Assembly interface ideal prestress distribution inversion design method for service performance traction and related device

Abstract

The invention belongs to the technical field of mechanical design, and discloses a service performance traction assembly interface ideal prestress distribution inversion design method and a related device, and the method comprises the steps: carrying out the feature mapping of assembly interface morphology-stress on a matching surface, and obtaining a contact pressure value and a relative coordinate of each node; according to the contact pressure value and the static contact stress, performing static ideal contact stress distribution inversion to obtain static ideal contact stress distribution; and according to the static ideal contact stress distribution, predicting the morphology layout driven by the static contact stress distribution of the assembly interface. According to the method, the morphology layout driven by the static contact stress distribution of the assembly interface can be assembled, machining can be carried out on the assembly interface according to the morphology layout of the assembly interface, the connection performance uniformity of the assembly interface in the dynamic service state can be effectively improved, and the purpose of improving the dynamic connection performance of high-end equipment parts such as aero-engines is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of mechanical design, and relates to a method for improving the performance of an assembly interface of mechanical equipment in a service state, and specifically to an inverse design method for an ideal prestress distribution of an assembly interface driven by service performance and related devices. Background Art

[0002] Mechanical connection structures are used to connect mechanical parts to each other. The assembly interface between the connection structures is the key carrier for carrying and ensuring that mechanical equipment can achieve the given functions. The contact stress distribution of the assembly interface is a key factor affecting the contact mechanical properties and even the service performance of the whole machine. For example, the uneven distribution of contact stress at the assembly interface and the small effective contact area are one of the main factors for high-cycle fatigue and excessive vibration of high-end equipment parts such as aircraft engines. Compared with the assembly process optimization of the assembly interface, focusing on the design of the assembly interface can directly affect the assembly connection performance. At present, the surface morphology design of the assembly interface has been increasingly applied. For example, the world's aviation engine manufacturing giants such as Rolls Royce in the UK have significantly improved the comprehensive performance of aviation engines by designing the radial texture "micro-spline" morphology on the assembly interface between the high-pressure turbine disk and the rear journal of the aircraft engine, effectively enhancing the competitiveness of its products in the international market. By simulating the load conditions of mechanical equipment in service, the optimal assembly interface morphology layout is inversely designed using deep learning technology. With the current high-precision processing as the design guarantee, the processing results are helpful to improve the dynamic performance of mechanical equipment, including reducing vibration, improving reliability and extending service life. It can be seen that starting from the mechanical nature of the formation of assembly connection performance, the technical route aimed at regulating physical performance indicators such as assembly contact stress is a research direction with great development potential.

[0003] However, since the assembly interfaces of the connection structures of large-scale equipment such as aerospace are subjected to service conditions such as high pressure, high and low temperature, strong oxidation, variable load and strong vibration impact, the assembly stress distribution state when the connection parts are assembled will be "redesigned". Therefore, the uniformity of contact stress distribution on the assembly interface pursued by the existing technology at the assembly completion stage is not the direct target form of contact stress distribution regulated by the active design of the assembly interface morphology layout. Summary of the invention

[0004] The present invention aims to address the problem in the prior art that it is impossible to determine the stress distribution relationship of the assembly interface between the service stage and the assembly completion stage, and it is impossible to design the assembly interface according to the service conditions of the assembly interface. The present invention aims to propose an inverse design method and related devices for an ideal prestress distribution of an assembly interface driven by service performance, which can reversely evolve the static ideal contact stress distribution of the assembly interface from the homogenized dynamic service state contact stress distribution, thereby facilitating the processing of the assembly interface.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is: An inverse design method for ideal prestress distribution of assembly interface driven by service performance includes the following steps: Perform feature mapping of assembly interface morphology-stress on the mating surface to obtain the contact pressure value and the relative coordinates of each node; According to the contact pressure value and the static contact stress, the static ideal contact stress distribution is inverted to obtain the static ideal contact stress distribution; Based on the static ideal contact stress distribution, the topography layout driven by the static contact stress distribution of the assembly interface is predicted.

[0006] A further improvement of the present invention is to perform characteristic mapping of assembly interface morphology-stress on the mating surface to obtain the contact pressure value and the relative coordinates of each node, including the following steps: (1.1) Divide the assembly interface into finite element meshes, construct a finite element model for assembly interface contact analysis, and apply boundary constraints and load conditions; (1.2) Perform transient finite element contact analysis and select a cycle Sampling, calculating the contact pressure value and average contact pressure of the nodes on the assembly interface during sampling; (1.3) Calculate the contact stress variance based on the average contact pressure; (1.4) Determine the contact stress variance , if satisfied , is the threshold value, or the number of optimization iterations satisfy , is the total number of nodes in the finite element calculation process of the assembly interface, the iteration is terminated, and the relative coordinates of each node are obtained.

