Methods, systems, equipment, and storage media for predicting residual stress in gear shot peening

By adopting a hierarchical dual-model architecture and utilizing the normalization of gear shot peening process parameters and the integration of feature vectors, the problem of low prediction efficiency of residual stress in gear shot peening is solved, and efficient and accurate prediction of residual stress distribution in gear shot peening is achieved.

CN122864076BActive Publication Date: 2026-11-10CENT SOUTH UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202611327948.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-08-31
Publication Date
2026-11-10
Estimated Expiration
2046-08-31

AI Technical Summary

Technical Problem

In the existing technology, the method for obtaining residual stress in gear shot peening has the problem of low efficiency. Especially in the process of parameter optimization, the test and inspection method relies on special equipment and performs point-by-point inspection, while the finite element model has a large amount of calculation, which leads to a significant increase in simulation time.

Method used

A hierarchical dual-model architecture is adopted. The first prediction model determines the impact number and impact velocity distribution of the normalized gear shot peening process parameters, constructs a fused feature vector, and inputs it into the second prediction model to predict the residual stress distribution, thereby improving prediction efficiency and accuracy.

Benefits of technology

By adopting a hierarchical dual-model architecture, the efficiency and accuracy of predicting residual stress in gear shot peening are improved, while reducing computation time and cost. It is suitable for optimizing process parameters for multiple shot types, multiple coverage areas, and complex tooth surfaces.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122864076B_ABST
    Figure CN122864076B_ABST
Patent Text Reader

Abstract

This application discloses a method, system, device, and storage medium for predicting residual stress in gear shot peening. The method includes determining a first impact parameter based on the normalized first process parameters of the gear shot to be peened, using a first prediction model. The normalized first process parameters include the normalized first shot diameter of the gear shot to be peened, and the first impact parameter includes the first number of impacts and the first impact velocity distribution value. Based on the normalized first shot diameter, the first number of impacts, and the first impact velocity distribution value, a first fused feature vector is determined. The first fused feature vector is input into a second prediction model to obtain the predicted residual stress distribution value output by the second prediction model, thereby improving the efficiency of gear shot peening residual stress prediction.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the technical field of residual stress prediction in gear shot peening, and in particular to methods, systems, equipment and storage media for predicting residual stress in gear shot peening. Background Technology

[0002] During shot peening, high-speed shot impacts the tooth surface, causing localized plastic deformation of the gear surface. After unloading, the surface is constrained by the surrounding elastic matrix, resulting in residual stress. This residual stress can effectively inhibit the initiation and propagation of fatigue cracks, thereby extending the service life of the gear.

[0003] In existing technologies, methods for obtaining residual stress in gear shot peening mainly include experimental testing methods and finite element methods. Among them, experimental testing methods directly obtain the residual stress value of the surface or near-surface layer through physical measurement. This type of method has high accuracy, but it usually relies on specialized testing equipment, sample preparation, and point-by-point testing. As a process design tool, it is more suitable for post-verification and not suitable for making a large number of rapid predictions during parameter optimization. On the other hand, the finite element model has a large computational load, especially when it is necessary to simulate multiple shot, multiple coverage ratios, complex tooth surfaces, and multiple sets of process parameters, the simulation time increases significantly, resulting in low efficiency. Summary of the Invention

[0004] This application aims to at least address the technical problems existing in the prior art. To this end, this application proposes a method, system, device, and storage medium for predicting residual stress in gear shot peening, which can improve the efficiency of predicting residual stress in gear shot peening.

[0005] A first aspect of this application provides a method for predicting residual stress in gear shot peening, comprising the following steps: Given the normalized first process parameters of the gear shot peening to be predicted, a first impact parameter is determined based on the normalized first process parameters through a first prediction model. The normalized first process parameters include the normalized first shot diameter of the gear shot peening to be predicted, and the first impact parameter includes the first number of impacts and the first impact velocity distribution value. Based on the normalized first projectile diameter, the first number of impacts, and the first impact velocity distribution value, a first fused feature vector is determined. The first fused feature vector is input into the second prediction model to obtain the predicted value of residual stress distribution output by the second prediction model.

[0006] The gear shot peening residual stress prediction method according to the embodiments of this application has at least the following beneficial effects: This application first uses normalized gear shot peening process parameters, relies on a first prediction model to obtain the corresponding impact number and impact velocity distribution of the gear shot peening to be predicted, then integrates the shot diameter and impact parameters to build a fused feature vector, and sends it to a second prediction model to predict residual stress. By using a hierarchical dual-model architecture to separately deduce the shot peening impact behavior and stress response, the efficiency and accuracy of gear shot peening residual stress prediction are improved.

