Laser shock remanufacturing method for aero-engine blades based on digital twin

Abstract

The present invention discloses a laser shock remanufacturing method for aircraft engine blades based on digital twins, which belongs to the field of deep transfer learning and includes: collecting multimodal data such as cutting force, vibration, and temperature to build a monitoring system for the uneven stress distribution caused by tool wear. The maximum mean difference method is used to align the features of data under different working conditions, and a wear diagnosis and stress prediction model is constructed in combination with deep transfer learning to identify stress concentration areas. Laser shock strengthening remanufacturing is implemented in stress concentration areas, shock parameters are optimized, and adversarial generative networks are combined to expand small sample data sets to improve remanufacturing adaptability. Incremental transfer learning is used to dynamically adjust the model based on the freeze-fine-tuning mechanism to achieve rapid optimization under new working conditions. By first monitoring tool wear and stress distribution, and then implementing laser shock strengthening, the remanufacturing strategy is optimized in combination with incremental transfer learning to improve the residual stress control accuracy and processing stability of the blades.

Description

Technical Field

[0001] The present invention belongs to the field of deep transfer learning, and in particular relates to a laser shock remanufacturing method for aircraft engine blades based on digital twins. Background Art

[0002] The manufacturing and remanufacturing process of aircraft engine blades involves high-precision CNC machining. The surface quality and residual stress state of the blades directly affect their service performance and service life. However, during the CNC machining process, the cumulative effect of tool wear can lead to fluctuations in cutting forces, changes in heat input, and uneven material deformation, which can cause an imbalance in the residual stress distribution on the blade surface and even cause local stress concentration problems. This stress unevenness may reduce the fatigue resistance of the blade and affect the effectiveness of subsequent laser shock peening and remanufacturing. Therefore, how to accurately monitor the tool wear state during the machining process, predict stress concentration areas, and optimize laser shock remanufacturing parameters are key to improving blade machining quality and service stability.

[0003] Traditional blade machining quality control relies primarily on empirical parameter adjustments, making it difficult to accurately predict the impact of tool wear on stress distribution under complex operating conditions, leading to limitations in stress optimization strategies. In recent years, with the development of intelligent manufacturing and data-driven technologies, the use of multiple sensors to monitor machining conditions in real time, combined with deep learning and transfer learning methods to model tool wear and stress evolution, has become an important approach to improving CNC machining accuracy. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides a laser shock remanufacturing method for aircraft engine blades based on digital twins, comprising:

[0005] Collecting multimodal data during the numerical control machining of an aero-engine blade, and preprocessing the multimodal data to obtain a preprocessed data set;

[0006] Constructing a tool wear diagnosis and stress distribution prediction model, inputting the preprocessed data set into the tool wear diagnosis and stress distribution prediction model, and obtaining tool wear diagnosis and stress distribution prediction results;

[0007] Based on the tool wear diagnosis and stress distribution prediction results, the stress concentration area caused by tool wear is identified and the laser shock peening strategy is optimized;

[0008] Based on the optimized laser shock peening strategy, laser shock peening is performed on the stress concentration area to obtain feedback data;

[0009] Based on the feedback data, the tool wear diagnosis and stress distribution prediction model is optimized by an incremental transfer learning method to obtain a multi-working condition adaptation model;

[0010] The remanufacturing of different blades is predicted based on the multi-operating condition adaptation model.

[0011] Preferably, the process of obtaining the preprocessed data set includes: collecting multimodal data during the CNC machining of aircraft engine blades, and reducing the dimension of the multimodal data based on sparse principal component analysis and variational autoencoder to obtain the preprocessed data set.

[0012] Preferably, the expression of the optimization objective function of the sparse principal component analysis is:

[0013] ;

[0014] in, is the projection matrix, is the sparse regularization parameter, is the original data matrix;

[0015] The calculation expression of the variational autoencoder is:

[0016] ;

[0017] in, is the variational posterior distribution, is a hidden variable, is the prior distribution, is the Kullback-Leibler divergence.

[0018] Preferably, the process of obtaining tool wear diagnosis and stress distribution prediction results includes:

[0019] Performing distribution alignment on the data features of different wear states in the preprocessed data set based on the maximum mean difference method to obtain an aligned data set;

[0020] A tool wear diagnosis and stress distribution prediction model is constructed based on a self-attention time series network, and the time series data set is input into the tool wear diagnosis and stress distribution prediction model for calculation to obtain the tool wear diagnosis and stress distribution prediction results.