[0007] A further improvement of the present invention is that if or , then execute step (1.5); (1.5) Set the area for active morphology adjustment, use the node relative coordinates as the optimization design variables in theoretical analysis, set the magnification factor and reduction factor, and in the new optimization iteration step, the optimization method and update method of the actively adjusted node relative coordinate values ​​are as follows: after the update, delete the current external load condition, calculate the contact stress of each node in the current morphology layout without external load based on the contact pressure value of the reference contact node, and determine the node relative coordinate value based on the contact pressure value of the reference contact node and the contact stress of the node; (1.6) Based on the node relative coordinate values ​​obtained in step (1.5), update the assembly interface contact analysis finite element model described in step (1.1).

[0008] A further improvement of the present invention is that the average contact pressure Calculated by the following formula:

[0009] Where: is the total number of nodes in the finite element calculation process of the assembly interface, For the assembly interface The second sampling time The contact pressure value of each node; Contact stress variance Calculated by the following formula: .

[0010] A further improvement of the present invention is that, according to the contact pressure value and the static contact stress, an inversion of the static ideal contact stress distribution is performed to obtain the static ideal contact stress distribution, which includes the following steps: (2.1) The second sampling time The contact pressure value of each node is used as the dynamic service state contact stress and the static ideal contact stress Preprocess the data set and divide the preprocessed data into a training set and a validation set; (2.2) Using the U-Net model as the basic network architecture, the BP neural network is used to establish the Multi-Bp-U-Net model; (2.3) Use the training set data and the validation set data to train the Multi-Bp-U-Net model to obtain a trained Multi-Bp-U-Net model; (2.4) The trained Multi-Bp-U-Net model is used to invert the dynamic service state contact stress distribution under uniform state to obtain the static ideal contact stress distribution under uniform state.

[0011] A further improvement of the present invention is that the Multi-Bp-U-Net model comprises: (2.2.1) a feature extraction module, comprising an input convolution layer and a BpUpDown layer, wherein the input convolution layer is used to extract input multidimensional data to adapt to the input dimension; and the BpUpDown is used to extract sequence data features according to the input multidimensional data adapting to the input dimension; (2.2.2) a contraction path module, comprising a plurality of BpUpDown layers, for downsampling the sequence feature information; (2.2.3) Extended path module, including multiple BpUpDown layers for upsampling processing; (2.2.4) Skip connection module, which is used to transfer low-level features of the lower layer to the upper layer by copying and splicing.

[0012] A further improvement of the present invention is that, based on the static ideal contact stress distribution, the topography layout driven by the static contact stress distribution of the assembly interface is predicted, comprising the following steps: (3.1) Preprocessing the static ideal contact stress and relative coordinate values ​​as a data set, and dividing the preprocessed data into a training set and a validation set; (3.2) Using the U-Net model as the basic network architecture, the Multi-Bp-U-Net model is established using the fully connected layer of the BP neural network; (3.3) Using the training set data and the validation set data in step (3.1) to train the Multi-Bp-U-Net model to obtain the trained Multi-Bp-U-Net model; (3.4) The static ideal contact stress distribution is input into the trained Multi-Bp-U-Net model to predict the morphology layout and obtain the morphology layout under the static ideal contact stress distribution state.

[0013] An inverse design system for ideal prestress distribution of assembly interface driven by service performance, including: The feature mapping module is used to perform feature mapping of assembly interface morphology-stress on the mating surface to obtain the contact pressure value and the relative coordinates of each node; A static ideal contact stress distribution inversion module is used to perform static ideal contact stress distribution inversion according to the contact pressure value and the static contact stress to obtain the static ideal contact stress distribution; The morphology layout prediction module is used to predict the morphology layout driven by the static contact stress distribution of the assembly interface according to the static ideal contact stress distribution.

[0014] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the inverse design method for ideal prestress distribution of an assembly interface driven by service performance is implemented.

