A method for predicting residual stress in automobile suspension spring production

Abstract

The present invention relates to the field of design optimization technology, specifically a method for predicting residual stress in the production of automobile suspension springs. The present invention first inputs the first spring characteristic parameter and the current spring shot peening parameter of the automobile suspension spring into a shot peening prediction integrated model, predicts the second spring characteristic parameter of the spring after shot peening, and determines whether the current spring shot peening parameter needs to be adjusted; secondly, if the current spring shot peening parameter needs to be adjusted, the spring shot peening parameters in the spring shot peening parameter database are screened to obtain a compliant spring shot peening parameter set, and obtain the spring shot peening comprehensive value corresponding to each of the spring shot peening parameters in the compliant spring shot peening parameter set; finally, according to the spring shot peening comprehensive value, the optimal spring shot peening parameter is obtained, and the spring is shot peened. Through this method, the present invention can accurately optimize the spring shot peening parameters in combination with various characteristics in the automobile suspension spring shot peening process.

Description

Technical Field

[0001] The invention relates to the technical field of design optimization, in particular to a method for predicting residual stress in the production of automobile suspension springs. Background Art

[0002] In automobile manufacturing, suspension springs are important components for load bearing and shock absorption, and their performance is directly related to the stability, safety and comfort of the entire vehicle. Suspension springs are usually made of steel or alloy materials through precision processing, which includes multiple process steps such as cold drawing, forging and heat treatment. These processes not only determine the geometric shape and mechanical properties of the suspension springs, but also introduce residual stress inside the spring material.

[0003] Residual stress refers to the stress that still exists in the material even after the external load is removed during the material processing. This stress is generated during processing, cooling and heat treatment. Residual stress is divided into residual tensile stress and residual compressive stress. Residual tensile stress usually weakens the fatigue resistance and crack growth resistance of the material, while residual compressive stress usually helps to improve the fatigue strength and crack growth resistance of the material.

[0004] For automobile suspension springs, the shot peening operation in the production of automobile suspension springs will produce residual compressive stress on the surface of the spring to reduce the influence of the surface residual tensile stress generated in the production process before the shot peening operation. Therefore, accurately predicting and controlling the surface residual stress of the spring after the shot peening operation in the production process is an important means to improve the performance of automobile suspension springs, extend the service life and ensure safety. However, in the current production process of automobile suspension springs, there are still some technical bottlenecks and challenges in the prediction and control of surface residual stress. The existing predictions of surface residual stress generated in the shot peening process are mostly based on shot peening parameters, while ignoring the influence of the characteristics of the spring itself on the surface residual stress, and the shot peening parameters are not controlled in combination with the surface roughness of the spring after shot peening, resulting in the inability to accurately optimize the shot peening parameters, which ultimately affects the quality and consistency of the product.

[0005] Therefore, a method for predicting residual stress in automobile suspension spring production is proposed. Summary of the invention

[0006] The purpose of the present invention is to provide a method for predicting residual stress in the production of automobile suspension springs. First, the first spring characteristic parameter of the automobile suspension spring and the current spring shot peening parameter are input into a shot peening prediction integrated model to predict the second spring characteristic parameter of the spring after shot peening, and to determine whether the current spring shot peening parameter needs to be adjusted; secondly, if the current spring shot peening parameter needs to be adjusted, the spring shot peening parameters in the spring shot peening parameter database are screened to obtain a compliant spring shot peening parameter set, and the spring shot peening comprehensive value corresponding to each of the spring shot peening parameters in the compliant spring shot peening parameter set is obtained; finally, according to the spring shot peening comprehensive value, the optimal spring shot peening parameter is obtained, and the spring is shot peened.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A method for predicting residual stress in automobile suspension spring production includes:

[0009] S10, obtaining a first spring characteristic parameter and a current spring shot peening parameter of a vehicle suspension spring;

[0010] S20, combining the first spring characteristic parameter and the current spring shot peening parameter, obtaining an integrated prediction value of a second spring characteristic parameter of the spring after shot peening through a shot peening prediction integrated model; the second spring characteristic parameter includes the surface residual stress, residual compressive stress layer thickness, surface roughness and shot peening coverage of the automobile suspension spring after shot peening;

[0011] S30, judging whether the current spring shot peening parameters need to be adjusted according to the second spring characteristic parameter integrated prediction value, and executing S40 when it is determined that the spring shot peening parameters do not need to be adjusted, otherwise, executing S50 to S70;

[0012] S40, taking the current spring shot peening parameters as the optimal spring shot peening parameters, and jumping to S70;

[0013] S50, screening the spring shot peening parameters in the spring shot peening parameter database to obtain a compliant spring shot peening parameter set;

[0014] S60, according to the spring shot peening parameters in the compliant spring shot peening parameter set, the optimal spring shot peening parameters are obtained; the specific steps include: according to each of the spring shot peening parameters, the corresponding shot peening compliance prediction value and shot peening energy consumption value are obtained; according to the shot peening compliance prediction value and the shot peening energy consumption value, the spring shot peening comprehensive value corresponding to each of the spring shot peening parameters is obtained; the spring shot peening parameter corresponding to the largest spring shot peening comprehensive value is used as the optimal spring shot peening parameter;

[0015] S70, performing shot peening operation on the spring according to the optimal spring shot peening parameters.

[0016] Furthermore, the first spring characteristic parameter of the automobile suspension spring includes hardness, material diameter, yield strength and elastic modulus of the automobile suspension spring.

[0017] Furthermore, the spring shot peening parameters include shot peening hardness, shot peening diameter, shot peening distance, shot peening air pressure, shot peening time and shot peening incident angle.

[0018] Further, the S20 includes: inputting the spring characteristic parameter and the current spring shot peening parameter into a trained shot peening prediction integrated model to obtain a second spring characteristic parameter integrated prediction value of the spring;

[0019] The shot peening prediction integrated model includes a parameter input unit, a prediction unit and an integrated result output unit;

[0020] The parameter input unit is used to input the spring characteristic parameters and the current spring shot peening parameters into the shot peening detection integrated model; the prediction unit includes 4*N trained sub-prediction models; the 4*N trained sub-prediction models obtain N surface residual stress prediction values, N residual compressive stress layer thickness prediction values, N surface roughness prediction values ​​and N shot peening coverage prediction values; the prediction values ​​output by the 4*N sub-prediction models are input into the integrated result output unit to obtain the integrated prediction value of the surface residual stress, the integrated prediction value of the residual compressive stress layer thickness, the integrated prediction value of the surface roughness and the integrated prediction value of the shot peening coverage of the spring after shot peening.

