Part deformation prediction method and device, electronic equipment and storage medium

Abstract

The invention discloses a part deformation prediction method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining the geometric information of a part, and enabling the geometric information to represent the geometric shape of the part; the residual stress of the part is obtained, wherein the residual stress represents the stress of any position in the part in the preset direction; inputting the geometric information and the residual stress into a deformation prediction model to obtain deformation prediction information; the deformation prediction model is used for predicting the deformation trend of the part according to the geometric information and residual stress of the part; the deformation prediction model comprises multiple layers of Fourier neural operators, and the multiple layers of Fourier neural operators are used for determining deformation prediction information by solving the partial differential relation between part deformation and residual stress and part geometry. And the deformation prediction information of the parts in different shapes is accurately determined by solving the partial differential relationship. According to the invention, the deformation prediction of various parts in different shapes can be realized through one deformation prediction model, and the universality and accuracy of the deformation prediction of the parts are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of part design and manufacturing, and particularly relates to a part deformation prediction method and device, an electronic device and a storage medium. BACKGROUND

[0002] Part deformation control is a common concern in the fields of part design and manufacturing. In the field of design, the structure of a part determines the ability of the part to resist deformation, so the quality control of the part is ensured through structural design optimization. In the field of manufacturing, the quality control of the part is mainly achieved through processing deformation control of the part, and the main measure of processing deformation control is process optimization. Predicting the processing deformation of a part is an important step in optimizing the processing technology and controlling the processing deformation of the part.

[0003] At present, the analysis method used in part deformation prediction can only predict parts with flat or constant cross-section structures, and cannot accurately predict the deformation trend of parts with different shapes when the shape of the part changes. How to provide a universal and accurate part deformation prediction scheme has become a problem to be solved. SUMMARY

[0004] The present application provides a part deformation prediction method and device, an electronic device and a storage medium to solve the problem that the current deformation analysis method can only predict parts of one shape.

[0005] According to an aspect of the present application, a part deformation prediction method is provided, comprising:

[0006] obtaining geometric information of a part, the geometric information representing the geometric shape of the part;

[0007] obtaining residual stress of the part, the residual stress representing the stress of any position in the part in a predetermined direction;

[0008] inputting the geometric information and the residual stress into a deformation prediction model to obtain deformation prediction information; the deformation prediction model is used to predict the deformation trend of the part according to the geometric information and the residual stress of the part; the deformation prediction model comprises a multi-layer Fourier neural operator, which is used to determine the deformation prediction information by solving the partial differential relationship between the part deformation, the residual stress and the part geometry.

[0009] According to another aspect of the present application, a part deformation prediction device is provided, comprising:

[0010] a geometric information acquisition module configured to acquire geometric information of a part, the geometric information representing the geometric shape of the part;

[0011] a residual stress obtaining module configured to obtain a residual stress of the part, the residual stress representing a stress of any position in the part in a preset direction;

[0012] a prediction module configured to input the geometric information and the residual stress into a deformation prediction model to obtain deformation prediction information, the deformation prediction model being configured to predict a deformation trend of the part according to the geometric information and the residual stress of the part, the deformation prediction model comprising a multi-layer Fourier neural operator, the multi-layer Fourier neural operator being configured to determine the deformation prediction information by solving a partial differential relationship between the deformation of the part and the residual stress and the geometry of the part.

[0013] According to another aspect of the present application, an electronic device is provided, the electronic device comprising:

[0014] at least one processor; and

[0015] a memory connected to the at least one processor in communication; wherein,

[0016] the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the deformation prediction method of the part according to any one of the embodiments of the present application.

[0017] According to another aspect of the present application, a computer readable storage medium is provided, the computer readable storage medium storing computer instructions for enabling a processor to implement the deformation prediction method of the part according to any one of the embodiments of the present application when executed by the processor.