[0015] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the inverse design method for ideal prestress distribution of an assembly interface driven by service performance is implemented.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The characteristic mapping step of assembly interface morphology-stress in the present invention regulates the contact stress state of the assembly interface through an active design method, which can achieve the local optimal solution of the contact stress distribution under the dynamic service state, and output a sequence data set containing rich characteristic information between morphology and stress. The present invention reversely evolves the static ideal contact stress distribution of the assembly interface from the uniform dynamic service state contact stress distribution, and proposes a processing guidance scheme driven by the assembly interface morphology layout, so as to improve the dynamic connection performance of mechanical equipment parts under the service state. The present invention can obtain the assembly interface morphology layout, and perform processing on the assembly interface according to the assembly interface morphology layout, which can effectively improve the uniformity of the connection performance of the assembly interface in the dynamic service state, and achieve the purpose of improving the dynamic connection performance of parts and components of high-end equipment such as aircraft engines.

[0017] Furthermore, deep learning technology is used to build a Multi-Bp-U-Net model to explore the intrinsic connection between the contact stress in the dynamic service state output by the finite element software and the static ideal contact stress. This can achieve the global optimal solution for the contact stress distribution under the dynamic service state, predict the morphology layout under the solution state, and provide corresponding processing guidance. The training and prediction efficiency of deep learning technology is much higher than that of traditional finite element calculations. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of bolt connection; Figure 2 It is a two-dimensional cross-sectional view of the finite element model of the bolt structure; Figure 3 It is the average value diagram of the contact stress distribution in the dynamic service state of the assembly interface; Figure 4 It is the structural diagram of the Multi-Bp-U-Net model; Figure 5 is the distribution diagram of static ideal contact stress; Figure 6 It is a comparison diagram between the predicted morphology layout driven by the static contact stress distribution of the assembly interface and the initial morphology layout; Figure 7 It is a comparison diagram of the average values ​​of the contact stress distribution in the dynamic service state before and after the inversion; Figure 8 Flow chart of the inverse design method for the ideal prestress distribution of the assembly interface for service performance traction; Fig. 9 Schematic diagram of the inverse design system for the ideal prestress distribution of the assembly interface for service performance traction; Among them, 1 is the upper assembly, 2 is the lower assembly, 3 is the bolts and nuts, 4 is the upper assembly interface, 5 is the design domain, and 6 is the lower assembly interface. DETAILED DESCRIPTION

[0019] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. The preferred embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thoroughly and comprehensively understood.

[0020] Starting from the uniformity of contact stress distribution at the assembly interface pursued by the connection structure in the dynamic service stage, the contact stress distribution form of the connection structure in the assembly completion stage (called "ideal prestress distribution") is "inverted", which is the direct target form of the contact stress distribution state regulated by the active design of the assembly interface. Therefore, the present invention reversely evolves the static ideal contact stress distribution of the assembly interface from the homogenized dynamic service state contact stress distribution, and proposes a method for predicting the morphology layout of the morphology prediction model driven by the static contact stress distribution of the assembly interface. Among them, the dynamic service state contact stress is the contact stress affected by both the external load and the assembly load. The static ideal contact stress distribution is the static contact stress distribution that reflects the most uniform state of the dynamic service state contact stress distribution. The static contact stress is the contact stress that is only affected by the assembly load.

[0021] See also Figure 8 The present invention provides an inversion design method for ideal prestress distribution of an assembly interface driven by service performance, which realizes the morphological layout of the assembly interface, including characteristic mapping of assembly interface morphology-stress, inversion of static ideal contact stress distribution, and morphological prediction steps driven by static contact stress distribution of the assembly interface, which are specifically as follows: 1. Performing feature mapping of assembly interface morphology-stress on the mating surface to obtain the contact pressure value and the relative coordinates of each node; the steps include: (1.1) Divide the assembly interface into finite element meshes, construct a finite element model for assembly interface contact analysis, and apply boundary constraints and load conditions; (1.2) Perform transient finite element contact analysis and select a cycle Sampling, calculating and outputting the first The second sampling time Contact pressure value of each node With average contact pressure ;

[0022] Where: is the total number of nodes during the finite element calculation of the assembly interface.