[0021] Furthermore, the integrated result output unit includes:

[0022] The integrated result output unit includes a surface residual stress integration module, a residual compressive stress layer thickness integration module, a surface roughness integration module and a shot peening coverage integration module; the surface residual stress integration module is used to integrate N surface residual stress prediction values ​​to obtain a surface residual stress integrated prediction value; the residual compressive stress layer thickness integration module is used to integrate N residual compressive stress layer thickness prediction values ​​to obtain a residual compressive stress layer thickness integrated prediction value; the surface roughness integration module is used to integrate N surface roughness prediction values ​​to obtain a surface roughness integrated prediction value; the shot peening coverage integration module is used to integrate N shot peening coverage prediction values ​​to obtain a shot peening coverage integrated prediction value;

[0023] The calculation formula of the integrated prediction value of the surface residual stress is: ;when , expressed as surface residual compressive stress; when , expressed as surface residual tensile stress;

[0024] in, It is expressed as the integrated predicted value of surface residual stress; It represents the predicted value of surface residual stress output by the i-th sub-prediction model; It is represented as the weight coefficient of the surface residual stress prediction value output by the i-th sub-prediction model;

[0025] The calculation formula of the weight coefficient is:

[0026] ;

[0027] in, It is represented as the average value of the surface residual stress prediction accuracy of the i-th sub-prediction model on the test set; It is represented as the variance of the prediction accuracy of the surface residual stress of the ith sub-prediction model on the test set; and They are respectively expressed as the prediction accuracy mean coefficient and the prediction accuracy variance coefficient; It is expressed as an exponential function with a natural constant as base;

[0028] The calculation formulas for the integrated prediction value of the residual compressive stress layer thickness, the integrated prediction value of the surface roughness and the integrated prediction value of the shot peening coverage rate may refer to the calculation formula for the integrated prediction value of the surface residual stress.

[0029] Further, the S30 includes: when the second spring characteristic parameter integrated prediction value meets the spring shot peening standard, calculating the shot peening compliance prediction value according to the second spring characteristic parameter integrated prediction value; otherwise, determining that the current spring shot peening parameters need to be adjusted; the calculation formula of the shot peening compliance prediction value is:

[0030] ;

[0031] in, It is expressed as the predicted value of shot peening compliance; It is expressed as the integrated predicted value of surface residual stress; Expressed as the set surface residual stress standard value, ; It is expressed as the integrated predicted value of residual compressive stress layer thickness; It is expressed as the integrated predicted value of surface roughness; It is expressed as the integrated prediction value of shot peening coverage; , , and They are expressed as residual stress coefficient, residual compressive stress layer thickness coefficient, roughness coefficient and shot peening coverage coefficient respectively;

[0032] When the shot peening compliance prediction value is greater than the set shot peening compliance standard value, it is determined that the current shot peening parameters do not need to be adjusted; otherwise, it is determined that the current spring shot peening parameters need to be adjusted.

[0033] Furthermore, the spring shot peening standard includes: , , , When both are satisfied, it indicates that the second spring characteristic parameters meet the spring shot peening standard; otherwise, it indicates that the second spring characteristic parameters do not meet the spring shot peening standard; wherein, , , and They are respectively expressed as the integrated prediction value of surface residual stress, the integrated prediction value of residual compressive stress layer thickness, the integrated prediction value of surface roughness and the integrated prediction value of shot peening coverage; , , and They are respectively represented by the set standard value of surface residual stress, standard value of residual compressive stress layer thickness, standard value of surface roughness and standard value of shot peening coverage.

[0034] Furthermore, the screening of spring shot peening parameters to obtain a compliant spring shot peening parameter set includes:

[0035] All the spring shot peening parameters in the spring shot peening parameter database are input into the shot peening prediction integrated model to obtain the integrated prediction value of the second spring characteristic parameter corresponding to each of the spring shot peening parameters; when the integrated prediction value of the second spring characteristic parameter meets the spring shot peening standard and the corresponding shot peening compliance prediction value is greater than the set shot peening compliance standard value, the spring shot peening parameter is classified into the compliant spring shot peening parameter set.

[0036] Furthermore, the calculation formula of the spring shot peening comprehensive value is:

[0037] ;

[0038] in, It is represented by the spring shot peening comprehensive value corresponding to the spring shot peening parameter of the jth item in the compliant spring shot peening parameter set; It is represented by the shot peening compliance prediction value corresponding to the spring shot peening parameters described in item j; It is expressed as the shot peening energy consumption value corresponding to the spring shot peening parameters described in the jth item; and They are respectively expressed as shot peening compliance coefficient and shot peening energy consumption coefficient.

[0039] Furthermore, the calculation formula of the shot peening energy consumption value is:

[0040] ;

[0041] in, It is represented by the shot peening time in the spring shot peening parameter of the jth item in the compliant spring shot peening parameter set; It is expressed as the shot peening gas pressure in the spring shot peening parameters described in Article j; It is expressed as the shot peening diameter in the spring shot peening parameters described in Article j; , , and They are respectively expressed as shot peening time parameter, shot peening air pressure parameter, shot peening diameter parameter and comprehensive energy consumption coefficient.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] 1. The present invention constructs a shot peening prediction integrated model. The shot peening prediction integrated model obtains the integrated prediction value of the second spring characteristic parameter output by multiple sub-prediction models through the prediction unit for the input first spring characteristic parameter and spring shot peening parameter, and then the prediction value output by different sub-prediction models is weighted through the integrated result output unit, and the spring shot peening parameter is obtained and applied to the automobile suspension spring corresponding to the first spring characteristic parameter to obtain the integrated prediction value of the second spring characteristic parameter. This method can combine the spring's own characteristics and spring shot peening parameters to achieve accurate prediction of the second spring characteristic parameters by multiple sub-prediction models, and integrate the prediction capabilities of different sub-prediction models to accurately obtain the integrated prediction value of the second spring characteristic parameter, and then accurately optimize the spring shot peening parameters.

[0044] 2. The present invention compares the integrated prediction values ​​of each parameter in the second spring characteristic parameters corresponding to the spring shot peening parameters separately, and obtains the shot peening compliance prediction value by combining the integrated prediction values ​​of each parameter; judges whether the current spring shot peening parameters need to be adjusted based on the results of the separate comparison of the integrated prediction values ​​of each parameter and the shot peening compliance prediction value. This method can more comprehensively evaluate the compliance of the second spring characteristic parameters of the automobile suspension spring after shot peening, and then accurately optimize the spring shot peening parameters.

[0045] 3. When the current spring shot peening parameters are unqualified, the present invention screens the spring shot peening parameters in the spring shot peening parameter database to obtain a compliant spring shot peening parameter set; combining the shot peening compliance prediction value and shot peening energy consumption value corresponding to each spring shot peening parameter in the compliant spring shot peening parameter set, the optimal spring shot peening parameters are obtained. This method can combine the influence of spring shot peening parameters on the performance of automobile suspension springs after shot peening and the shot peening energy consumption value corresponding to the spring shot peening parameters, thereby achieving the goal of reducing the energy consumption value while ensuring the performance of automobile suspension springs, improving the shot peening efficiency of automobile suspension springs, and then accurately optimizing the spring shot peening parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A flow chart of a method for predicting residual stress in automobile suspension spring production provided by an embodiment of the present invention;

[0047] Figure 2 An interface diagram of a vehicle suspension spring shot peening parameter optimization system provided by an embodiment of the present invention;

[0048] Figure 3 A schematic diagram of the structure of the shot peening prediction integrated model provided by an embodiment of the present invention;

[0049] Figure 4 A flow chart for determining whether the current spring shot peening parameters need to be adjusted is provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] The present invention provides a method for predicting residual stress in automobile suspension spring production. The specific method flow chart is shown in Figure 1 , the technical solution is as follows:

[0052] S10, obtaining a first spring characteristic parameter and a current spring shot peening parameter of a vehicle suspension spring;

[0053] S20, combining the first spring characteristic parameter and the current spring shot peening parameter, obtaining an integrated prediction value of a second spring characteristic parameter of the spring after shot peening through a shot peening prediction integrated model; the second spring characteristic parameter includes the surface residual stress, residual compressive stress layer thickness, surface roughness and shot peening coverage of the automobile suspension spring after shot peening;