[0018] The technical scheme of the embodiment of the present application acquires geometric information of a part, the geometric information representing a geometric shape of the part; acquires residual stress of the part, the residual stress representing stress of any position in the part in a preset direction; inputs the geometric information and the residual stress into a deformation prediction model to obtain deformation prediction information; the deformation prediction model is used for predicting a deformation trend of the part according to the geometric information and the residual stress of the part; and the deformation prediction model comprises a multilayer Fourier neural operator, which is used for determining the deformation prediction information by solving a partial differential relationship between part deformation and residual stress and part geometry. Compared with the current analysis method which cannot predict the deformation of parts of various shapes, the technical scheme provided by the embodiment of the present application can predict the deformation trend of the part through the deformation prediction model after the geometric information and the residual stress of the part are acquired. The shape of the part is different in different processing steps, for example, the thickness of the part locally or a certain plane gradually decreases in multiple turning operations, the shape of the part changes accordingly, and the part geometry and the part bending stiffness also change accordingly. The prediction model can perform partial differential calculation according to the residual stress and the part bending stiffness to obtain the part deformation, and accurately determine the deformation prediction information of the part of different shapes by solving the above partial differential relationship. The deformation prediction of parts of various different shapes can be realized through one deformation prediction model, and the generality and accuracy of the part deformation prediction are improved.

[0019] It should be understood that the content described in this part is not intended to identify key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0020] 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 as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0021] Figure 1 is a flowchart of a part deformation prediction method provided by the embodiment of the present application;

[0022] Figure 2 is a schematic diagram of the relationship between the stress field and the geometry of the part and the deformation mechanism of the part provided by the embodiment of the present application;

[0023] Figure 3 is a structural schematic diagram of a prediction model provided by the embodiment of the present application;

[0024] Figure 4 is a schematic diagram of the application of the prediction model provided by the embodiment of the present application;

[0025] Figure 5 is a structural schematic diagram of a part deformation prediction device provided by an embodiment of the present application;

[0026] Figure 6 is a structural schematic diagram of an electronic device implementing a part deformation prediction method of an embodiment of the present application. DETAILED DESCRIPTION

[0027] In order to enable persons skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by persons skilled in the art without creative labor should fall within the scope of protection of the present application.

[0028] It should be noted that the terms "first", "second", and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" 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 does not have to be limited to only those steps or units clearly listed, but can include other steps or units that are not clearly listed or inherent to the process, method, product, or device.

[0029] Part deformation control is a common concern in the field of part design and manufacturing. In the field of design, the structure of the part determines the ability of the part to resist deformation, so that the quality control of the part is ensured through structure design optimization. In the field of manufacturing, the quality control of the part is mainly achieved through part machining deformation control, and the main measure of machining deformation control is process optimization. Predicting the machining deformation of the part is an important step in optimizing the machining process and controlling the machining deformation of the part.

[0030] Traditional part deformation prediction methods include analytical methods and numerical methods. Analytical methods use established mathematical models to accurately represent the deformation of structural parts during processing. However, most analytical methods require simplifying the target problem within the scope of classical mechanics models, which can only be used for flat and constant cross-section structural parts. The simplification process loses prediction accuracy and limits the scope of practical application. Numerical methods such as finite element method and finite volume method are the most widely used and effective methods. However, when numerical methods are used to solve large-scale and complex problems such as large and complex structural parts, the huge amount of calculation seriously affects the calculation efficiency, making it difficult to meet the need for a large number of iterative calculations to obtain the best results in the process of structural design optimization and process optimization. Existing deformation prediction methods widely use data-driven methods based on machine learning, which have strong nonlinear fitting capabilities and good accuracy and stability in deformation prediction problems. However, the prediction model obtained by this method is usually aimed at the problem of residual stress field under the same part geometry, and still needs to be retrained for new part geometry independent of the training data.

[0031] It can be seen that the current analysis method used in part deformation prediction can only predict flat or constant cross-section structural parts. When the shape of the part changes, it cannot accurately predict the deformation trend of parts with different shapes. Numerical methods require a large amount of calculation and cannot be applied to complex structural parts. How to provide a universal and accurate part deformation prediction scheme has become a problem to be solved.