[0007] A second aspect of this application provides a gear shot peening residual stress prediction system, the gear shot peening residual stress prediction system comprising: The first impact parameter determination module is used to determine the first impact parameter based on the normalized first process parameter of the gear shot peening to be predicted, through a first prediction model, when the normalized first process parameter of the gear shot peening to be predicted is determined. The normalized first process parameter includes the normalized first shot diameter of the gear shot peening to be predicted, and the first impact parameter includes the first number of impacts and the first impact velocity distribution value. The first fusion feature vector determination module is used to determine the first fusion feature vector based on the normalized first projectile diameter, the first number of impacts, and the first impact velocity distribution value. The residual stress distribution prediction module is used to input the first fused feature vector into the second prediction model to obtain the residual stress distribution prediction value output by the second prediction model.

[0008] This system first uses normalized gear shot peening process parameters, relies on the first prediction model to obtain the corresponding impact number and impact velocity distribution of the gear shot peening to be predicted, then integrates the shot diameter and impact parameters to build a fused feature vector, and sends it to the second prediction model to predict residual stress. By using a hierarchical dual-model architecture to separately deduce the shot peening impact behavior and stress response, the efficiency and accuracy of gear shot peening residual stress prediction are improved.

[0009] A third aspect of this application provides an electronic device including at least one processor and a memory for communicatively connecting to the processor; the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to cause the at least one processor to perform a gear shot peening residual stress prediction method as described in the first aspect of this application.

[0010] A fourth aspect of this application provides a computer-readable storage medium storing computer-executable instructions for causing a computer to perform a gear shot peening residual stress prediction method as described in the first aspect of this application.

[0011] It should be noted that the beneficial effects of the third and fourth aspects of this application compared with the prior art are the same as the beneficial effects of the above-mentioned method for predicting residual stress in gear shot peening compared with the prior art, and will not be described in detail here.

[0012] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0013] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1 This is a schematic flowchart of an embodiment of the gear shot peening residual stress prediction method provided in this application; Figure 2 This is a schematic diagram of an embodiment of the gear shot peening residual stress prediction system provided in this application; Figure 3 This is a schematic diagram of the structure of an embodiment of the electronic device provided in this application. Detailed Implementation

[0014] The embodiments of this application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0015] In the description of this application, the use of terms such as "first," "second," etc., is for the purpose of distinguishing technical features only and should not be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated or the order of the technical features indicated.

[0016] In the description of this application, it should be understood that the orientation descriptions, such as up, down, etc., are based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0017] In the description of this application, it should be noted that, unless otherwise explicitly defined, terms such as "setup," "installation," and "connection" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this application in conjunction with the specific content of the technical solution.

[0018] During shot peening, high-speed shot impacts the tooth surface, causing localized plastic deformation of the gear surface. After unloading, the surface is constrained by the surrounding elastic matrix, resulting in residual stress. This residual stress can effectively inhibit the initiation and propagation of fatigue cracks, thereby extending the service life of the gear.

[0019] In existing technologies, methods for obtaining residual stress in gear shot peening mainly include experimental testing methods and finite element methods. Among them, experimental testing methods directly obtain the residual stress value of the surface or near-surface layer through physical measurement. This type of method has high accuracy, but it usually relies on specialized testing equipment, sample preparation, and point-by-point testing. As a process design tool, it is more suitable for post-verification and not suitable for making a large number of rapid predictions during parameter optimization. On the other hand, the finite element model has a large computational load, especially when it is necessary to simulate multiple shot, multiple coverage ratios, complex tooth surfaces, and multiple sets of process parameters, the simulation time increases significantly, resulting in low efficiency.

[0020] To address the aforementioned technical deficiencies, embodiments of this application provide a method, system, device, and storage medium for predicting residual stress in gear shot peening.

[0021] Please see Figure 1 This is a flowchart illustrating a method for predicting residual stress in gear shot peening, provided in an embodiment of this application. This method is applied to electronic devices, such as servers. Figure 1 As shown, the method for predicting residual stress in gear shot peening includes: Step S101: Given the normalized first process parameters of the gear shot peening to be predicted, the first impact parameter is determined based on the normalized first process parameters and through the first prediction model. The normalized first process parameters include the normalized first shot diameter of the gear shot peening to be predicted, and the first impact parameter includes the first number of impacts and the first impact velocity distribution value. The aforementioned first process parameters may include the first initial velocity of the first shot, the first shot flow rate, the first nozzle moving speed, and the first shot diameter of the gear to be shot peened.