[0021] Preferably, the expression of the maximum mean difference method is:

[0022] ;

[0023] in, is the feature map mapped to the reproducing kernel Hilbert space, is the reproducing kernel Hilbert space, is the source domain data distribution, is the target domain data distribution, is the source domain sample, is the target domain sample.

[0024] Preferably, based on the tool wear diagnosis and stress distribution prediction results, the process of identifying stress concentration areas caused by tool wear and optimizing the laser shock peening strategy includes:

[0025] constructing a stress concentration area data set based on the tool wear diagnosis and stress distribution prediction results;

[0026] Calculating the correlation of data in the stress concentration area data set, and constructing a stress concentration area identification model based on the correlation;

[0027] Calculating the changing trends of various physical variables during the machining process based on the stress concentration area identification model, establishing a stress evolution model, and screening areas with significant stress anomalies based on the stress evolution model;

[0028] Acquire a laser shock peening region based on the region with significant stress anomaly, and optimize laser shock parameters based on the distribution of the laser shock peening region;

[0029] The laser shock peening strategy is optimized based on the optimized laser shock parameters.

[0030] Preferably, the process of optimizing the laser shock parameters based on the distribution of the laser shock peening area includes:

[0031] Set different combinations of laser shock parameters, conduct experimental verification, and establish a mapping relationship between the effects of parameters on surface residual stress;

[0032] Optimize parameters based on the mapping relationship, perform laser shock, record the plasma pressure changes and shock wave propagation characteristics in the shock area, and analyze the shock wave attenuation law under different laser parameters;

[0033] The finite element simulation method is used to verify the distribution of residual stress, and corrections are made based on experimental measurement data to obtain optimized laser shock peening process parameters.

[0034] Preferably, based on the feedback data, the tool wear diagnosis and stress distribution prediction model is optimized by an incremental transfer learning method, and the process of obtaining a multi-working condition adaptation model also includes: in the process of updating the model parameters, a freeze-fine-tuning strategy is adopted to freeze some layers of the deep neural network, update the parameters of specific layers, and for the low-level feature extraction layer, keep the parameters unchanged so that the low-level feature extraction layer continues to extract general features; for the high-level decision layer, parameter fine-tuning is allowed to make the model adapt to the new working condition data.

[0035] Preferably, in the incremental transfer learning method, the expression of the optimization objective is:

[0036] ;

[0037] in, is the source condition model loss, The deviation of the distribution calculated as the maximum mean difference, and is the weight coefficient, Train the model parameters for the source condition, Train the model parameters for the target operating conditions.

[0038] Compared with the prior art, the present invention has the following advantages and technical effects:

[0039] 1. This invention, based on the impact of tool wear on stress distribution during CNC machining of aero-engine blades, collects multimodal data such as cutting force, vibration, temperature field, and stress distribution. It then uses sparse principal component analysis (SPCA) and variational autoencoders (VAE) for dimensionality reduction and feature extraction to construct a tool wear diagnosis model. By calculating changes in tool wear rate and stress concentration areas, it accurately identifies stress anomalies during machining. The maximum mean difference (MMD) method is then used to optimize the feature distribution under different operating conditions, improving the model's adaptability to complex operating conditions and enabling efficient and accurate stress anomaly detection.

[0040] 2. This invention uses incremental transfer learning to build a deep learning model, enabling cross-condition knowledge transfer for different blade models, tool wear states, and impact parameters. Using the MMD method to align the source and target domain data, combined with a freeze-and-fine-tune strategy, this model improves its learning capabilities under new operating conditions while maintaining the learning results from the original machining conditions. This improves prediction accuracy and computational efficiency, ensuring the generalization of intelligent monitoring across multiple operating conditions.