[0032] Figure 1 is a flowchart of a part deformation prediction method provided by an embodiment of the present application. The present embodiment can be applied to the case of predicting the deformation of a part, which can be a part of an aircraft, a ship, or other transportation carriers. The method can be executed by a part deformation prediction device, which can be implemented in the form of hardware and / or software. The part deformation prediction device can be configured in a personal computer, a server, or other electronic devices. As shown in the figure, the method comprises the following steps. Figure 1

[0033] S101, obtain the geometric information of the part, which represents the geometric shape of the part.

[0034] Optionally, the step of obtaining the geometric information of the part can be implemented as follows.

[0035] By measuring the part, the size information is obtained; and the geometric matrix is obtained according to the size information and the convolution network.

[0036] ​The geometry information of the part can be obtained by measurement. The geometry information, also referred to as geometry input, is used to describe the geometry shape of the part. For example, the distance between each adjacent contour point of the part can be measured, and the three-dimensional coordinates of each contour point can be determined after determining the reference coordinate origin. The three-dimensional coordinates of each contour point can be constructed into a coordinate matrix in a predetermined order. The coordinate matrix can be used as the geometry information, and the coordinate points in the coordinate matrix can represent the geometry shape of the part.

[0037] The three-dimensional image of the part in the three-dimensional space can also be obtained by scanning the part through image recognition technology. The geometry information of the part can be determined according to the three-dimensional image. The geometry information can represent the geometry shape of the part through the three-dimensional coordinates of the coordinate points.

[0038] After obtaining the geometry information, the geometry information can be processed through the convolution and deconvolution network to obtain a geometry matrix representing the geometry information. The geometry matrix obtained through the convolution and deconvolution network can more accurately describe the shape of the part through more position points, and the accuracy of the shape expression of the part is improved.

[0039] In S102, residual stress of the part is obtained, and the residual stress represents stress of any position in the part in a predetermined direction.

[0040] Figure 2 A schematic diagram of the relationship between the stress field and the geometry of the part and the deformation mechanism of the part is provided for the embodiments of the present application. As shown in Figure 2 , the part has residual stress σ x in the x-axis direction and residual stress σ y in the y-axis direction. The shape of the part in combination with the residual stress can generate bending moment M x in the x-axis direction and bending moment M y in the y-axis direction. The bending moment causes the deformation of the part.

[0041] Taking a typical wall plate part in the field of aviation manufacturing as an example, the main feature of this type of part is a slot cavity opened at the top of the part. The initial shape of the part can be rectangular, and the slot is opened at the top of the part by a lathe. Each time the lathe is turned, the depth of the slot cavity is correspondingly increased. As the part is continuously machined, the residual stress and the bending moment at each position of the part change accordingly.

[0042] For example, Figure 2 , the part is a thin plate element, and the ratio between the thickness of the part and the minimum side length is less than 0.1, which is suitable for analysis by the thin plate theory. Affected by the residual stress σ x and σ y , the bending moment M x and M y are concentrated to cause the deformation trend of the part. According to the above mechanical model, the relationship between the initial residual stress and the bending moment can be obtained as follows:

[0043] M x (x, y) = Function1 (σ x (x, y, z)) (Expression 1)

[0044] M y (x, y) = Function2 (σ y (x, y, z)) (Expression 2)

[0045] When the part is not externally constrained, the bending moment causes the part to deform, and the deformation u and the moment exhibit a partial differential relationship, as follows:

[0046] u(x, y) = Function3 (M x (x, y), M y (x, y), B(x, y)) (Expression 3)

[0047] where B is the bending stiffness of the part, and u is the deformation of the part. The moment in the above expression is substituted with the relationship with the initial residual stress, to obtain the partial differential relationship of the deformation of the part and the residual stress, as follows:

[0048] u(x, y) = Function4 (σ x (x, y), σ y (x, y), B(x, y)) (Expression 4)

[0049] According to Expression 4, the deformation at a position of the part can be obtained by partial differentiation according to the residual stress at the position and the bending stiffness of the part. It should be noted that the bending stiffness exists as an influencing factor of deformation in the deformation mechanism description, and is actually directly affected by the geometry of the part, and the influence of the geometry on the deformation process is not limited to the bending stiffness.