[0054] S30, judging whether the current spring shot peening parameters need to be adjusted according to the second spring characteristic parameter integrated prediction value, and executing S40 when it is determined that the spring shot peening parameters do not need to be adjusted, otherwise, executing S50 to S70;

[0055] S40, taking the current spring shot peening parameters as the optimal spring shot peening parameters, and jumping to S70;

[0056] S50, screening the spring shot peening parameters in the spring shot peening parameter database to obtain a compliant spring shot peening parameter set;

[0057] S60, according to the spring shot peening parameters in the compliant spring shot peening parameter set, the optimal spring shot peening parameters are obtained; the specific steps include: according to each of the spring shot peening parameters, the corresponding shot peening compliance prediction value and shot peening energy consumption value are obtained; according to the shot peening compliance prediction value and the shot peening energy consumption value, the spring shot peening comprehensive value corresponding to each of the spring shot peening parameters is obtained; the spring shot peening parameter corresponding to the largest spring shot peening comprehensive value is used as the optimal spring shot peening parameter;

[0058] S70, performing shot peening operation on the spring according to the optimal spring shot peening parameters.

[0059] The characteristics of automobile suspension springs after shot peening are not only related to the shot peening parameters but also to the characteristics of the springs themselves. This application combines the characteristics of the springs themselves with the spring shot peening parameters, and predicts the characteristic parameters of the second spring after shot peening through a shot peening prediction integrated model; at the same time, combined with the energy consumption value during the shot peening process, the spring shot peening parameters are optimized. This method can accurately optimize the spring shot peening parameters by combining various characteristics during the shot peening process of automobile suspension springs.

[0060] This application applies a method for predicting residual stress in automobile suspension spring production to an automobile suspension spring shot peening parameter optimization system. The interface diagram of this system is shown in FIG. Figure 2 , the system includes a parameter acquisition unit, a second spring characteristic parameter prediction unit, a spring shot peening parameter adjustment judgment unit, an optimal spring shot peening parameter acquisition unit and a shot peening unit;

[0061] The parameter acquisition unit is used to acquire a first spring characteristic parameter and a current spring shot peening parameter of a vehicle suspension spring; the second spring characteristic parameter prediction unit is used to combine the first spring characteristic parameter and the current spring shot peening parameter, and acquire an integrated prediction value of a second spring characteristic parameter of the spring after shot peening through a shot peening prediction integrated model; the spring shot peening parameter adjustment judgment unit is used to judge whether the current spring shot peening parameter needs to be adjusted according to the second spring characteristic parameter integrated prediction value, and when it is determined that the spring shot peening parameter does not need to be adjusted, the current spring shot peening parameter is input into the shot peening unit as the optimal spring shot peening parameter, otherwise, the optimal spring shot peening parameter is acquired through the optimal spring shot peening parameter acquisition unit; the optimal spring shot peening parameter acquisition unit is used to screen the spring shot peening parameters in the spring shot peening parameter database, acquire a compliant spring shot peening parameter set, and acquire the optimal spring shot peening parameters based on the spring shot peening parameters in the compliant spring shot peening parameter set; the shot peening unit is used to perform a shot peening operation on the spring according to the optimal spring shot peening parameters.

[0062] Embodiment 1

[0063] Reference Figure 1In S10, the present invention collects first spring characteristic parameters and current spring shot peening parameters of two automobile suspension springs produced by Factory A;

[0064] Furthermore, the first spring characteristic parameters of the automobile suspension spring include the hardness, material diameter, yield strength and elastic modulus of the automobile suspension spring; the material diameter is the average diameter of the steel wire in the automobile suspension spring. In this embodiment, the first spring characteristic parameters of two automobile suspension springs are obtained, and specific reference is made to Table 1.

[0065] Table 1 Table of characteristic parameters of the first spring of the automobile suspension spring in Example 1

[0066]

[0067] The spring's own characteristics will also affect the characteristic parameter value of the second spring after spring shot peening. In this embodiment, this step obtains the hardness, material diameter, yield strength and elastic modulus of the automobile suspension spring, providing a data basis for the subsequent prediction of the characteristic parameter value of the second spring.

[0068] Furthermore, the spring shot peening parameters include shot peening hardness, shot peening diameter, shot peening distance, shot peening air pressure, shot peening time and shot peening incident angle. This embodiment obtains the current spring shot peening parameters corresponding to two automobile suspension springs, and specifically refers to Table 2.

[0069] Table 2 Current spring shot peening parameter table corresponding to the automobile suspension spring of Example 1

[0070]

[0071] The difference in spring shot peening parameters will directly affect the characteristic parameter value of the second spring after the automobile suspension spring is shot peened. In this embodiment, the spring shot peening parameters are obtained in this step, which can provide a data basis for the subsequent prediction of the characteristic parameter value of the second spring.

[0072] Reference Figure 1 In S20, combining the first spring characteristic parameter and the current spring shot peening parameter, an integrated prediction value of a second spring characteristic parameter of the spring after shot peening is obtained through a shot peening prediction integrated model.

[0073] Furthermore, the second spring characteristic parameters include surface residual stress, residual compressive stress layer thickness, surface roughness and shot peening coverage of the automobile suspension spring after shot peening.

[0074] Further, the step of predicting the second spring characteristic parameter of the spring after shot peening comprises:

[0075] Input the spring characteristic parameter and the current spring shot peening parameter into the trained shot peening prediction integrated model to obtain the second spring characteristic parameter integrated prediction value of the spring;

[0076] Reference Figure 3 , the shot peening prediction integrated model includes a parameter input unit, a prediction unit and an integrated result output unit;

[0077] The parameter input unit is used to input the spring characteristic parameters and the current spring shot peening parameters into the shot peening detection integrated model; the prediction unit includes 4*N trained sub-prediction models; the 4*N trained sub-prediction models include N sub-prediction models for surface residual stress prediction, N sub-prediction models for residual compressive stress layer thickness prediction, N sub-prediction models for surface roughness prediction and N sub-prediction models for shot peening coverage prediction; according to the 4*N trained sub-prediction models, N surface residual stress prediction values, N residual compressive stress layer thickness prediction values, N surface roughness prediction values ​​and N shot peening coverage prediction values ​​are obtained; the prediction values ​​output by the 4*N sub-prediction models are input into the integrated result output unit to obtain the integrated prediction value of the surface residual stress, the integrated prediction value of the residual compressive stress layer thickness, the integrated prediction value of the surface roughness and the integrated prediction value of the shot peening coverage of the spring after shot peening. The N is 3 or 4, etc., and in this embodiment, N is 4; in this embodiment, the four sub-prediction models for surface residual stress prediction include a BP neural network model, a Transformer model, a support vector machine model and XGBoost; the four sub-prediction models for residual compressive stress layer thickness prediction include a BP neural network model, a Transformer model, a support vector machine model and XGBoost; the four sub-prediction models for surface roughness prediction include a BP neural network model, a Transformer model, a support vector machine model and XGBoost; the four sub-prediction models for shot peening coverage prediction include a BP neural network model, a Transformer model, a support vector machine model and XGBoost.