[0022] The normalized first process parameters mentioned above may also include the normalized first initial velocity of the projectile, the normalized first projectile flow rate, and the normalized first nozzle moving speed.

[0023] The first prediction model mentioned above can be a convolutional neural network model containing two prediction heads.

[0024] The above impact velocity distribution values ​​can be used to characterize the statistical distribution set of instantaneous impact velocities of each shot impacting the tooth surface of the gear to be predicted during the shot peening process.

[0025] In step S101, the normalized first process parameter for determining the gear shot peening to be predicted can be obtained by manually setting the first process parameter for the gear shot peening to be predicted, normalizing the first process parameter, and then obtaining the normalized first process parameter.

[0026] In step S101, the first impact parameter is determined by the first prediction model based on the normalized first process parameters. This can be achieved by mapping the normalized first process parameters to the corresponding first high-dimensional feature vector through a fully connected mapping layer after extracting the tooth surface position of the gear shot peening to be predicted. The tooth surface position of the gear shot peening to be predicted and the first high-dimensional feature vector are then input into the first prediction model to obtain the first impact parameter output by the first prediction model.

[0027] Step S102: Determine the first fused feature vector based on the normalized first projectile diameter, first number of impacts, and first impact velocity distribution value; Step S103: Input the first fused feature vector into the second prediction model to obtain the predicted value of residual stress distribution output by the second prediction model.

[0028] The second prediction model mentioned above can be a convolutional neural network model containing three prediction heads.

[0029] The aforementioned predicted residual stress distribution values ​​may include predicted residual stress distribution values ​​for the surface layer, the middle layer, and the deep layer. Specifically, the predicted residual stress distribution values ​​for the surface layer are used to characterize the predicted residual stress distribution values ​​within a first depth range preset according to actual needs, with the depth value along the tooth surface normal direction being the predicted residual stress distribution values ​​within a second depth range preset according to actual needs, and the predicted residual stress distribution values ​​for the deep layer are used to characterize the predicted residual stress distribution values ​​within a third depth range preset according to actual needs, with the depth value along the tooth surface normal direction being the predicted residual stress distribution values ​​within a third depth range preset according to actual needs.

[0030] This application first uses normalized gear shot peening process parameters, relies on a first prediction model to obtain the corresponding impact number and impact velocity distribution of the gear shot peening to be predicted, then integrates the shot diameter and impact parameters to build a fused feature vector, and sends it to a second prediction model to predict residual stress. By using a hierarchical dual-model architecture to separately deduce the shot peening impact behavior and stress response, the efficiency and accuracy of gear shot peening residual stress prediction are improved.

[0031] In some embodiments, the training process of the first prediction model may include steps S201 to S203: Step S201: After obtaining the second process parameters and first tooth surface positions of several historical gear shot peenings, shot peening simulation is performed using the discrete element method based on all the second process parameters and all the first tooth surface positions to obtain the second impact parameters corresponding to each historical gear shot peening, wherein the second process parameters are the process parameters corresponding to each historical gear shot peening. The aforementioned second process parameters may include the initial velocity of the second shot, the flow rate of the second shot, the moving speed of the second nozzle, and the diameter of the second shot in the historical gear shot peening.

[0032] The aforementioned second impact parameter may include the second impact number and the second impact velocity distribution value.

[0033] In step S201, after obtaining the second process parameters and first tooth surface positions of several historical gear shot peenings, shot peening simulation is performed based on all the second process parameters and all the first tooth surface positions using the discrete element method to obtain the second impact parameter corresponding to each historical gear shot peening. This can be achieved by manually setting the second process parameters and first tooth surface positions of several historical gear shot peenings, inputting all the second process parameters and all the first tooth surface positions into the discrete element simulation software, and then performing shot peening simulation using the discrete element simulation software (it should be noted that the above discrete element method is used to characterize the shot peening simulation performed by the discrete element simulation software, which can be EDEM) to obtain the second impact parameter corresponding to each historical gear shot peening.

[0034] Step S202: After normalizing the second process parameters to obtain normalized second process parameters, each normalized second process parameter is mapped to a corresponding high-dimensional feature vector through a fully connected mapping layer. The normalized second process parameters mentioned above may include the normalized second shot diameter, normalized second shot initial velocity, normalized second shot flow rate, and normalized second nozzle moving speed of the historical gear shot peening (it should be noted that a normalized second process parameter (i.e., which may include the normalized second shot diameter, normalized second shot initial velocity, normalized second shot flow rate, and normalized second nozzle moving speed mentioned above) corresponds to a high-dimensional feature vector).

