Using the boundary condition sample data set, finite element analysis is performed on the positioning structure simulation model to obtain the boundary load-target point displacement sample data set; Using the boundary load-target point displacement sample data set to train and verify the constructed neural network model to obtain a prediction model; Forward inputting the set boundary load conditions into the prediction model to obtain the target point displacement prediction value; Taking the measured displacement data of the target point as a constraint condition, according to the predicted displacement value of the target point, using the prediction model to perform reverse prediction, and obtaining the predicted displacement value of the positioning structure terminal based on the set boundary load condition; The boundary condition sample data set includes 6 boundary condition values: 、 、 、 、 ,in, 、 Corresponding to the forces on the x, y, and z axes respectively, 、 、 Corresponding to the bending moment loads on the x, y, and z axes respectively.

2. The method according to claim 1, characterized in that The method further comprises: An experimental design method based on Latin hypercube sampling is used to obtain the boundary condition sample data set.

3. The method according to claim 1, characterized in that The method further comprises: Laser measurement is used to obtain the measured model of the positioning structure and the measured displacement data of the target point.

4. The method according to claim 3, characterized in that The method further comprises: Based on the measured model of the positioning structure, a simulation model of the positioning structure is constructed.

5. The method according to claim 4, characterized in that The method further comprises: Tetrahedral grid units are used to mesh the measured model of the positioning structure.

6. The method according to any one of claims 1 to 5, characterized in that The neural network model includes a multi-layer perception mechanism neural network model.

7. An inversion prediction device for positioning structure terminal deformation, characterized in that: The device comprises: A sample data set acquisition module is used to perform finite element analysis on the positioning structure simulation model using the boundary condition sample data set to obtain a boundary load-target point displacement sample data set; A prediction model construction module is used to train and verify the constructed neural network model using the boundary load-target point displacement sample data set to obtain a prediction model; A target point displacement prediction module is used to forward input the set boundary load conditions into the prediction model to obtain a target point displacement prediction value; The terminal displacement reverse prediction module is used to use the measured displacement data of the target point as a constraint condition, perform reverse prediction using the prediction model according to the displacement prediction value of the target point, and obtain the predicted displacement value of the positioning structure terminal based on the set boundary load condition; The boundary condition sample data set includes 6 boundary condition values: 、 、 、 、 ,in, 、 Corresponding to the forces on the x, y, and z axes respectively, 、 、 Corresponding to the bending moment loads on the x, y, and z axes respectively.

8. The device according to claim 7, characterized in that The device further comprises: A measured data acquisition module, configured to acquire a measured model of the positioning structure and measured displacement data of the target point using laser measurement; A boundary condition acquisition module, configured to acquire the boundary condition sample data set by adopting a Latin hypercube sampling experimental design method; The simulation model construction module is used to construct the positioning structure simulation model based on the positioning structure measured model.

9. An electronic device for inverse prediction of deformation of a positioning structure terminal, characterized in that: include: one or more processors; a storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

Images