[0050] Optionally, obtaining the residual stress of the part can be implemented as:

[0051] obtaining stress field information of the part; and determining a residual stress matrix according to the stress field information.

[0052] The stress field information of any position of the part can be obtained according to the stress field testing method provided by the prior art. The stress field information is the residual stress in the x-axis direction and the y-axis direction at the any position. The residual stress can be the stress existing at a position of the part in a static state before the part is processed. The residual stress field matrix is obtained by combining the stress field information at each position of the part. The residual stress field matrix is adapted to the input of the deformation prediction model, so that the deformation prediction model can more quickly identify the stress field information of the part and improve the deformation prediction efficiency.

[0053] S103, input the geometric information and the residual stress into a deformation prediction model to obtain deformation prediction information; the deformation prediction model is used to predict the deformation trend of the part according to the geometric information and the residual stress of the part; the deformation prediction model comprises a multilayer Fourier neural operator, and the multilayer Fourier neural operator is used to determine the deformation prediction information by solving the partial differential relationship between the part deformation, the residual stress and the part geometry.

[0054] Figure 3 A structural diagram of a deformation prediction model provided by an embodiment of the present application is shown. The deformation prediction model comprises a plurality of Fourier neural operators (FNO), a convolution layer for convolution processing of geometric input, and a fully connected layer. The plurality of Fourier neural operators are more accurate and efficient when solving partial differential equations.

[0055] Since the object to be solved is a partial differential equation, it is difficult to accurately approximate and solve directly using a general neural network structure due to factors such as multivariable, strong nonlinearity, and variable problem space. Therefore, a Fourier neural operator (FNO) structure is introduced into the model, and the spectral function form of the partial differential equation is approximated through calculation and feature extraction in a multilayer Fourier space. The neural operator network is an iterative structure divided by layers: v0, b1, … v n . The relationship of the network layer G i from v j to v j can be written as:

[0056] G j = Function5 (W j , K j , b j , σ j ) (Expression 5)

[0057] where W j is the linear term of the layer, b j is the bias term, K j is the integral kernel operator, and σ j is the nonlinear activation function. In the Fourier neural network, the integral kernel operator K j is represented as an operator K j (φ) defined in the Fourier space, and includes Fourier transform and inverse Fourier transform:

[0058] K j (φ) = Function6 (R j (φ)) (Expression 6)

[0059] where R j(φ) is a filter function in the Fourier space, and φ is a parameter related to the Fourier transform.

[0060] The partial differential relationship of the stress field is solved through the multilayer Fourier neural operator to obtain a solving result.

[0061] The multilayer Fourier neural operator structure provided by the deformation prediction model provided by the embodiment of the present application can accurately output the deformation trend information corresponding to a part of any geometric shape based on the solving of the partial differential equation.

[0062] Further, before the geometric information and the residual stress are input to the deformation prediction model, the method further comprises:

[0063] The training data is determined according to the simulation data and the actual measurement data, and the deformation prediction model is trained according to the training data.

[0064] The training data is determined according to the simulation data and the actual measurement data, and the deformation prediction model is trained according to the training data.

[0065] Different geometric shape parts are constructed in a simulation environment, initial geometric information and initial residual stress of the parts are obtained, a machining process of the parts is simulated in the simulation environment, deformation information of the parts in different machining steps is collected, and the training data is determined according to the obtained initial geometric information, initial residual stress and deformation information.

[0066] Optionally, only the deformation information of the parts when the machining is completed can be collected.

[0067] The deformation prediction model provided by the embodiment of the present application is trained based on a plurality of part data with different geometric shapes. The collection of the training data is performed in a simulation environment. A corresponding blank is imported through a simulation software, and a corresponding initial residual stress and a material removal process simulation step are set. Finally, the deformation data in the simulation result is collected as label data, and the deformation data of the parts is obtained by simulating a real machining scene. Through this method, a large amount of training data containing various geometric features can be quickly obtained, and the material and time costs required by actual machining experiments are effectively saved.