[0035] Step S203: With the first initial prediction model constructed, the first initial prediction model is iteratively trained based on all high-dimensional feature vectors, all first tooth surface positions and all second impact parameters to obtain the first prediction model.

[0036] In step S203, the first initial prediction model is iteratively trained based on all high-dimensional feature vectors, all first tooth surface positions, and all second impact parameters to obtain the first prediction model. This can be achieved by inputting all high-dimensional feature vectors, all first tooth surface positions, and all second impact parameters (as the true values ​​during the training process of the first initial prediction model) into the first initial prediction model and iteratively training the first initial prediction model using the backpropagation method until the number of iterations reaches the preset number of iterations set according to actual needs, thus obtaining the first prediction model.

[0037] This application obtains impact parameter samples under historical gear shot peening conditions through discrete element method simulation. After normalizing the historical process parameters, they are converted into high-dimensional feature vectors through a fully connected mapping layer. Combined with tooth surface position information, the first initial prediction model is iteratively trained to obtain the first prediction model. This avoids the problems of high cost and long cycle caused by a large number of real shot peening physical experiments. It fully learns the nonlinear mapping relationship between process parameters, tooth surface position and shot peening impact parameters, improves the model's generalization ability to different gear shot peening conditions, ensures the reliability of impact parameter output results under subsequent prediction conditions, and lays a good data and model foundation for accurate prediction of residual stress.

[0038] In some embodiments, the training process of the second prediction model may include steps S301 to S305: Step S301: Given the determined positions of several second tooth surfaces for each historical gear shot peening, each high-dimensional feature vector is copied to obtain several copies of the copied high-dimensional feature vectors, wherein the copied high-dimensional feature vectors correspond one-to-one with the positions of the second tooth surfaces. In step S301, when several second tooth surface positions of each historical gear shot peening are determined, each high-dimensional feature vector is copied to obtain several copies of the high-dimensional feature vector. The one-to-one correspondence between the copied high-dimensional feature vector and the second tooth surface position can be achieved by dividing each historical gear shot peening area from the tooth root to the tooth tip into a uniform region according to a preset number of second tooth surface positions as required by actual needs, and then copying each high-dimensional feature vector and corresponding the copied high-dimensional feature vector with the second tooth surface position.

[0039] Step S302: Input all copied high-dimensional feature vectors and all second tooth surface positions into the first prediction model to obtain the third impact parameter corresponding to each second tooth surface position output by the first prediction model; The aforementioned third impact parameter may include the number of third impacts and the third impact velocity distribution value.

[0040] Step S303: After sorting the positions of several second tooth surfaces of each historical gear shot peening in ascending order from tooth tip to tooth root to obtain the sorted second tooth surface position sequence corresponding to each historical gear shot peening, the fourth impact parameter of each second tooth surface position is determined based on the sorted second tooth surface position sequence and the third impact parameter. Step S304: Determine the second fused feature vector based on the normalized second projectile diameter and the fourth impact parameter; The aforementioned fourth impact parameter may include the fourth impact number and the fourth impact velocity distribution value.

[0041] Step S305: With the second initial prediction model constructed, the second initial prediction model is iteratively trained based on all the second fusion feature vectors to obtain the second prediction model.

[0042] This application achieves one-to-one matching of process features with each tooth surface position by replicating high-dimensional feature vectors. It uses a first prediction model to obtain impact parameters corresponding to different tooth surface positions, and corrects the fourth impact parameter along the position sequence from tooth tip to tooth root. Then, it combines the normalized shot diameter to construct a second fused feature vector, which is used to iteratively train the second initial prediction model to obtain the second prediction model. This second prediction model can fully take into account the shot peening impact differences at different positions from the tooth tip to the tooth root, explore the spatial correlation between tooth surface position, shot conditions and impact behavior, enrich the spatial dimension of the model training samples, improve the model's learning effect on residual stress in different tooth surface regions of the gear, and effectively improve the prediction accuracy of the second prediction model in the entire tooth surface range of residual stress distribution.

[0043] In some embodiments, step S303 may include steps S401 to S402: Step S401: Sort all the third impact parameters corresponding to each historical gear shot peening in ascending order according to the number of third impacts to obtain the sorted impact number sequence corresponding to each historical gear shot peening. Step S402: Take the first sorted impact number of each sorted impact number sequence as the fourth impact parameter of the first sorted second tooth surface position in the corresponding historical gear shot peening second tooth surface position sequence; take the second sorted impact number of each sorted impact number sequence as the fourth impact parameter of the second sorted second tooth surface position in the corresponding historical gear shot peening second tooth surface position sequence, and so on, until the fourth impact parameter of each second tooth surface position is obtained.