[0078] Furthermore, the training process of the shot peening prediction integrated model includes:

[0079] A historical shot peening data set is obtained; the historical shot peening data set includes M historical shot peening data; each of the historical shot peening data includes a first spring characteristic parameter of an automobile suspension spring, a corresponding spring shot peening parameter, and a second spring characteristic parameter measured after the shot peening is completed; wherein, when the surface residual stress is compressive stress, the stress is negative; when the surface residual stress is tensile stress, the stress sign is positive; the parameter input unit divides the historical shot peening data set to obtain a historical surface residual stress data set, a historical residual compressive stress layer thickness data set, a historical surface roughness data set, and a historical shot peening coverage data set;

[0080] The historical surface residual stress data set includes M historical surface residual stress data; each of the historical surface residual stress data includes the first spring characteristic parameter of the automobile suspension spring, the corresponding spring shot peening parameter and the surface residual stress measured after the shot peening is completed; the historical residual compressive stress layer thickness data set includes M historical residual compressive stress layer thickness data; each of the historical residual compressive stress layer thickness data includes the first spring characteristic parameter of the automobile suspension spring, the corresponding spring shot peening parameter and the residual compressive stress layer thickness measured after the shot peening is completed; the historical surface roughness data set includes M historical surface roughness data; each of the historical surface roughness data includes the first spring characteristic parameter of the automobile suspension spring, the corresponding spring shot peening parameter and the surface roughness measured after the shot peening is completed; the historical shot peening coverage rate data set includes M historical shot peening coverage rate data; each of the historical shot peening coverage rate data includes the first spring characteristic parameter of the automobile suspension spring, the corresponding spring shot peening parameter and the shot peening coverage rate measured after the shot peening is completed;

[0081] The sub-prediction model for surface residual stress prediction is trained using the historical surface residual stress dataset; the sub-prediction model for residual compressive stress layer thickness prediction is trained using the historical residual compressive stress dataset; the sub-prediction model for surface roughness prediction is trained using the historical surface roughness dataset; the sub-prediction model for shot peening coverage prediction is trained using the historical shot peening coverage dataset; when the loss function values ​​corresponding to each of the sub-prediction models converge, the training of the shot peening prediction integrated model is completed, and the trained shot peening prediction integrated model is obtained.

[0082] The process of training a sub-prediction model for surface residual stress prediction using the historical surface residual stress dataset includes:

[0083] The historical surface residual stress data set is divided into a training set and a test set; the sub-prediction model is trained with the historical surface residual stress data in the training set, and the prediction effect of the sub-prediction model is verified with the test set; the loss function value corresponding to the sub-prediction model after each round of training is calculated; when the loss function value converges, it means that the training is completed, and the trained sub-prediction model for surface residual stress prediction is obtained; the sub-prediction model includes a BP neural network model, a Transformer model, a support vector machine model and XGBoost;

[0084] The calculation formula of the loss function value is:

[0085] ;

[0086] in, It is represented by the loss function value corresponding to the qth round of training of the sub-prediction model; It is expressed as the number of historical surface residual stress data in the test set; It represents the predicted value of surface residual stress corresponding to the k-th historical surface residual stress data on the test set obtained by the sub-prediction model after the q-th round of training; It is expressed as the actual value of the surface residual stress corresponding to the kth historical surface residual stress data.

[0087] Furthermore, the training process of the sub-prediction model for residual compressive stress layer thickness prediction, the sub-prediction model for surface roughness prediction and the sub-prediction model for shot peening coverage prediction can refer to the training process of the sub-prediction model for surface residual stress prediction.

[0088] This step of the present embodiment constructs a shot peening prediction integrated model. The shot peening prediction integrated model obtains the integrated prediction value of the second spring characteristic parameter output by multiple sub-prediction models through the prediction unit for the input first spring characteristic parameter and spring shot peening parameter, and then the prediction value output by different sub-prediction models is weighted through the integrated result output unit, and the spring shot peening parameter is applied to the automobile suspension spring corresponding to the first spring characteristic parameter to obtain the integrated prediction value of the second spring characteristic parameter. This method can combine the spring's own characteristics and spring shot peening parameters to achieve accurate prediction of the second spring characteristic parameter by multiple sub-prediction models, and integrate the prediction capabilities of different sub-prediction models to accurately obtain the integrated prediction value of the second spring characteristic parameter.

[0089] Furthermore, the integrated result output unit includes:

[0090] The integrated result output unit includes a surface residual stress integration module, a residual compressive stress layer thickness integration module, a surface roughness integration module and a shot peening coverage integration module; the surface residual stress integration module is used to integrate N surface residual stress prediction values ​​to obtain a surface residual stress integrated prediction value; the residual compressive stress layer thickness integration module is used to integrate N residual compressive stress layer thickness prediction values ​​to obtain a residual compressive stress layer thickness integrated prediction value; the surface roughness integration module is used to integrate N surface roughness prediction values ​​to obtain a surface roughness integrated prediction value; the shot peening coverage integration module is used to integrate N shot peening coverage prediction values ​​to obtain a shot peening coverage integrated prediction value;

[0091] The calculation formula of the integrated prediction value of the surface residual stress is: ;when , expressed as surface residual compressive stress; when , expressed as surface residual tensile stress;

[0092] in, It is expressed as the integrated predicted value of surface residual stress; It represents the predicted value of surface residual stress output by the i-th sub-prediction model; It is represented as the weight coefficient of the surface residual stress prediction value output by the i-th sub-prediction model;

[0093] The calculation formula of the weight coefficient is:

[0094] ;

[0095] in, It is represented as the average value of the surface residual stress prediction accuracy of the i-th sub-prediction model on the test set; It is represented as the variance of the prediction accuracy of the surface residual stress of the ith sub-prediction model on the test set; It is expressed as an exponential function with a natural constant as base; and They are respectively expressed as the prediction accuracy mean coefficient and the prediction accuracy variance coefficient; and The default values ​​are all 0.5, and can be modified according to actual applications; the calculation formulas for the integrated prediction value of the residual compressive stress layer thickness, the integrated prediction value of the surface roughness, and the integrated prediction value of the shot peening coverage rate can refer to the calculation formula for the integrated prediction value of the surface residual stress.

[0096] In this step of the present embodiment, different weights are assigned to the sub-prediction models according to the mean and variance of the prediction accuracy of different sub-prediction models in the test. This method not only takes into account the prediction accuracy of the sub-prediction model, but also takes into account the prediction stability of the sub-prediction model, and then assigns corresponding weights to improve the prediction effect of the shot peening prediction integrated model.

[0097] Furthermore, the present embodiment obtains first spring characteristic parameters, corresponding spring shot peening parameters, and corresponding second spring characteristic parameters after shot peening of five automobile suspension springs;

[0098] The first spring characteristic parameters and corresponding spring shot peening parameters of five automobile suspension springs are input into the shot peening prediction integrated model to obtain the integrated prediction value of the second spring characteristic parameter; the integrated prediction value of the second spring characteristic parameter corresponding to the five automobile suspension springs is compared with the actual value of the second spring characteristic parameter to obtain the prediction accuracy, and the calculation formula is:

[0099] ;

[0100] in, It is expressed as the prediction accuracy of the shot peening prediction integrated model for the vth parameter in the second spring characteristic parameter of the uth automobile suspension spring; the value range of u is 1 to 5; the value range of v is 1 to 4; v=1 corresponds to the surface residual stress in the second spring characteristic parameter; v=2 corresponds to the thickness of the residual compressive stress layer in the second spring characteristic parameter; v=3 corresponds to the surface roughness in the second spring characteristic parameter; v=4 corresponds to the shot peening coverage in the second spring characteristic parameter; It represents the integrated prediction value of the vth parameter in the second spring characteristic parameter of the uth automobile suspension spring output by the shot peening prediction integrated model; It is represented as the actual value of the vth parameter in the second spring characteristic parameter of the uth automobile suspension spring. The prediction accuracies corresponding to the four parameters in the second spring characteristic parameter are averaged to obtain the average prediction accuracies corresponding to the four parameters; in order to verify the effectiveness of the shot peening prediction integrated model, this embodiment also selected three prediction models to predict the second spring characteristic parameters corresponding to five automobile suspension springs, and obtained the average prediction accuracies corresponding to the four parameters in the second spring characteristic parameter. For specific results, refer to Table 3.