[0068] The simulation data and the actual data can be sampled based on a voxel sampling method, the sampling space size and the resolution in each direction are set according to the approximate size range of the part to be predicted, and the recorded data is mapped to a standard space as an input and output layer of a neural network. For the geometric information of the part, the average condition of the material distribution in the sampling unit is divided into two cases of having material and not having material to record.

[0069] The actual measurement data of the physical parts can be obtained. The simulation data obtained by simulation and the actual measurement data are combined to obtain the training data of the deformation prediction model.

[0070] Furthermore, the deformation prediction model is trained according to the training data, including:

[0071] The mini-batch method and ADAM method are used to optimize the convergence speed and fitting effect of the deformation prediction model.

[0072] The model training process uses the mini-batch gradient descent method and the Adaptive Moment Estimation (ADAM) method to optimize the model convergence speed and fitting effect. The model training process uses the L2 loss function and limits the complexity of the model parameters. The loss function L is expressed as follows:

[0073] L=Function7(er x ,er y ,er z , M p )(Expression 7)

[0074] Where er represents the error between the model prediction value and the true value in each direction, for example, er x Indicates the error between the model prediction value and the true value in the x-axis direction, er y Indicates the error between the model prediction value and the true value in the y-axis direction, er z Indicates the error between the model prediction value and the true value in the z-axis direction, M p Represents model parameters.

[0075] You can choose whether to use mini-batch optimization according to your needs. You can also replace the ADAM method with other learning rate optimization algorithms.

[0076] The above optimization method can achieve model convergence more quickly and improve the training efficiency of the deformation prediction model.

[0077] Figure 4 This is a schematic diagram of the model application provided by an embodiment of the present invention. The residual stress field of a part can be obtained using stress measurement calculation methods and specialized equipment. Part geometry data is collected using probes, scanning, and other methods, and then input into the trained model after voxel sampling, ultimately outputting predicted data for the part's machining deformation field. After machining is complete and the fixture is released, the part undergoes machining deformation, which is measured and compared with the model's predicted results.

[0078] Further, after inputting the geometric information and residual stress into the deformation prediction model to obtain deformation prediction information, the method further includes:

[0079] According to the prediction result and the actual deformation information, the prediction result is compared; if the comparison result is inaccurate prediction, the deformation prediction model is retrained.

[0080] The training needs sufficient data support, and it is difficult to adjust the model in real time according to the real-time processing effect. It is found through research that the prediction failure scene generally needs to expand the geometric features included in the model training data set to further expand the model universality; or improve the voxel grid precision. Based on this, the retraining can be performed according to more initial geometric information and initial residual stress of parts with different geometric shapes and deformation information obtained through simulation as expanded training data, and the prediction model is trained according to the expanded training data. The retraining can also improve the voxel grid precision, so that the obtained geometric information of the part is more accurate, the geometric matrix dimension is higher in precision, and the geometric shape of the part can be more accurately recognized, and the prediction accuracy of the deformation prediction model is improved.

[0081] If it is found through the prediction result comparison that the prediction is inaccurate, the deformation prediction model is retrained. Through the prediction result comparison, the model prediction effect can be better detected, the deformation prediction model can be adjusted in time, and the accuracy of the deformation prediction model can be improved.