[0044] This application sorts the third impact parameters corresponding to each historical gear shot peening in ascending order according to the number of impacts, and then matches them one by one with the tooth surface position sequence sorted from the tooth tip to the tooth root to obtain the fourth impact parameter. This achieves an ordered mapping between the impact parameters and the spatial position of the tooth surface, corrects the position matching disorder caused by the direct output of impact parameters by the model, fits the actual distribution law of the impact of shot peening on the gear tooth surface from the tooth tip to the tooth root, ensures that the correspondence between the impact parameters and the tooth surface position in the training samples is real and reasonable, improves the quality of the training samples of the subsequent second prediction model, and thus helps to improve the prediction accuracy of the residual stress distribution of the entire gear tooth surface.

[0045] In some embodiments, step S305 may include steps S501 to S502: Step S501: Based on all the fourth impact parameters, perform simulation using finite element software to obtain the first residual stress distribution value corresponding to each fourth impact parameter; In step S501, the above-mentioned simulation based on all fourth impact parameters using finite element software to obtain the first residual stress distribution value corresponding to each fourth impact parameter can be achieved by inputting all fourth impact parameters into the finite element simulation software to obtain the first residual stress distribution value corresponding to each fourth impact parameter through the finite element simulation software.

[0046] The aforementioned first residual stress distribution value may include a first surface residual stress distribution value, a first middle layer residual stress distribution value, and a first deep layer residual stress distribution value. The first surface residual stress distribution value is used to characterize the first residual stress distribution value within a first depth range preset according to actual needs, with the depth value along the tooth surface normal direction being the first residual stress distribution value within a second depth range preset according to actual needs. The first deep layer residual stress distribution value is used to characterize the first residual stress distribution value within a third depth range preset according to actual needs, with the depth value along the tooth surface normal direction being the first residual stress distribution value within a third depth range preset according to actual needs.

[0047] Step S502: Based on all the second fusion feature vectors and all the first residual stress distribution values, iteratively train the second initial prediction model to obtain the second prediction model.

[0048] This application utilizes finite element simulation to obtain the first residual stress distribution value under historical gear shot peening conditions based on the fourth impact parameter. The second initial prediction model is then iteratively trained by pairing the second fused feature vector with the residual stress distribution value output by the finite element simulation to obtain the second prediction model. By using finite element simulation to obtain label data, large-scale physical shot peening tests are unnecessary, reducing the experimental and time costs of sample acquisition. Simultaneously, it can realistically reflect the stress evolution law of the gear tooth surface under shot peening impact, improving the reliability of the training labels. This allows the second prediction model to fully learn the nonlinear mapping relationship between the fused features and the residual stress distribution, further improving the model's prediction accuracy and adaptability to working conditions for gear shot peening residual stress distribution.

[0049] In some embodiments, step S502 may include steps S601 to S602: Step S601: Input all second fused feature vectors and all first residual stress distribution values ​​into the second initial prediction model to obtain the first total loss value output by the second initial prediction model; In step S601, the above-mentioned input of all second fused feature vectors and all first residual stress distribution values ​​into the second initial prediction model to obtain the first total loss value output by the second initial prediction model can be the input of all second fused feature vectors and all first residual stress distribution values ​​into the second initial prediction model to obtain the second residual stress distribution value output by the second initial prediction model. (It should be noted that each second residual stress distribution value is used to characterize the residual stress prediction value corresponding to each fourth impact parameter. The above-mentioned second residual stress distribution value may include the second surface residual stress distribution value, the second middle layer residual stress distribution value, and the second deep layer residual stress distribution value. Among them, the second surface residual stress distribution value is used to characterize the depth value along the tooth surface normal direction. The depth value is within a first depth range preset according to actual needs (which can be taken as a value). The second residual stress distribution value, the second middle layer residual stress distribution value is used to characterize the depth along the tooth surface normal direction. The depth value is a second depth range preset according to actual needs (which can be taken as a value). The second residual stress distribution value, and the second deep residual stress distribution value, are used to characterize the depth along the tooth surface normal direction. The value is within a third depth range preset according to actual needs (which can be taken as a value). The second residual stress distribution value is used to calculate the first total loss value using the second initial prediction model, based on the first and second residual stress distribution values. The calculation process of the first total loss value includes: The difference is obtained by subtracting the corresponding second residual stress distribution value from the first residual stress distribution value, and then the difference is dimensionless to obtain the corresponding residual stress distribution difference value. The first total loss value is calculated using the following formula: ; in, The first parameter value is preset according to actual needs. For the first The sum of the differences in the distribution of all residual stresses in the layer. For the first The loss value of the layer, The first one is pre-set according to actual needs The second parameter value of the layer, This is the first total loss value.