[0041] 3. This invention addresses stress concentration caused by tool wear by using a data-driven approach to precisely identify strengthening areas and optimize laser shock parameters to ensure uniform residual stress distribution. Combining plasma pressure monitoring with shock wave propagation analysis, the laser shock power, pulse width, and number of shocks are dynamically adjusted to improve strengthening uniformity and stability. The strengthening effect is quality-assessed through fatigue life testing, microstructure analysis, and surface roughness measurement. A feedback optimization model is then used to further enhance the reliability and service stability of blade remanufacturing. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0043] Figure 1Schematic diagram of a method flow in an embodiment of the present invention. DETAILED DESCRIPTION

[0044] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0045] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0046] Example 1

[0047] like Figure 1 As shown, this embodiment provides a laser shock remanufacturing method for aircraft engine blades based on digital twins, including:

[0048] S1. Collect multimodal data during the CNC machining of aero-engine blades and set the data set ,in The cutting force data, including normal force, tangential force and feed force, are monitored in real time using a force sensor; For vibration signal data, the tool-workpiece dynamic contact characteristics are recorded by a three-axis accelerometer to characterize machining stability; To obtain temperature field data, an infrared thermal imager is used to record the temperature distribution on the blade surface and the characteristics of the heat-affected zone (HAZ); The stress distribution data is obtained by combining strain gauges and digital speckle pattern interferometry (DSPI) to monitor the evolution trend of residual stress on the blade surface during machining.

[0049] Due to the high dimension of the data, the existence of redundant information and the uneven distribution of different modal data, sparse principal component analysis (SPCA) and variational autoencoder (VAE) are used to reduce the dimension, extract the key features in the processing process, and improve data utilization and computational efficiency. Assume that the original data matrix is ,in is the number of samples, For data dimensions, SPCA introduces Regularization constraints, the optimization objective function is as follows:

[0050] ;

[0051] is the projection matrix, is the sparse regularization parameter, which ensures that the data after dimensionality reduction still retains the most discriminative feature information.

[0052] For nonlinear feature relationships, VAE is further used for feature extraction. Assume that the input data Mapping to latent variables through encoder , obeys a certain prior distribution , which is then reconstructed by the decoder into , minimize the evidence lower bound (ELBO):

[0053] ;

[0054] in, is the variational posterior distribution, By jointly optimizing SPCA and VAE, the dimensionality reduction features of different modal data retain global distribution information while also possessing local discriminative capabilities, providing data support for subsequent tool wear diagnosis and stress distribution modeling.

[0055] S2. Based on the multimodal data collected during the CNC machining process, a tool wear diagnosis and stress distribution prediction model is constructed to identify stress concentration areas caused by tool wear. Set the tool wear status data set ,in is the cutting force data at different wear stages, is the corresponding vibration signal data, is the temperature field distribution on the tool surface during the wear process, The residual stress measurement data is obtained by analyzing the influence of tool wear on different physical fields and constructing a mapping model between tool wear state and stress distribution.

[0056] Since the data distribution of different tool wear stages has inter-domain deviation, the maximum mean difference (MMD) is used to align the distribution of data features of different wear states. Assume that the source domain data distribution is , the target domain data distribution is , the maximum mean difference is calculated as follows:

[0057] ;

[0058] in, is the feature map mapped to the reproducing kernel Hilbert space, is the reproducing kernel Hilbert space, is the source domain data distribution, is the target domain data distribution, is the source domain sample, The target domain samples are optimized by minimizing MMD to improve the cross-condition prediction capability.

[0059] In terms of stress distribution prediction, the Self-Attention Temporal Network (SATN) is used to model the dynamic evolution of stress concentration areas during wear. The time series data of tool wear is set as ,in is the temperature change, is the vibration signal, is the cutting force, is the stress distribution. The SATN model uses the self-attention mechanism to calculate the time step The weighted feature representation of is:

[0060] ;

[0061] in, are query, key and value matrices respectively, is the scaling factor of the feature dimension. By introducing the self-attention mechanism, the model can capture the long-term impact of tool wear on stress concentration areas and improve the prediction accuracy of stress distribution based on wear state.

[0062] Combining MMD transfer learning with SATN time series modeling, cross-working condition tool wear diagnosis and stress distribution prediction are realized, providing a basis for identifying high-precision stress concentration areas in laser shock peening.

[0063] S3. Based on the tool wear diagnosis and stress distribution prediction results, identify the stress concentration area caused by tool wear and optimize the laser shock peening strategy. Set the stress concentration area data set ,in is the residual stress distribution data, is the temperature field data, is the abnormal point of cutting force, The data is driven by analyzing the high stress areas on the blade surface under different tool wear conditions and screening the areas suitable for laser shock peening.

[0064] By calculating the correlation between tool wear characteristics, cutting force characteristics, temperature field characteristics and stress distribution, a stress concentration area identification model is constructed. Assume that the formation of stress concentration area has a mapping relationship with these characteristic variables. ,in represents the distribution of stress concentration areas, Represents each processing feature variable. Normalization is used to keep the features of different data modes within the same numerical range, thereby improving the stability of model prediction.