[0082] The part deformation prediction method provided by the embodiment of the present application comprises the following steps: obtaining geometric information of a part, wherein the geometric information represents a geometric shape of the part; obtaining residual stress of the part, wherein the residual stress represents stress of any position in the part in a preset direction; inputting the geometric information and the residual stress into a deformation prediction model to obtain deformation prediction information; the deformation prediction model is used to predict a deformation trend of the part according to the geometric information and the residual stress of the part; and the deformation prediction model comprises a multilayer Fourier neural operator, which is used to determine the deformation prediction information by solving a partial differential relationship between part deformation, residual stress and part geometry. Compared with the current analysis method which cannot predict the deformation of parts of various shapes, the part deformation prediction method provided by the embodiment of the present application can predict the deformation trend of the part through the deformation prediction model after obtaining the geometric information and the residual stress of the part. The shape of the part is different in different processing steps, for example, the thickness of a part or a plane thereof gradually decreases during multiple turning operations, the shape of the part changes accordingly, and the part geometry and the bending stiffness of the part also change accordingly. The prediction model can perform partial differential calculation according to the residual stress and the bending stiffness of the part to obtain the deformation of the part, and accurately determine the deformation prediction information of the part of different shapes by solving the above partial differential relationship. The deformation prediction of parts of various different shapes can be realized through one deformation prediction model, and the generality and accuracy of part deformation prediction are improved.

[0083] Figure 5 Figure 1 is a structural schematic diagram of a part deformation prediction device provided by the embodiment of the present application, which is suitable for the case of predicting the deformation of a part. The part can be a part in a transport carrier such as an airplane or a ship. Figure 5 As shown in the figure, the device comprises a geometric information acquisition module 21, a residual stress acquisition module 22 and a prediction module 23.

[0084] The geometric information acquisition module 21 is used to acquire geometric information of a part, wherein the geometric information represents a geometric shape of the part.

[0085] The residual stress acquisition module 22 is used to acquire residual stress of the part, wherein the residual stress represents stress of any position in the part in a preset direction.

[0086] The prediction module 23 is used to input the geometric information and the residual stress into a deformation prediction model to obtain deformation prediction information; the deformation prediction model is used to predict a deformation trend of the part according to the geometric information and the residual stress of the part; and the deformation prediction model comprises a multilayer Fourier neural operator, which is used to determine the deformation prediction information by solving a partial differential relationship between part deformation, residual stress and part geometry.

[0087] On the basis of the above-mentioned embodiments, optionally, the geometric information acquisition module 21 is configured to:

[0088] The size information is obtained by measuring the part;

[0089] The geometric matrix is obtained according to the size information and the convolution network.

[0090] On the basis of the above-mentioned embodiments, optionally, the residual stress acquisition module 22 is configured to:

[0091] Obtain the stress field information of the part;

[0092] The residual stress matrix is determined according to the stress field information.

[0093] On the basis of the above-mentioned embodiments, optionally, the model training module is further configured to determine training data according to simulation data and actual measurement data before the geometric information and the residual stress are input into the deformation prediction model;

[0094] The deformation prediction model is trained according to the training data.

[0095] On the basis of the above-mentioned embodiments, optionally, the model training module is configured to:

[0096] Parts with different geometric shapes are constructed in a simulation environment, and initial geometric information and initial residual stress of the parts are obtained;

[0097] The processing process of the parts is simulated in the simulation environment, and deformation information of the parts at different processing steps is collected;

[0098] The training data is determined according to the obtained initial geometric information, initial residual stress and deformation information.

[0099] On the basis of the above-mentioned embodiments, optionally, the model training module is configured to:

[0100] The convergence speed and fitting effect of the deformation prediction model are optimized using the mini-batch method and the ADAM method.

[0101] On the basis of the above-mentioned embodiments, optionally, the model training module is further configured to compare the prediction results according to the prediction results and the actual deformation information after the geometric information and the residual stress are input into the deformation prediction model to obtain deformation prediction information.

[0102] If the comparison result is inaccurate prediction, the deformation prediction model is retrained.