[0050] Step S602: If the first total loss value is less than the preset loss threshold, the second initial prediction model is used as the second prediction model.

[0051] The aforementioned preset loss threshold can be a value set in advance according to actual needs.

[0052] This application calculates a first total loss value by inputting a second fused feature vector and the corresponding first residual stress distribution value into a second initial prediction model. The second prediction model is obtained by using a total loss value less than a preset loss threshold as the model convergence criterion. This objectively quantifies the deviation between the model prediction output and the real simulation label, enabling controllable judgment of the model training effect, avoiding underfitting problems, ensuring that the trained second prediction model has stable prediction capabilities, and improving the accuracy and reliability of the gear shot peening residual stress distribution prediction results.

[0053] In some embodiments, step S102 may include steps S701 to S704: Step S701: Extract the normalized feature vectors of the first projectile diameter and the first impact number, and use them as the first feature vector; In step S701, the above-mentioned extracted and normalized feature vectors of the first projectile diameter and the first impact number can be used as the first feature vectors.

[0054] Step S702: Extract the feature vector of the first impact velocity distribution value as the second feature vector; In step S702, the feature vector of the first impact velocity distribution value extracted above can be used as the second feature vector. This second feature vector can be the feature vector of the first impact velocity distribution value extracted through a convolutional neural network model.

[0055] Step S703: Concatenate the first feature vector and the second feature vector to obtain the concatenated feature vector; Step S704: Reduce the dimensionality of the concatenated feature vectors to obtain the first fused feature vector.

[0056] In step S304, the calculation process for determining the second fusion feature vector based on the normalized second projectile diameter and the fourth impact parameter is similar to the calculation process for determining the first fusion feature vector based on the normalized first projectile diameter, the first impact number, and the first impact velocity distribution value in step S102, and will not be repeated here.

[0057] This application extracts the normalized projectile diameter and the number of first impacts as the first feature vector, and extracts the first impact velocity distribution value as the second feature vector. After feature concatenation, dimensionality reduction is performed to obtain the first fused feature vector. By splitting, extracting, concatenating, and fused features related to shot peening impact with different physical meanings, this application not only fully retains the key information of projectile size, number of impacts, and impact velocity distribution, but also eliminates redundant noise in the concatenated features, reduces the feature dimension, reduces the computational burden of subsequent prediction models, alleviates the overfitting risk caused by high-dimensional features, and strengthens the feature's ability to represent the physical process of shot peening. This is beneficial to further improve the prediction accuracy of residual stress distribution in gear shot peening.

[0058] Additionally, refer to Figure 2 One embodiment of this application provides a gear shot peening residual stress prediction system, including a first impact parameter determination module 1100, a first fused feature vector determination module 1200, and a residual stress distribution prediction module 1300, wherein: The first impact parameter determination module 1100 is used to determine the first impact parameter based on the normalized first process parameter of the gear shot peening to be predicted, through a first prediction model, when the normalized first process parameter of the gear shot peening to be predicted is determined. The normalized first process parameter includes the normalized first shot diameter of the gear shot peening to be predicted, and the first impact parameter includes the first number of impacts and the first impact velocity distribution value. The first fusion feature vector determination module 1200 is used to determine the first fusion feature vector based on the normalized first projectile diameter, first number of impacts and first impact velocity distribution value. The residual stress distribution prediction module 1300 is used to input the first fused feature vector into the second prediction model to obtain the residual stress distribution prediction value output by the second prediction model.

[0059] This system first uses normalized gear shot peening process parameters, relies on the first prediction model to obtain the corresponding impact number and impact velocity distribution of the gear shot peening to be predicted, then integrates the shot diameter and impact parameters to build a fused feature vector, and sends it to the second prediction model to predict residual stress. By using a hierarchical dual-model architecture to separately deduce the shot peening impact behavior and stress response, the efficiency and accuracy of gear shot peening residual stress prediction are improved.

[0060] It should be noted that the system embodiments described above are based on the same inventive concept as the method embodiments described above. Therefore, the relevant content of the method embodiments described above is also applicable to the system embodiments described above, and will not be repeated here.

[0061] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with relevant regulations. The acquisition, storage, use and processing of data in the technical solution of this application all comply with the relevant provisions of national laws and regulations.