[0023] (1.3) Based on the average contact pressure , by calculating the contact stress variance as a measure of the uniformity of the contact stress at the assembly interface , then the optimization design goal is min ;

[0024] (1.4) Determine the contact pressure value of the reference contact node , generally take the point where the contact pressure value is 0 in the assembly completion stage; (1.5) Determine the connection performance uniformity optimization design objective function , if satisfied , or optimize the number of iterations satisfy , the iteration is terminated, and the relative coordinates of each node are output and saved; if both of the above two conditions are not met, execute step (1.6); (1.6) Set the area for active morphology adjustment, i.e., design domain 5, and use the node relative coordinates as the optimization design variables for theoretical analysis, and set the amplification factor and reduction factor , in the new optimization iteration step The optimization and update methods of the actively adjusted node relative coordinate values ​​are as follows: after updating, the current external load condition is deleted and the contact pressure value of the reference contact node is calculated. , calculate the contact stress at each node , according to the contact pressure value of the reference contact node Contact stress at the node Determine the relative coordinate value of the node. , , The data is output and saved in the form of sequence data for later use.

[0025]

[0026]

[0027] Where: For the The contact node is The relative coordinate values ​​of the iteration; For the The contact stress variance of iterations; is the variable coefficient parameter; is the contact pressure value of the reference contact node; (1.7) Based on the node relative coordinate values ​​obtained in step (1.6), the assembly interface finite element model described in step (1) is updated and a new finite element contact analysis is performed.

[0028] 2. According to the contact pressure value and the static contact stress, the static ideal contact stress distribution is inverted to obtain the static ideal contact stress distribution. The steps include: (2.1) Save the sequence data as described in Contact pressure value of each node As the dynamic service state contact stress and static ideal contact stress The preprocessed data is divided into a training set and a validation set.

[0029] (2.2) Taking the U-Net model as the basic network architecture, the fully connected layer of the BP neural network is used to replace it and a Multi-Bp-U-Net model is established.

[0030] (2.3) Using the training set data and the validation set data in step (2.1) to train the Multi-Bp-U-Net model, to obtain the trained Multi-Bp-U-Net model.

[0031] (2.4) The dynamic service state contact stress distribution under the uniform state is inverted using the trained Multi-Bp-U-Net model, and the obtained static contact stress distribution is called the static ideal contact stress distribution under this state.

[0032] Wherein, the Multi-Bp-U-Net model includes: (2.2.1) a feature extraction module, comprising an input convolution layer and a BpUpDown layer, wherein the input convolution layer is used to extract input multi-dimensional data to adapt to the input dimension; the BpUpDown layer uses full connection instead of one-dimensional convolution to extract sequence data features; (2.2.2) a contraction path module, which includes a plurality of BpUpDown layers simulating convolution and pooling operations, and is used to downsample the feature information obtained by the feature extraction module; (2.2.3) Extended path module, which includes multiple BpUpDown layers, simulating the deconvolution layer operation for upsampling processing; (2.2.4) A skip connection module, which transfers low-level features of the lower layer to the upper layer by copying and splicing.

[0033] 3. Based on the static ideal contact stress distribution, predict the topography layout driven by the static contact stress distribution of the assembly interface, including the following steps: (3.1) The static ideal contact stress described in the saved sequence data With relative coordinate values The preprocessed data is divided into a training set and a validation set.

[0034] (3.2) Taking the U-Net model as the basic network architecture, the fully connected layer of the BP neural network is used to replace it and a Multi-Bp-U-Net model is established.

[0035] (3.3) Using the training set data and the validation set data in step (3.1) to train the Multi-Bp-U-Net model, to obtain the trained Multi-Bp-U-Net model.

[0036] (3.4) Using the trained Multi-Bp-U-Net model to input the static ideal contact stress distribution, the morphology layout under this state is predicted to obtain the morphology layout under the static ideal contact stress distribution state.

[0037] The assembly interface topography-stress feature mapping step can not only even out the distribution of contact pressure on the mating surface to achieve optimization, but also increase the contact pressure variance to expand the data set and improve the model generalization ability.

[0038] The static ideal contact stress distribution inversion model includes a BpUpDown module, which can replace the one-dimensional convolution operation, can more comprehensively understand the global change trend of the assembly interface morphology and stress, and is more suitable for the output structure of the finite element method of the present invention. The BpUpDown module uses a multi-path parallel method to calculate the loss function, which has a greater probability of ensuring that at least one path can perform feature mapping in a more effective way, thereby improving the generalization ability of the overall model.

[0039] See also Figure 1 The topography layout design can be selected on the upper assembly interface 1, the lower assembly interface 2, or designed on both the upper assembly interface 4 and the lower assembly interface 6.

[0040] The beneficial effects of the present invention are further illustrated by taking the most common bolt and nut 3 assembly structure (including an upper assembly 1 and a lower assembly 2) and a single vibration load in the engineering field as an example.

[0041] The two-dimensional cross-sectional view of the finite element model of the bolt structure is as follows: Figure 2 shown.