[0103] The part deformation prediction device provided by the embodiment of the present application comprises a geometric information acquisition module 21, a residual stress acquisition module 22 and a prediction module 23. The geometric information acquisition module 21 is configured to acquire geometric information of a part, wherein the geometric information represents a geometric shape of the part. The residual stress acquisition module 22 is configured to acquire residual stress of the part, wherein the residual stress represents stress of an arbitrary position in the part in a preset direction. The prediction module 23 is configured to input the geometric information and the residual stress into a deformation prediction model to obtain deformation prediction information. The deformation prediction model is configured to predict a deformation trend of the part according to the geometric information and the residual stress of the part. The deformation prediction model comprises a multilayer Fourier neural operator, and the multilayer Fourier neural operator is configured to determine the deformation prediction information by solving a partial differential relationship between part deformation and residual stress and part geometry. Compared with the current analysis method which cannot predict the deformation of parts of various shapes, the part deformation prediction device provided by the embodiment of the present application can predict the deformation trend of the part through the deformation prediction model after the geometric information and the residual stress of the part are acquired. The shape of the part is different in different processing steps. For example, in the process of multiple turning operations, the thickness of a part locally or on a certain plane gradually decreases, the shape of the part changes, and the part geometry also changes. The prediction model can perform partial differential calculation according to the residual stress and the part geometry to obtain the deformation of the part, and accurately determine the deformation prediction information of the part of different shapes by solving the above partial differential relationship. The deformation prediction of parts of various shapes can be realized through one deformation prediction model, and the generality and accuracy of part deformation prediction are improved.

[0104] The part deformation prediction device provided by the embodiment of the present application can execute the part deformation prediction method provided by any embodiment of the present application, and has the function modules and beneficial effects corresponding to the execution method.

[0105] Figure 6 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. The electronic device 10 is intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not intended to limit the implementations of the present application described and / or claimed in this document.

[0106] As Figure 6As shown, the electronic device 10 includes at least one processor 11, and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., communicatively connected to the at least one processor 11, where the memory stores a computer program executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or loaded into the random access memory (RAM) 13 from the storage unit 18. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0107] Various components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc., an output unit 17, such as various types of displays, a speaker, etc., a storage unit 18, such as a magnetic disk, an optical disk, etc., and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0108] The processor 11 can be various general and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 performs various methods and processes described above, such as the part deformation prediction method.

[0109] In some embodiments, the part deformation prediction method can be implemented as a computer program tangibly embodied in a computer readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the part deformation prediction method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to perform the part deformation prediction method by any other appropriate means, such as by means of firmware.

[0110] Various implementations of the systems and techniques described above can be realized in digital electronic circuitry, integrated circuitry, specially designed application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0111] Computer programs implementing the part deformation prediction method of the present application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the computer program running on the processor implements the functions / operations specified in the flow diagrams and / or block diagrams. The computer program can execute entirely on a machine, partly on a machine, partly on a machine and partly on a remote machine or entirely on a remote machine or server.

[0112] The embodiments of the present application also provide a computer readable storage medium, which stores computer instructions for causing a processor to execute a part deformation prediction method, the method comprising:

[0113] obtaining geometric information of the part, the geometric information representing a geometric shape of the part;

[0114] obtaining residual stress of the part, the residual stress representing stress in a preset direction at any position in the part;

[0115] inputting the geometric information and the residual stress into a deformation prediction model to obtain deformation prediction information; the deformation prediction model is used to predict a deformation trend of the part according to the geometric information and the residual stress of the part; the deformation prediction model comprises a multi-layer Fourier neural operator, and the multi-layer Fourier neural operator is used to determine the deformation prediction information by solving a partial differential relationship between part deformation, residual stress, and part geometry.

[0116] In the above embodiments, optionally, the obtaining of the geometric information of the part comprises:

[0117] obtaining the size information by measuring the part;

[0118] obtaining a geometry matrix according to the size information and the convolution network.

[0119] On the basis of the above-mentioned embodiments, optionally, the residual stress of the part is obtained, comprising:

[0120] obtaining stress field information of the part;

[0121] determining a residual stress matrix according to the stress field information.

[0122] On the basis of the above-mentioned embodiments, optionally, before inputting the geometry information and the residual stress into the deformation prediction model, further comprising:

[0123] determining training data according to simulation data and actual measurement data;

[0124] training the deformation prediction model according to the training data.

[0125] On the basis of the above-mentioned embodiments, optionally, determining training data according to simulation data and actual measurement data, comprising:

[0126] constructing parts with different geometrical shapes in a simulation environment, obtaining initial geometry information and initial residual stress of the parts;

[0127] simulating the machining process of the parts in the simulation environment, collecting deformation information of the parts at different machining steps;

[0128] determining training data according to the obtained initial geometry information, initial residual stress and deformation information.