[0062] like Figure 3 One embodiment of this application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned method for predicting residual stress in gear shot peening. The electronic device includes: At least one memory; At least one processor; At least one program; The program is stored in memory, and the processor executes at least one program to implement a gear shot peening residual stress prediction method according to the above embodiments of this application.

[0063] Electronic devices can be any smart terminal, including mobile phones, tablets, personal digital assistants (PDAs), and in-vehicle computers.

[0064] The electronic devices according to embodiments of this application will now be described in detail.

[0065] The processor 1600 can be implemented using a general-purpose central processing unit (CPU), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 1700 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 1700 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program code is stored in the memory 1700 and is called and executed by the processor 1600 to execute a gear shot peening residual stress prediction method according to an embodiment of this application.

[0066] The input / output interface 1800 is used to implement information input and output. The communication interface 1900 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 2000 transmits information between various components of the device (e.g., processor 1600, memory 1700, input / output interface 1800, and communication interface 1900); The processor 1600, memory 1700, input / output interface 1800 and communication interface 1900 are connected to each other within the device via bus 2000.

[0067] This application embodiment also provides a storage medium, which is a computer-readable storage medium storing computer-executable instructions for causing a computer to execute the above-described gear shot peening residual stress prediction method.

[0068] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0069] The above is a detailed description of the preferred embodiments of this application. However, the embodiments of this application are not limited to the above-described implementation methods. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the embodiments of this application. All such equivalent modifications or substitutions are included within the scope defined by the claims of the embodiments of this application.

Claims

1. A method for predicting residual stress in gear shot peening, characterized in that, The method for predicting residual stress in gear shot peening includes: Given a normalized first process parameter for the gear shot peening to be predicted, a first impact parameter is determined based on the normalized first process parameter using a first prediction model. The normalized first process parameter includes the normalized first shot diameter of the gear shot peening to be predicted, and the first impact parameter includes the first number of impacts and the first impact velocity distribution value. The training process of the first prediction model includes: Given the second process parameters and first tooth surface positions of several historical gear shot peenings, shot peening simulation is performed using the discrete element method based on all the second process parameters and all the first tooth surface positions to obtain the second impact parameters corresponding to each historical gear shot peening, wherein the second process parameters are the process parameters corresponding to each historical gear shot peening. After normalizing the second process parameters to obtain normalized second process parameters, each normalized second process parameter is mapped to a corresponding high-dimensional feature vector through a fully connected mapping layer. In constructing a first initial prediction model, the first initial prediction model is iteratively trained based on all the high-dimensional feature vectors, all the first tooth surface positions, and all the second impact parameters to obtain the first prediction model. Based on the normalized first projectile diameter, the first number of impacts, and the first impact velocity distribution value, a first fused feature vector is determined. The first fused feature vector is input into the second prediction model to obtain the predicted value of residual stress distribution output by the second prediction model. The normalized second process parameter includes the normalized second shot diameter of the historical gear shot peening. The training process of the second prediction model includes: Given that several second tooth surface positions are determined for each historical gear shot peening, each high-dimensional feature vector is copied to obtain several copies of the copied high-dimensional feature vectors, wherein the copied high-dimensional feature vectors correspond one-to-one with the second tooth surface positions. All the copied high-dimensional feature vectors and all the second tooth surface positions are input into the first prediction model to obtain the third impact parameter corresponding to each second tooth surface position output by the first prediction model; In the case of sorting several second tooth surface positions of each historical gear shot peening in ascending order from tooth tip to tooth root to obtain the sorted second tooth surface position sequence corresponding to each historical gear shot peening, a fourth impact parameter is determined for each second tooth surface position based on the sorted second tooth surface position sequence and the third impact parameter. Based on the normalized second projectile diameter and the fourth impact parameter, the second fused feature vector is determined; In constructing a second initial prediction model, the second initial prediction model is iteratively trained based on all the second fused feature vectors to obtain the second prediction model.

2. The method for predicting residual stress in gear shot peening according to claim 1, characterized in that, The third impact parameter includes the third impact number and the third impact velocity distribution value. The determination of the fourth impact parameter for each second tooth surface position based on the sorted second tooth surface position sequence and the third impact parameter includes: Sort all the third impact parameters corresponding to each historical gear shot peening in ascending order according to the number of third impacts to obtain the sorted impact number sequence corresponding to each historical gear shot peening. The first sorted impact number in each sorted impact number sequence is used as the fourth impact parameter for the first sorted second tooth surface position in the corresponding historical gear shot peening sequence; the second sorted impact number in each sorted impact number sequence is used as the fourth impact parameter for the second sorted second tooth surface position in the corresponding historical gear shot peening sequence, and so on, until the fourth impact parameter for each second tooth surface position is obtained.