[0101] Table 3 Example 1 Shot Peening Prediction Integrated Model Prediction Effect Verification Table

[0102]

[0103] In Table 3, the model number 1 is the shot peening prediction integrated model constructed by the present invention; the model number 2 is based on the model number 1, and only the spring shot peening parameters are used as the input of the model; the model number 3 is based on the model number 2, each sub-prediction model is integrated with equal weights; the model number 4 is the BP neural network used to predict the characteristic parameters of the second spring. It can be seen from Table 3 that the shot peening prediction integrated model provided by the present invention can effectively improve the prediction accuracy of the characteristic parameters of the second spring.

[0104] Reference Figure 1 In S30, it is determined whether the current spring shot peening parameters need to be adjusted according to the second spring characteristic parameters.

[0105] Further, refer to Figure 4 When the second spring characteristic parameter integrated prediction value meets the spring shot peening standard, the shot peening compliance prediction value is calculated according to the second spring characteristic parameter integrated prediction value; otherwise, it is determined that the current spring shot peening parameters need to be adjusted; the calculation formula of the shot peening compliance prediction value is:

[0106] ;

[0107] in, It is expressed as the predicted value of shot peening compliance; It is expressed as the integrated predicted value of surface residual stress; Expressed as the set surface residual stress standard value, ; It is expressed as the integrated predicted value of residual compressive stress layer thickness; It is expressed as the integrated predicted value of surface roughness; It is expressed as the integrated prediction value of shot peening coverage; , , and They are expressed as residual stress coefficient, residual compressive stress layer thickness coefficient, roughness coefficient and shot peening coverage coefficient respectively; , , and The default value is 0.25, which can be modified according to the actual application;

[0108] When the shot peening compliance prediction value is greater than the set shot peening compliance standard value, it is determined that the current shot peening parameters do not need to be adjusted; otherwise, it is determined that the current spring shot peening parameters need to be adjusted.

[0109] In this embodiment, this step is to compare the integrated prediction values ​​of each parameter in the second spring characteristic parameter corresponding to the spring shot peening parameter separately, and to obtain the shot peening compliance prediction value in combination with the integrated prediction values ​​of each parameter; and to judge whether the current spring shot peening parameters need to be adjusted based on the results of the separate comparison of the integrated prediction values ​​of each parameter and the shot peening compliance prediction value. This method can more comprehensively evaluate the compliance of the second spring characteristic parameters of the automobile suspension spring after shot peening.

[0110] Furthermore, the spring shot peening standard includes: , , , When both are satisfied, it indicates that the second spring characteristic parameters meet the spring shot peening standard; otherwise, it indicates that the second spring characteristic parameters do not meet the spring shot peening standard; wherein, , , and They are respectively expressed as the integrated prediction value of surface residual stress, the integrated prediction value of residual compressive stress layer thickness, the integrated prediction value of surface roughness and the integrated prediction value of shot peening coverage; , , and They are respectively represented by the set standard value of surface residual stress, standard value of residual compressive stress layer thickness, standard value of surface roughness and standard value of shot peening coverage.

[0111] This step of the present embodiment compares the various parameter values ​​in the characteristic parameters of the second spring with the corresponding standard values. This method can ensure that the various parameter values ​​in the characteristic parameters of the second spring after shot peening meet the basic standards and avoid that a single parameter is ignored during comprehensive comparison.

[0112] Reference Figure 1 In S40 and S50, when it is determined that the spring shot peening parameters do not need to be adjusted, the current spring shot peening parameters are used as the optimal spring shot peening parameters, and the process jumps to S70; otherwise, the spring shot peening parameters in the spring shot peening parameter database are screened to obtain a compliant spring shot peening parameter set.

[0113] Furthermore, the screening of spring shot peening parameters to obtain a compliant spring shot peening parameter set includes:

[0114] All the spring shot peening parameters in the spring shot peening parameter database are input into the shot peening prediction integrated model to obtain the integrated prediction value of the second spring characteristic parameter corresponding to each of the spring shot peening parameters; when the integrated prediction value of the second spring characteristic parameter meets the spring shot peening standard and the corresponding shot peening compliance prediction value is greater than the set shot peening compliance standard value, the spring shot peening parameter is classified into the compliant spring shot peening parameter set.

[0115] In this step of the present embodiment, when the current spring shot peening parameters need to be adjusted, spring shot peening parameters that meet the standards are screened out from the spring shot peening parameter database to construct a spring shot peening parameter data set. This method can ensure that the characteristic parameters of the second spring after the automobile suspension spring is shot peened by each spring shot peening parameter in the compliant spring shot peening parameter set meet the standards.

[0116] Reference Figure 1 In S60, optimal spring shot peening parameters are obtained according to the spring shot peening parameters in the compliant spring shot peening parameter set.

[0117] Furthermore, the steps for obtaining the optimal spring shot peening parameters include:

[0118] According to each of the spring shot peening parameters, the corresponding shot peening compliance prediction value and shot peening energy consumption value are obtained; according to the shot peening compliance prediction value and the shot peening energy consumption value, the spring shot peening comprehensive value corresponding to each of the spring shot peening parameters is obtained; the spring shot peening parameter corresponding to the largest spring shot peening comprehensive value is used as the optimal spring shot peening parameter.

[0119] Furthermore, the calculation formula of the spring shot peening comprehensive value is:

[0120] ;

[0121] in, It is represented by the spring shot peening comprehensive value corresponding to the spring shot peening parameter of the jth item in the compliant spring shot peening parameter set; It is represented by the shot peening compliance prediction value corresponding to the spring shot peening parameters described in item j; It is expressed as the shot peening energy consumption value corresponding to the spring shot peening parameters described in the jth item; and They are expressed as shot peening compliance coefficient and shot peening energy consumption coefficient respectively; and The default value is 0.5, and it can be modified according to the actual application.

[0122] In this step of the present embodiment, when the current spring shot peening parameters are unqualified, the spring shot peening parameters in the spring shot peening parameter database are screened to obtain a compliant spring shot peening parameter set; the optimal spring shot peening parameters are obtained by combining the shot peening compliance prediction value and shot peening energy consumption value corresponding to each spring shot peening parameter in the compliant spring shot peening parameter set. This method can combine the influence of spring shot peening parameters on the performance of automobile suspension springs after shot peening and the shot peening energy consumption value corresponding to the spring shot peening parameters, thereby achieving the goal of reducing the energy consumption value while ensuring the performance of automobile suspension springs, and improving the shot peening efficiency of automobile suspension springs.