[0129] On the basis of the above-mentioned embodiments, optionally, training the deformation prediction model according to the training data, comprising:

[0130] using the mini-batch method and the ADAM method to optimize the convergence speed and fitting effect of the deformation prediction model.

[0131] On the basis of the above-mentioned embodiments, optionally, after inputting the geometry information and the residual stress into the deformation prediction model to obtain deformation prediction information, further comprising:

[0132] comparing the prediction results according to the prediction results and the actual deformation information;

[0133] If the comparison results are inaccurate, the deformation prediction model is retrained. In the context of the present invention, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. Computer-readable storage media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of a machine-readable storage medium will include one or more lines of electrical connections, portable computer disks, hard disk drives, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fibers, portable compact disc read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0134] To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0135] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0136] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.

[0137] It should be understood that the various forms of flow shown above can be reordered, added to, or have steps deleted. For example, the steps described in the present application can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions of the present application can be achieved, which are not limited herein.

[0138] The above detailed description does not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for predicting deformation of a part, characterized in that: include: Acquiring geometric information of a part, wherein the geometric information represents a geometric shape of the part; Obtaining residual stress of a part, where the residual stress represents stress at any position in the part in a preset direction; Inputting the geometric information and residual stress into a deformation prediction model to obtain deformation prediction information; The deformation prediction model is used to predict the deformation trend of the part based on the geometric information and residual stress of the part; the deformation prediction model includes a multi-layer Fourier neural operator, and the multi-layer Fourier neural operator is used to determine the deformation prediction information by solving the partial differential relationship between the part deformation, residual stress and part geometry.

2. The method according to claim 1, characterized in that The obtaining of geometric information of the part includes: By measuring the parts, the size information is obtained; A geometric matrix is ​​obtained according to the size information and the convolutional network.

3. The method according to claim 2, characterized in that Obtain residual stresses of parts, including: Obtain the stress field information of the parts; A residual stress matrix is ​​determined according to the stress field information.

4. The method according to claim 3, characterized in that Before inputting the geometric information and residual stress into the deformation prediction model, the method further includes: Determine training data based on simulation data and actual measurement data; The deformation prediction model is trained according to the training data.

5. The method according to claim 4, characterized in that Determine the training data based on simulation data and actual measurement data, including: Constructing parts of different geometric shapes in a simulation environment and obtaining initial geometric information and initial residual stress of the parts; Simulating the machining process of the part in a simulation environment and collecting deformation information of the part in different machining steps; Training data is determined according to the obtained initial geometric information, the initial residual stress and the deformation information.

6. The method according to claim 5, characterized in that Training the deformation prediction model according to the training data includes: The mini-batch method and ADAM method are used to optimize the convergence speed and fitting effect of the deformation prediction model training process.

7. The method according to claim 6, characterized in that After inputting the geometric information and residual stress into the deformation prediction model to obtain deformation prediction information, the method further includes: Compare the predicted results with the actual deformation information; If the comparison result shows that the prediction is inaccurate, the deformation prediction model is retrained.

8. A device for predicting deformation of a part, characterized in that: include: A geometric information acquisition module, configured to acquire geometric information of a part, wherein the geometric information represents the geometric shape of the part; A residual stress acquisition module is used to obtain the residual stress of a part, where the residual stress represents the stress at any position in the part in a preset direction; A prediction module, configured to input the geometric information and residual stress into a deformation prediction model to obtain deformation prediction information; The deformation prediction model is used to predict the deformation trend of the part based on the geometric information and residual stress of the part; the deformation prediction model includes a multi-layer Fourier neural operator, and the multi-layer Fourier neural operator is used to determine the deformation prediction information by solving the partial differential relationship between the part deformation, residual stress and part geometry.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor. The computer program is executed by the at least one processor so that the at least one processor can perform the part deformation prediction method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the part deformation prediction method according to any one of claims 1 to 7 when executed.

Images