[0065] For the identified stress concentration areas, the changing trends of various physical variables during the processing are calculated, a stress evolution model is established, and areas with significant stress anomalies are screened and marked as laser shock strengthening areas. Assume that the change amplitude of the stress anomaly area is ,when When the set threshold is exceeded, it is considered that stress concentration exists in the area and needs to be optimized through laser shock peening.

[0066] Combined with experimental data, the identified laser shock peening areas were verified to ensure that the selected areas conformed to the stress evolution laws of the machining process. Finally, a distribution map of the blade stress concentration areas was generated, providing a basis for subsequent laser shock peening parameter optimization.

[0067] S4. Laser shock peening is performed on stress concentration areas to improve the uniformity of residual stress distribution on the blade surface and enhance fatigue resistance. Set up laser shock remanufacturing data set ,in is the residual stress data on the blade surface after impact, is the plasma pressure distribution during the impact process, is the impact force signal, is the dynamic vibration characteristic of the impact area.

[0068] Optimize laser shock parameters, including laser power, according to the distribution of stress concentration areas , spot diameter , pulse width and number of impacts Different parameter combinations are set, experimental verification is conducted, and the mapping relationship between the influence of parameters on surface residual stress is established. Laser shock parameters There is a nonlinear relationship:

[0069] ;

[0070] By fitting this relationship with experimental data, the laser shock parameters are optimized to make the residual stress distribution on the blade surface more uniform and avoid new stress concentration phenomena.

[0071] During the laser shock process, an online monitoring system is used to record the plasma pressure changes and shock wave propagation characteristics in the shock area, and analyze the shock wave attenuation law under different laser parameters. Affects the microstructural changes of materials, when Below the critical value When the laser power is too low, the strengthening effect may be insufficient, and it is necessary to increase the laser power or adjust the pulse width to improve the impact energy transfer efficiency.

[0072] Finite element simulation was used to verify the distribution of residual stress under different impact conditions, and corrections were made based on experimental measurement data. Ultimately, the optimized laser shock peening process parameters were obtained to ensure uniform distribution of residual stress on the blade surface and improve remanufacturing quality.

[0073] S5. Based on the data feedback from the laser shock remanufacturing process, the incremental transfer learning strategy is used to optimize the model to improve its adaptability under different blade models, damage states and shock parameters. Set the model training data set ,in is the residual stress data on the blade surface after impact, is the plasma pressure distribution, is the impact force signal, The dynamic vibration characteristics of the impact area. Due to data distribution deviations between different blade remanufacturing conditions, a directly trained model may perform poorly under new conditions. Therefore, incremental transfer learning is used for optimization to ensure that the model maintains good generalization capabilities even after adding new data.

[0074] Use the maximum mean difference (MMD) to align the new working condition data to ensure the distribution of the source working condition data Distribution of target working condition data In the incremental learning process, the source condition training model parameters are defined and target working condition training model parameters , the optimization objectives are as follows:

[0075] ;

[0076] in, is the source condition model loss, The deviation of the distribution calculated as the maximum mean difference, and is the weight coefficient.

[0077] During the model parameter update process, a freeze-and-fine-tune strategy is employed. This freezes some layers of the deep neural network and updates only the parameters of specific layers. This prevents new data from disrupting the original model and ensures stability during incremental learning. For low-level feature extraction layers, their parameters remain unchanged, allowing them to continue extracting common features. For high-level decision-making layers, parameters are fine-tuned to adapt the model to new operating data and improve prediction accuracy.

[0078] After incremental transfer learning optimization, the model can maintain high-precision prediction capabilities under different laser shock remanufacturing conditions, adapt to the remanufacturing needs of different blades, and ensure processing stability and residual stress control effects.

[0079] S6. After laser shock remanufacturing is completed, the blade surface residual stress distribution, fatigue strength and microstructure changes are evaluated, and the shock process parameters are optimized based on data feedback. Set the blade quality evaluation data set ,in is the residual stress data of the blade after remanufacturing, is the fatigue life measurement result, Information about changes in the material microstructure, Surface roughness characteristics.

[0080] Analyze the residual stress changes before and after laser shock strengthening for different blade areas and calculate the stress improvement rate To evaluate the impact strengthening effect.