[0123] Furthermore, the calculation formula of the shot peening energy consumption value is:

[0124] ;

[0125] in, It is represented by the shot peening time in the spring shot peening parameter of the jth item in the compliant spring shot peening parameter set; It is expressed as the shot peening gas pressure in the spring shot peening parameters described in Article j; It is expressed as the shot peening diameter in the spring shot peening parameters described in Article j; , , and They are respectively expressed as shot peening time parameter, shot peening air pressure parameter, shot peening diameter parameter and comprehensive energy consumption coefficient; , , and The default value is 0.25, and it can be modified according to the actual application.

[0126] The shot peening time parameter, shot peening air pressure parameter and shot peening diameter parameter in the shot peening process will affect the energy consumption in the shot peening process. In this embodiment, the step obtains the shot peening energy consumption value based on the shot peening time parameter, shot peening air pressure parameter and shot peening diameter parameter in the shot peening process. This method can provide a data basis for the selection of the optimal spring shot peening parameters.

[0127] Reference Figure 1In S70, the spring is shot peened according to the optimal spring shot peening parameters.

[0128] Furthermore, after the shot peening is completed, the automobile suspension spring can also be tested to obtain the actual shot peening compliance; when the difference between the actual shot peening compliance and the shot peening compliance prediction value is less than the set difference threshold, the actual shot peening compliance, the corresponding spring characteristic parameters and the spring shot peening parameters are saved in the historical shot peening data set, and the shot peening detection integrated model is retrained; otherwise, a shot peening abnormality warning is output; in this embodiment, the difference between the actual shot peening compliance and the shot peening compliance prediction value of the two automobile suspension springs produced by Factory A after shot peening is less than the set difference threshold, therefore, the actual shot peening compliance, the corresponding spring characteristic parameters and the spring shot peening parameters are saved in the historical shot peening data set, and the shot peening detection integrated model is retrained.

[0129] Embodiment 2

[0130] Reference Figure 1 In S10, the present invention collects first spring characteristic parameters and current spring shot peening parameters of two automobile suspension springs produced by factory B;

[0131] Furthermore, the first spring characteristic parameters of the automobile suspension spring include the hardness, material diameter, yield strength and elastic modulus of the automobile suspension spring; the material diameter is the average diameter of the steel wire in the automobile suspension spring. In this embodiment, the first spring characteristic parameters of two automobile suspension springs are obtained, and specific reference is made to Table 4.

[0132] Table 4 Table of characteristic parameters of the first spring of the automobile suspension spring of Example 2

[0133]

[0134] Furthermore, the spring shot peening parameters include shot peening hardness, shot peening diameter, shot peening distance, shot peening air pressure, shot peening time and shot peening incident angle. This embodiment obtains the current spring shot peening parameters corresponding to two automobile suspension springs, and specifically refers to Table 5.

[0135] Table 5 Current spring shot peening parameters corresponding to the automobile suspension spring of Example 2

[0136]

[0137] Reference Figure 1 In S20, combining the first spring characteristic parameter and the current spring shot peening parameter, an integrated prediction value of a second spring characteristic parameter of the spring after shot peening is obtained through a shot peening prediction integrated model.

[0138] Furthermore, the second spring characteristic parameters include surface residual stress, residual compressive stress layer thickness, surface roughness and shot peening coverage of the automobile suspension spring after shot peening.

[0139] Further, the step of predicting the second spring characteristic parameter of the spring after shot peening comprises:

[0140] Input the spring characteristic parameter and the current spring shot peening parameter into the trained shot peening prediction integrated model to obtain the second spring characteristic parameter integrated prediction value of the spring;

[0141] The shot peening prediction integrated model includes a parameter input unit, a prediction unit and an integrated result output unit;

[0142] The parameter input unit is used to input the spring characteristic parameters and the current spring shot peening parameters into the shot peening detection integrated model; the prediction unit includes 4*N trained sub-prediction models; the 4*N trained sub-prediction models include N sub-prediction models for surface residual stress prediction, N sub-prediction models for residual compressive stress layer thickness prediction, N sub-prediction models for surface roughness prediction and N sub-prediction models for shot peening coverage prediction; according to the 4*N trained sub-prediction models, N surface residual stress prediction values, N residual compressive stress layer thickness prediction values, N surface roughness prediction values ​​and N shot peening coverage prediction values ​​are obtained; the prediction values ​​output by the 4*N sub-prediction models are input into the integrated result output unit to obtain the surface residual stress integrated prediction value, residual compressive stress layer thickness integrated prediction value, surface roughness integrated prediction value and shot peening coverage integrated prediction value of the spring after shot peening. The N is 3 or 4, etc.; the sub-prediction models include BP neural network model, Transformer model and support vector machine model, etc.

[0143] Furthermore, the integrated result output unit includes:

[0144] The integrated result output unit includes a surface residual stress integration module, a residual compressive stress layer thickness integration module, a surface roughness integration module and a shot peening coverage integration module; the surface residual stress integration module is used to integrate N surface residual stress prediction values ​​to obtain a surface residual stress integrated prediction value; the residual compressive stress layer thickness integration module is used to integrate N residual compressive stress layer thickness prediction values ​​to obtain a residual compressive stress layer thickness integrated prediction value; the surface roughness integration module is used to integrate N surface roughness prediction values ​​to obtain a surface roughness integrated prediction value; the shot peening coverage integration module is used to integrate N shot peening coverage prediction values ​​to obtain a shot peening coverage integrated prediction value;

[0145] The calculation formula of the integrated prediction value of the surface residual stress is: ;when , expressed as surface residual compressive stress; when , expressed as surface residual tensile stress;

[0146] in, It is expressed as the integrated predicted value of surface residual stress; It represents the predicted value of surface residual stress output by the i-th sub-prediction model; It is represented as the weight coefficient of the surface residual stress prediction value output by the i-th sub-prediction model;

[0147] The calculation formula of the weight coefficient is:

[0148] ;

[0149] in, It is represented as the average value of the surface residual stress prediction accuracy of the i-th sub-prediction model on the test set; It is represented as the variance of the prediction accuracy of the surface residual stress of the ith sub-prediction model on the test set; It is expressed as an exponential function with a natural constant as base; and They are respectively expressed as the prediction accuracy mean coefficient and the prediction accuracy variance coefficient; and The default values ​​are all 0.5, and can be modified according to actual applications; the calculation formulas for the integrated prediction value of the residual compressive stress layer thickness, the integrated prediction value of the surface roughness, and the integrated prediction value of the shot peening coverage rate can refer to the calculation formula for the integrated prediction value of the surface residual stress.

[0150] Furthermore, the present embodiment obtains first spring characteristic parameters, corresponding spring shot peening parameters, and corresponding second spring characteristic parameters after shot peening of five automobile suspension springs;

[0151] The first spring characteristic parameters and corresponding spring shot peening parameters of ten automobile suspension springs are input into the shot peening prediction integrated model to obtain the integrated prediction value of the second spring characteristic parameter; the integrated prediction value of the second spring characteristic parameter corresponding to five automobile suspension springs is compared with the actual value of the second spring characteristic parameter to obtain the prediction accuracy, and the calculation formula is:

[0152] ;

[0153] in, It is expressed as the prediction accuracy of the vth parameter in the second spring characteristic parameter of the uth automobile suspension spring by the shot peening prediction integrated model; the value range of u is 1 to 10; the value range of v is 1 to 4; v=1 corresponds to the surface residual stress in the second spring characteristic parameter; v=2 corresponds to the thickness of the residual compressive stress layer in the second spring characteristic parameter; v=3 corresponds to the surface roughness in the second spring characteristic parameter; v=4 corresponds to the shot peening coverage in the second spring characteristic parameter; It represents the integrated prediction value of the vth parameter in the second spring characteristic parameter of the uth automobile suspension spring output by the shot peening prediction integrated model; It is represented as the actual value of the vth parameter in the second spring characteristic parameter of the uth automobile suspension spring. The prediction accuracies corresponding to the four parameters in the second spring characteristic parameter are averaged to obtain the average prediction accuracies corresponding to the four parameters; in order to verify the effectiveness of the shot peening prediction integrated model, this embodiment also selected three prediction models to predict the second spring characteristic parameters corresponding to ten automobile suspension springs, and obtained the average prediction accuracies corresponding to the four parameters in the second spring characteristic parameter. For specific results, refer to Table 6.