3. The method for predicting residual stress in gear shot peening according to claim 1, characterized in that, The step of iteratively training the second initial prediction model based on all the second fused feature vectors to obtain the second prediction model includes: Based on all the fourth impact parameters, simulation was performed using finite element software to obtain the first residual stress distribution value corresponding to each of the fourth impact parameters. Based on all the second fused feature vectors and all the first residual stress distribution values, the second initial prediction model is iteratively trained to obtain the second prediction model.

4. The method for predicting residual stress in gear shot peening according to claim 3, characterized in that, The step of iteratively training the second initial prediction model based on all the second fused feature vectors and all the first residual stress distribution values ​​to obtain the second prediction model includes: Input all the second fused feature vectors and all the first residual stress distribution values ​​into the second initial prediction model to obtain the first total loss value output by the second initial prediction model; If the first total loss value is less than a preset loss threshold, the second initial prediction model is used as the second prediction model.

5. The method for predicting residual stress in gear shot peening according to claim 1, characterized in that, The determination of the first fused feature vector based on the normalized first projectile diameter, the first number of impacts, and the first impact velocity distribution value includes: Extract the feature vectors of the normalized first projectile diameter and the first number of impacts as the first feature vector; Extract the feature vector of the first impact velocity distribution value as the second feature vector; The first feature vector and the second feature vector are concatenated to obtain the concatenated feature vector. The concatenated feature vector is dimensionality reduced to obtain the first fused feature vector.

6. A gear shot peening residual stress prediction system, characterized in that, The gear shot peening residual stress prediction system includes: The first impact parameter determination module is used to determine a first impact parameter based on the normalized first process parameters of the gear shot peening to be predicted, using a first prediction model. The normalized first process parameters include the normalized first shot diameter of the gear shot peening to be predicted. The first impact parameter includes a first number of impacts and a first impact velocity distribution value. The training process of the first prediction model includes: Given the second process parameters and first tooth surface positions of several historical gear shot peenings, shot peening simulation is performed using the discrete element method based on all the second process parameters and all the first tooth surface positions to obtain the second impact parameters corresponding to each historical gear shot peening, wherein the second process parameters are the process parameters corresponding to each historical gear shot peening. After normalizing the second process parameters to obtain normalized second process parameters, each normalized second process parameter is mapped to a corresponding high-dimensional feature vector through a fully connected mapping layer. In constructing a first initial prediction model, the first initial prediction model is iteratively trained based on all the high-dimensional feature vectors, all the first tooth surface positions, and all the second impact parameters to obtain the first prediction model. The first fusion feature vector determination module is used to determine the first fusion feature vector based on the normalized first projectile diameter, the first number of impacts, and the first impact velocity distribution value. The residual stress distribution prediction module is used to input the first fused feature vector into the second prediction model to obtain the predicted value of residual stress distribution output by the second prediction model. The normalized second process parameter includes the normalized second shot diameter of the historical gear shot peening. The training process of the second prediction model includes: Given that several second tooth surface positions are determined for each historical gear shot peening, each high-dimensional feature vector is copied to obtain several copies of the copied high-dimensional feature vectors, wherein the copied high-dimensional feature vectors correspond one-to-one with the second tooth surface positions. All the copied high-dimensional feature vectors and all the second tooth surface positions are input into the first prediction model to obtain the third impact parameter corresponding to each second tooth surface position output by the first prediction model; In the case of sorting several second tooth surface positions of each historical gear shot peening in ascending order from tooth tip to tooth root to obtain the sorted second tooth surface position sequence corresponding to each historical gear shot peening, a fourth impact parameter is determined for each second tooth surface position based on the sorted second tooth surface position sequence and the third impact parameter. Based on the normalized second projectile diameter and the fourth impact parameter, the second fused feature vector is determined; In constructing a second initial prediction model, the second initial prediction model is iteratively trained based on all the second fused feature vectors to obtain the second prediction model.

7. An electronic device, characterized in that, It includes at least one processor and a memory for communicatively connecting with the processor; the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to cause the at least one processor to perform a gear shot peening residual stress prediction method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions for causing a computer to perform a gear shot peening residual stress prediction method as described in any one of claims 1 to 5.

Citation Information

Patent Citations

  • Gear shot blasting tooth surface residual stress calculation method and system

    CN121118507A

  • Coupling plasticity and friction heat generation gear shot peening residual stress prediction method, system, equipment and medium

    CN122347031A