[0042] The preload applied to the structure is 24kN, and the tangential displacement load on the upper assembly interface is , the lower assembly interface is fixed. The average value of the contact stress distribution of the assembly interface in the dynamic service state is as follows: Figure 3 shown.

[0043] The Multi-Bp-U-Net model is constructed as follows Figure 4The average value of the dynamic service state contact stress under uniform state is input to the Multi-Bp-U-Net model, and the corresponding static ideal contact stress distribution is output as follows. Figure 5 shown.

[0044] Comparison of the topography layout output by the topography prediction model driven by the static contact stress distribution of the assembly interface and the initial topography layout. Figure 6 shown.

[0045] Comparison of the average values ​​of the contact stress distribution in the dynamic service state before and after inversion Figure 7 As shown. It can be seen that the inversion design method of the ideal prestress distribution of the assembly interface driven by the service performance effectively realizes the inversion of the static ideal contact stress distribution of the assembly interface under the uniform dynamic service state contact stress distribution, and proposes a processing guidance scheme driven by the assembly interface morphology layout. The uniformity of the dynamic service state contact stress of this invention is improved by 66.11%, and the contact stress variance is increased from 81.089MPa to 2 Down to 27.483MPa 2 ; The peak and range of pressure on the assembly interface were significantly reduced, and the average maximum contact stress was reduced from 76.366MPa to 23.987MPa.

[0046] See also Fig. 9 Another embodiment of the present invention provides an inverse design system for ideal prestress distribution of assembly interface driven by service performance, comprising: The feature mapping module is used to perform feature mapping of the assembly interface morphology-stress on the mating surface to obtain the contact pressure value and the relative coordinates of each node; A static ideal contact stress distribution inversion module is used to perform static ideal contact stress distribution inversion according to the contact pressure value and the static contact stress to obtain the static ideal contact stress distribution; The morphology layout prediction module is used to predict the morphology layout driven by the static contact stress distribution of the assembly interface according to the static ideal contact stress distribution.

[0047] Another embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the inverse design method for ideal prestress distribution of assembly interface driven by service performance when executing the computer program.

[0048] Another embodiment of the present invention provides a computer-readable storage medium storing a computer program, characterized in that: when the computer program is executed by a processor, the inverse design method for ideal prestress distribution of assembly interface driven by service performance is implemented.

[0049] The present invention reversely evolves the static ideal contact stress distribution of the assembly interface from the homogenized dynamic service state contact stress distribution, and proposes a processing guidance scheme driven by the assembly interface morphology layout, so as to improve the dynamic connection performance of mechanical equipment parts in the service state.

[0050] Deep learning technology is used to build a Multi-Bp-U-Net model to explore the intrinsic connection between the dynamic service state contact stress output by the finite element software and the static ideal contact stress, so as to achieve the global optimal solution of the contact stress distribution under the dynamic service state, predict the morphology layout under the solution state, and provide corresponding processing guidance: the prediction efficiency is much higher than that of traditional finite element calculations.

[0051] The above description is only for the best embodiment of the present invention, but it should not be understood as limiting the claims. The present invention is not limited to the above embodiments, and its specific structure is allowed to be changed. However, all changes made within the protection scope of the independent claims of the present invention are within the protection scope of the present invention.

[0052] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

Claims

1. A service performance driven assembly interface ideal prestress distribution inversion design method, characterized in that: The following steps are involved: Perform feature mapping of assembly interface morphology-stress on the mating surface to obtain the contact pressure value and the relative coordinates of each node; According to the contact pressure value and the static contact stress, the static ideal contact stress distribution is inverted to obtain the static ideal contact stress distribution; Based on the static ideal contact stress distribution, the topography layout driven by the static contact stress distribution of the assembly interface is predicted.

2. The inverse design method for ideal prestress distribution of assembly interface for service performance traction according to claim 1 is characterized in that: The characteristic mapping of the assembly interface morphology-stress on the mating surface is performed to obtain the contact pressure value and the relative coordinates of each node, including the following steps: (1.1) Divide the assembly interface into finite element meshes, construct a finite element model for assembly interface contact analysis, and apply boundary constraints and load conditions; (1.2) Perform transient finite element contact analysis and select a cycle Sampling, calculating the contact pressure value and average contact pressure of the nodes on the assembly interface during sampling; (1.3) Calculate the contact stress variance based on the average contact pressure; (1.4) Determine the contact stress variance , if satisfied , is the threshold value, or the number of optimization iterations satisfy , is the total number of nodes in the finite element calculation process of the assembly interface, the iteration is terminated, and the relative coordinates of each node are obtained.