[0154] Table 6 Example 2 Shot Peening Prediction Integrated Model Prediction Effect Verification Table

[0155]

[0156] In Table 6, the model number 1 is the shot peening prediction integrated model constructed by the present invention; the model number 2 is based on the model number 1, and only the spring shot peening parameters are used as the input of the model; the model number 3 is based on the model number 2, each sub-prediction model is integrated with equal weights; the model number 4 is the BP neural network used to predict the characteristic parameters of the second spring. It can be seen from Table 6 that the shot peening prediction integrated model provided by the present invention can effectively improve the prediction accuracy of the characteristic parameters of the second spring.

[0157] Reference Figure 1 In S30, it is determined whether the current spring shot peening parameters need to be adjusted according to the second spring characteristic parameters.

[0158] Further, when the second spring characteristic parameter integrated prediction value meets the spring shot peening standard, the shot peening compliance prediction value is calculated according to the second spring characteristic parameter integrated prediction value; otherwise, it is determined that the current spring shot peening parameters need to be adjusted; the calculation formula of the shot peening compliance prediction value is:

[0159] ;

[0160] in, It is expressed as the predicted value of shot peening compliance; It is expressed as the integrated predicted value of surface residual stress; Expressed as the set surface residual stress standard value, ; It is expressed as the integrated predicted value of residual compressive stress layer thickness; It is expressed as the integrated predicted value of surface roughness; It is expressed as the integrated prediction value of shot peening coverage; , , and They are expressed as residual stress coefficient, residual compressive stress layer thickness coefficient, roughness coefficient and shot peening coverage coefficient respectively; , , and The default value is 0.25, which can be modified according to the actual application;

[0161] When the shot peening compliance prediction value is greater than the set shot peening compliance standard value, it is determined that the current shot peening parameters do not need to be adjusted; otherwise, it is determined that the current spring shot peening parameters need to be adjusted.

[0162] Furthermore, the spring shot peening standard includes: , , , When both are satisfied, it indicates that the second spring characteristic parameters meet the spring shot peening standard; otherwise, it indicates that the second spring characteristic parameters do not meet the spring shot peening standard; wherein, , , and They are respectively expressed as the integrated prediction value of surface residual stress, the integrated prediction value of residual compressive stress layer thickness, the integrated prediction value of surface roughness and the integrated prediction value of shot peening coverage; , , and They are respectively represented by the set standard value of surface residual stress, standard value of residual compressive stress layer thickness, standard value of surface roughness and standard value of shot peening coverage.

[0163] Reference Figure 1 In S40, when it is determined that the spring shot peening parameters need to be adjusted, the spring shot peening parameters in the spring shot peening parameter database are screened to obtain a compliant spring shot peening parameter set; otherwise, the current spring shot peening parameters are used as the optimal spring shot peening parameters, and jump to S60.

[0164] Furthermore, the screening of spring shot peening parameters to obtain a compliant spring shot peening parameter set includes:

[0165] All the spring shot peening parameters in the spring shot peening parameter database are input into the shot peening prediction integrated model to obtain the integrated prediction value of the second spring characteristic parameter corresponding to each of the spring shot peening parameters; when the integrated prediction value of the second spring characteristic parameter meets the spring shot peening standard and the corresponding shot peening compliance prediction value is greater than the set shot peening compliance standard value, the spring shot peening parameter is classified into the compliant spring shot peening parameter set.

[0166] Reference Figure 1 In S50, optimal spring shot peening parameters are obtained according to the spring shot peening parameters in the compliant spring shot peening parameter set.

[0167] Furthermore, the steps for obtaining the optimal spring shot peening parameters include:

[0168] According to each of the spring shot peening parameters, the corresponding shot peening compliance prediction value and shot peening energy consumption value are obtained; according to the shot peening compliance prediction value and the shot peening energy consumption value, the spring shot peening comprehensive value corresponding to each of the spring shot peening parameters is obtained; the spring shot peening parameter corresponding to the largest spring shot peening comprehensive value is used as the optimal spring shot peening parameter.

[0169] Furthermore, the calculation formula of the spring shot peening comprehensive value is:

[0170] ;

[0171] in, It is represented by the spring shot peening comprehensive value corresponding to the spring shot peening parameter of the jth item in the compliant spring shot peening parameter set; It is represented by the shot peening compliance prediction value corresponding to the spring shot peening parameters described in item j; It is expressed as the shot peening energy consumption value corresponding to the spring shot peening parameters described in the jth item; and They are expressed as shot peening compliance coefficient and shot peening energy consumption coefficient respectively; and The default value is 0.5, and it can be modified according to the actual application.

[0172] Furthermore, the calculation formula of the shot peening energy consumption value is:

[0173] ;

[0174] in, It is represented by the shot peening time in the spring shot peening parameter of the jth item in the compliant spring shot peening parameter set; It is expressed as the shot peening gas pressure in the spring shot peening parameters described in Article j; It is expressed as the shot peening diameter in the spring shot peening parameters described in Article j; , , and They are respectively expressed as shot peening time parameter, shot peening air pressure parameter, shot peening diameter parameter and comprehensive energy consumption coefficient; , , and The default value is 0.25, and it can be modified according to the actual application.

[0175] Reference Figure 1 In S60, the spring is shot peened according to the optimal spring shot peening parameters.

[0176] Furthermore, after the shot peening is completed, the automobile suspension spring can also be tested to obtain the actual shot peening compliance; when the difference between the actual shot peening compliance and the shot peening compliance prediction value is less than the set difference threshold, the actual shot peening compliance, the corresponding spring characteristic parameters and the spring shot peening parameters are saved in the historical shot peening data set, and the shot peening detection integrated model is retrained; otherwise, a shot peening abnormality warning is output; in this embodiment, the difference between the actual shot peening compliance and the shot peening compliance prediction value of the two automobile suspension springs produced by Factory A after shot peening is less than the set difference threshold, therefore, the actual shot peening compliance, the corresponding spring characteristic parameters and the spring shot peening parameters are saved in the historical shot peening data set, and the shot peening detection integrated model is retrained.