[0081] ;

[0082] when Exceeding the set optimization threshold If the strengthening effect is not achieved, the laser shock parameters need to be adjusted. , re-optimize the impact strategy.

[0083] High cycle fatigue (HCF) testing was used to measure the fatigue life of the blades and record the fatigue crack growth rate. The crack initiation is analyzed. Assuming that the crack propagation follows the Paris formula, the fatigue life is estimated as follows:

[0084] ;

[0085] in, and is the material constant, is the stress intensity factor range. If the fatigue life is lower than the set threshold, it indicates that the laser shock peening is insufficient or the parameters are unreasonable, and the shock power or shock coverage needs to be adjusted.

[0086] Measure surface quality and calculate surface roughness using white light interferometry , ensuring that the surface roughness after laser shock treatment meets aviation blade remanufacturing standards. Based on data feedback from different processing conditions, an adaptive optimization model for process parameters was constructed to dynamically adjust the laser shock strategy to ensure consistent remanufacturing quality for different blade models and damage conditions. Ultimately, a blade remanufacturing quality assessment system was established to improve blade reliability and service life.

[0087] Furthermore, the multimodal data collection in step S1 specifically includes the following:

[0088] During the CNC machining of aircraft engine blades, multimodal data such as cutting force, vibration, temperature field, and stress distribution are collected to provide an optimization basis for tool wear diagnosis (S2), stress concentration area identification (S3), and laser shock peening (S4). Due to the different distribution characteristics of different data modalities, directly using the raw data may affect the stability of the model. Therefore, after data collection, sparse principal component analysis (SPCA) is used to reduce the data dimension and extract key features from the high-dimensional data. At the same time, variational autoencoders (VAEs) are used to further enhance the expression of nonlinear features, reduce data redundancy, and improve data quality. The data after dimensionality reduction is represented as:

[0089] ;

[0090] in, is the original data matrix, The dimensionality reduction results are input into the tool wear diagnosis model (S2) to ensure the stability and generalization ability of subsequent predictions.

[0091] Furthermore, the tool wear diagnosis and stress distribution prediction in step S2 specifically include the following:

[0092] Based on the data collected by S1, a tool wear diagnostic model is constructed to predict stress concentration areas caused by wear. Tool wear is a dynamic process. As processing time passes, the degree of wear gradually increases, affecting the stress distribution. The tool wear rate is calculated as follows:

[0093] ;

[0094] in, is the current tool wear amount, is the time interval. This rate is used to determine the current state of the tool and, combined with stress data, predict the changing trend of high-stress areas. Due to differences in data distribution under different machining conditions, maximum mean difference (MMD) is used for feature alignment in S2 to adapt the model to different blade machining conditions. MMD is calculated as follows:

[0095] ;

[0096] in, is the feature map mapped to the reproducing kernel Hilbert space, is the reproducing kernel Hilbert space, is the source domain data distribution, is the target domain data distribution, is the source domain sample, The target domain samples are optimized by minimizing MMD to improve the cross-condition prediction capability.

[0097] Furthermore, the stress concentration area identification in step S3 specifically includes the following:

[0098] Based on the tool wear state and stress distribution predicted by S2, high stress concentration areas are accurately identified to provide optimization strategies for subsequent laser shock peening (S4). ) is calculated as follows:

[0099] ;

[0100] in, and Represent the residual stress values ​​at the current moment and the previous moment respectively. If the stress exceeds a set threshold, the area is marked as a high stress concentration area and included in the laser shock peening optimization range. In addition, based on data-driven classification and analysis of the changing trends of cutting forces, temperature fields, and vibration signals during the processing process, the S4 laser shock peening strategy can accurately act on the areas most in need of optimization, rather than blindly applying laser shock.

[0101] Furthermore, the laser shock peening in step S4 specifically includes the following:

[0102] Based on the high stress area identified by S3, the specific location of laser shock peening is determined and the shock parameters are optimized. The key parameters of laser shock peening include laser power ( ), spot diameter ( ), pulse width ( ) and number of impacts ( ), residual stress The relationship with these parameters is as follows:

[0103] ;

[0104] By optimizing this relationship through experimental data and finite element simulation, the stress distribution on the blade surface after strengthening is made more uniform, and the fatigue resistance is improved. During the laser shock process, the plasma pressure and shock wave propagation characteristics are monitored in real time to ensure the stability of laser shock strengthening. If the shock wave energy decays rapidly, it may be necessary to adjust the shock power or coverage to improve the strengthening effect. In addition, a closed-loop optimization control system is established to monitor the stress distribution on the blade surface after strengthening and compare it with the predicted results of S3. If the strengthening effect does not meet the optimization requirements, the laser shock parameters are automatically adjusted and secondary optimization is performed to achieve the optimal residual stress improvement rate.