3. The inverse design method for ideal prestress distribution of assembly interface for service performance traction according to claim 2 is characterized in that: like or , then execute step (1.5); (1.5) Set the area for active morphology adjustment, use the node relative coordinates as the optimization design variables in theoretical analysis, set the magnification factor and reduction factor, and in the new optimization iteration step, the optimization method and update method of the actively adjusted node relative coordinate values ​​are as follows: after the update, delete the current external load condition, calculate the contact stress of each node in the current morphology layout without external load based on the contact pressure value of the reference contact node, and determine the node relative coordinate value based on the contact pressure value of the reference contact node and the contact stress of the node; (1.6) Based on the node relative coordinate values ​​obtained in step (1.5), update the assembly interface contact analysis finite element model described in step (1.1).

4. The inverse design method for ideal prestress distribution of assembly interface for service performance traction according to claim 1 is characterized in that: Average contact pressure Calculated by the following formula: Where: is the total number of nodes in the finite element calculation process of the assembly interface, For the assembly interface The second sampling time The contact pressure value of each node; Contact stress variance Calculated by the following formula: 。 5. The inverse design method for ideal prestress distribution of assembly interface for service performance traction according to claim 1 is characterized in that: According to the contact pressure value and the static contact stress, the static ideal contact stress distribution is inverted to obtain the static ideal contact stress distribution, including the following steps: (2.1) The second sampling time The contact pressure value of each node is used as the dynamic service state contact stress and the static ideal contact stress Preprocess the data set and divide the preprocessed data into a training set and a validation set; (2.2) Using the U-Net model as the basic network architecture, the BP neural network is used to establish the Multi-Bp-U-Net model; (2.3) Use the training set data and the validation set data to train the Multi-Bp-U-Net model to obtain a trained Multi-Bp-U-Net model; (2.4) The trained Multi-Bp-U-Net model is used to invert the dynamic service state contact stress distribution under uniform state to obtain the static ideal contact stress distribution under uniform state.

6. The inverse design method for ideal prestress distribution of assembly interface for service performance traction according to claim 5 is characterized in that: The Multi-Bp-U-Net model includes: (2.2.1) a feature extraction module, comprising an input convolution layer and a BpUpDown layer, wherein the input convolution layer is used to extract input multidimensional data to adapt to the input dimension; and the BpUpDown is used to extract sequence data features according to the input multidimensional data adapting to the input dimension; (2.2.2) a contraction path module, comprising a plurality of BpUpDown layers, for downsampling the sequence feature information; (2.2.3) Extended path module, including multiple BpUpDown layers for upsampling processing; (2.2.4) Skip connection module, which is used to transfer low-level features of the lower layer to the upper layer by copying and splicing.

7. The inverse design method for ideal prestress distribution of assembly interface for service performance traction according to claim 1 is characterized in that: According to the static ideal contact stress distribution, the topography layout driven by the static contact stress distribution of the assembly interface is predicted, including the following steps: (3.1) Preprocessing the static ideal contact stress and relative coordinate values ​​as a data set, and dividing the preprocessed data into a training set and a validation set; (3.2) Using the U-Net model as the basic network architecture, the Multi-Bp-U-Net model is established using the fully connected layer of the BP neural network; (3.3) Using the training set data and the validation set data in step (3.1) to train the Multi-Bp-U-Net model to obtain the trained Multi-Bp-U-Net model; (3.4) The static ideal contact stress distribution is input into the trained Multi-Bp-U-Net model to predict the morphology layout and obtain the morphology layout under the static ideal contact stress distribution state.

8. A service performance driven assembly interface ideal prestress distribution inversion design system, characterized by: include: The feature mapping module is used to perform feature mapping of assembly interface morphology-stress on the mating surface to obtain the contact pressure value and the relative coordinates of each node; A static ideal contact stress distribution inversion module is used to perform static ideal contact stress distribution inversion according to the contact pressure value and the static contact stress to obtain the static ideal contact stress distribution; The morphology layout prediction module is used to predict the morphology layout driven by the static contact stress distribution of the assembly interface according to the static ideal contact stress distribution.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the inverse design method for ideal prestress distribution of the assembly interface driven by service performance as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the inverse design method for ideal prestress distribution of an assembly interface driven by service performance as described in any one of claims 1 to 7 is implemented.

Images