[0177] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for predicting residual stress in automobile suspension spring production, characterized in that: include: S10, obtaining a first spring characteristic parameter and a current spring shot peening parameter of a vehicle suspension spring; S20, combining the first spring characteristic parameter and the current spring shot peening parameter, obtaining an integrated prediction value of a second spring characteristic parameter of the spring after shot peening through a shot peening prediction integrated model; the second spring characteristic parameter includes the surface residual stress, residual compressive stress layer thickness, surface roughness and shot peening coverage of the automobile suspension spring after shot peening; S30, judging whether the current spring shot peening parameters need to be adjusted according to the second spring characteristic parameter integrated prediction value, and executing S40 when it is determined that the spring shot peening parameters do not need to be adjusted, otherwise, executing S50 to S70; S40, taking the current spring shot peening parameters as the optimal spring shot peening parameters, and jumping to S70; S50, screening the spring shot peening parameters in the spring shot peening parameter database to obtain a compliant spring shot peening parameter set; S60, according to the spring shot peening parameters in the compliant spring shot peening parameter set, the optimal spring shot peening parameters are obtained; the specific steps include: according to each of the spring shot peening parameters, the corresponding shot peening compliance prediction value and shot peening energy consumption value are obtained; according to the shot peening compliance prediction value and the shot peening energy consumption value, the spring shot peening comprehensive value corresponding to each of the spring shot peening parameters is obtained; the spring shot peening parameter corresponding to the largest spring shot peening comprehensive value is used as the optimal spring shot peening parameter; S70, performing shot peening operation on the spring according to the optimal spring shot peening parameters; The first spring characteristic parameters of the automobile suspension spring include hardness, material diameter, yield strength and elastic modulus of the automobile suspension spring; The spring shot peening parameters include shot peening hardness, shot peening diameter, shot peening distance, shot peening air pressure, shot peening time and shot peening incident angle; The S20 includes: inputting the spring characteristic parameter and the current spring shot peening parameter into a trained shot peening prediction integrated model to obtain a second spring characteristic parameter integrated prediction value of the spring; The shot peening prediction integrated model includes a parameter input unit, a prediction unit and an integrated result output unit; The parameter input unit is used to input the spring characteristic parameters and the current spring shot peening parameters into the shot peening detection integrated model; the prediction unit includes 4*N trained sub-prediction models; the 4*N trained sub-prediction models obtain N surface residual stress prediction values, N residual compressive stress layer thickness prediction values, N surface roughness prediction values ​​and N shot peening coverage prediction values; the prediction values ​​output by the 4*N sub-prediction models are input into the integrated result output unit to obtain the integrated prediction value of the surface residual stress, the integrated prediction value of the residual compressive stress layer thickness, the integrated prediction value of the surface roughness and the integrated prediction value of the shot peening coverage of the spring after shot peening.

2. The method for predicting residual stress in automobile suspension spring production according to claim 1, characterized in that: The integrated result output unit comprises: The integrated result output unit includes a surface residual stress integration module, a residual compressive stress layer thickness integration module, a surface roughness integration module and a shot peening coverage integration module; the surface residual stress integration module is used to integrate N surface residual stress prediction values ​​to obtain a surface residual stress integrated prediction value; the residual compressive stress layer thickness integration module is used to integrate N residual compressive stress layer thickness prediction values ​​to obtain a residual compressive stress layer thickness integrated prediction value; the surface roughness integration module is used to integrate N surface roughness prediction values ​​to obtain a surface roughness integrated prediction value; the shot peening coverage integration module is used to integrate N shot peening coverage prediction values ​​to obtain a shot peening coverage integrated prediction value; The calculation formula of the integrated prediction value of the surface residual stress is: It is expressed as surface residual compressive stress; Expressed as surface residual tensile stress; in, It is expressed as the integrated predicted value of surface residual stress; It represents the predicted value of surface residual stress output by the i-th sub-prediction model; It is represented as the weight coefficient of the surface residual stress prediction value output by the i-th sub-prediction model; The calculation formula of the weight coefficient is: in, It is represented as the average value of the surface residual stress prediction accuracy of the i-th sub-prediction model on the test set; It is represented as the variance of the prediction accuracy of the surface residual stress of the ith sub-prediction model on the test set; and They are respectively expressed as the prediction accuracy mean coefficient and the prediction accuracy variance coefficient; It is expressed as an exponential function with a natural constant as base; The calculation formulas for the integrated prediction value of the residual compressive stress layer thickness, the integrated prediction value of the surface roughness and the integrated prediction value of the shot peening coverage rate may refer to the calculation formula for the integrated prediction value of the surface residual stress.

3. The method for predicting residual stress in automobile suspension spring production according to claim 1, characterized in that: The step S30 includes: when the second spring characteristic parameter integrated prediction value meets the spring shot peening standard, calculating the shot peening compliance prediction value according to the second spring characteristic parameter integrated prediction value; otherwise, determining that the current spring shot peening parameters need to be adjusted; the calculation formula of the shot peening compliance prediction value is: in, It is expressed as the predicted value of shot peening compliance; It is expressed as the integrated predicted value of surface residual stress; Expressed as the set surface residual stress standard value, It is expressed as the integrated predicted value of residual compressive stress layer thickness; It is expressed as the integrated predicted value of surface roughness; It is expressed as the integrated prediction value of shot peening coverage; They are expressed as residual stress coefficient, residual compressive stress layer thickness coefficient, roughness coefficient and shot peening coverage coefficient respectively; When the shot peening compliance prediction value is greater than the set shot peening compliance standard value, it is determined that the current shot peening parameters do not need to be adjusted; otherwise, it is determined that the current spring shot peening parameters need to be adjusted.

4. The method for predicting residual stress in automobile suspension spring production according to claim 3, characterized in that: The spring shot peening standards include: When both are satisfied, it indicates that the second spring characteristic parameters meet the spring shot peening standard; otherwise, it indicates that the second spring characteristic parameters do not meet the spring shot peening standard; wherein, They are respectively expressed as the integrated prediction value of surface residual stress, the integrated prediction value of residual compressive stress layer thickness, the integrated prediction value of surface roughness and the integrated prediction value of shot peening coverage; They are respectively represented by the set standard value of surface residual stress, standard value of residual compressive stress layer thickness, standard value of surface roughness and standard value of shot peening coverage.

5. The method for predicting residual stress in automobile suspension spring production according to claim 1, characterized in that: The screening of spring shot peening parameters to obtain a set of compliant spring shot peening parameters includes: All the spring shot peening parameters in the spring shot peening parameter database are input into the shot peening prediction integrated model to obtain the integrated prediction value of the second spring characteristic parameter corresponding to each of the spring shot peening parameters; when the integrated prediction value of the second spring characteristic parameter meets the spring shot peening standard and the corresponding shot peening compliance prediction value is greater than the set shot peening compliance standard value, the spring shot peening parameter is classified into the compliant spring shot peening parameter set.

6. The method for predicting residual stress in automobile suspension spring production according to claim 1, characterized in that: The calculation formula for the comprehensive value of spring shot peening is: in, It is represented by the spring shot peening comprehensive value corresponding to the spring shot peening parameter of the jth item in the compliant spring shot peening parameter set; It is represented by the shot peening compliance prediction value corresponding to the spring shot peening parameters described in item j; It is expressed as the shot peening energy consumption value corresponding to the spring shot peening parameters described in the jth item; and They are respectively expressed as shot peening compliance coefficient and shot peening energy consumption coefficient.

7. The method for predicting residual stress in automobile suspension spring production according to claim 6, characterized in that: The calculation formula of the shot peening energy consumption value is: in, It is represented by the shot peening time in the spring shot peening parameter of the jth item in the compliant spring shot peening parameter set; Expressed as the shot peening gas pressure in the spring shot peening parameters described in Article j; It is expressed as the shot peening diameter in the spring shot peening parameters described in Article j; They are respectively expressed as shot peening time parameter, shot peening air pressure parameter, shot peening diameter parameter and comprehensive energy consumption coefficient.

Images