[0105] Furthermore, the incremental transfer learning optimization in step S5 specifically includes the following:

[0106] Under different blade models, damage states, and impact parameter changes, an incremental transfer learning strategy is used to continuously optimize the model and improve prediction accuracy. The optimization objectives are as follows:

[0107] ;

[0108] in, Calculate the difference in characteristic distribution between the source and target conditions, Controlling the adjustment range of model parameters ensures that the model remains stable while adapting to new data. To prevent the model from forgetting previously learned knowledge, a freeze-and-fine-tune strategy is adopted during training. This freezes the low-level feature extraction layers and only updates the parameters of the high-level decision-making layers to improve adaptability to new working conditions.

[0109] Furthermore, the blade quality assessment and feedback optimization in step S6 specifically includes the following:

[0110] After laser shock remanufacturing is completed, to ensure the strengthening effect, it is necessary to conduct residual stress assessment on the blade surface, fatigue life testing and microstructure analysis, and optimize the shock process parameters based on the measurement data. The stress improvement rate is calculated as follows:

[0111] ;

[0112] when Below optimization threshold When adjusting the laser shock parameters Secondary optimization was performed. Furthermore, high-cycle fatigue (HCF) testing was used to assess the blade's fatigue life, and its reliability was analyzed using crack growth rate. Based on the feedback data, a data-driven optimization model was constructed. This feedback from S6 was used to optimize the incremental transfer learning process in S5, enabling the model to continuously adapt to new processing conditions and improve the quality stability of blade remanufacturing.

[0113] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A laser shock remanufacturing method for aircraft engine blades based on digital twin, characterized in that: include: Collecting multimodal data during the numerical control machining of an aero-engine blade, and preprocessing the multimodal data to obtain a preprocessed data set; Constructing a tool wear diagnosis and stress distribution prediction model, inputting the preprocessed data set into the tool wear diagnosis and stress distribution prediction model, and obtaining tool wear diagnosis and stress distribution prediction results; Based on the tool wear diagnosis and stress distribution prediction results, the stress concentration area caused by tool wear is identified and the laser shock peening strategy is optimized; Based on the optimized laser shock peening strategy, laser shock peening is performed on the stress concentration area to obtain feedback data; Based on the feedback data, the tool wear diagnosis and stress distribution prediction model is optimized by an incremental transfer learning method to obtain a multi-working condition adaptation model; Predicting the remanufacturing of different blades based on the multi-condition adaptation model; The process of obtaining tool wear diagnosis and stress distribution prediction results includes: Performing distribution alignment on the data features of different wear states in the preprocessed data set based on the maximum mean difference method to obtain an aligned data set; Constructing a tool wear diagnosis and stress distribution prediction model based on a self-attention temporal network, inputting the aligned data set into the tool wear diagnosis and stress distribution prediction model for calculation, and obtaining the tool wear diagnosis and stress distribution prediction results; Based on the tool wear diagnosis and stress distribution prediction results, the process of identifying stress concentration areas caused by tool wear and optimizing the laser shock peening strategy includes: constructing a stress concentration area data set based on the tool wear diagnosis and stress distribution prediction results; Calculating the correlation of data in the stress concentration area data set, and constructing a stress concentration area identification model based on the correlation; Calculating the changing trends of various physical variables during the machining process based on the stress concentration area identification model, establishing a stress evolution model, and screening areas with significant stress anomalies based on the stress evolution model; Acquire a laser shock peening region based on the region with significant stress anomaly, and optimize laser shock parameters based on the distribution of the laser shock peening region; The laser shock peening strategy is optimized based on the optimized laser shock parameters.

2. The method according to claim 1, characterized in that The process of obtaining the preprocessed data set includes: collecting multimodal data during the numerical control machining of aircraft engine blades, and reducing the dimension of the multimodal data based on sparse principal component analysis and variational autoencoder to obtain the preprocessed data set.

3. The method according to claim 2, characterized in that The expression of the optimization objective function of the sparse principal component analysis is: ； in, is the projection matrix, is the sparse regularization parameter, is the original data matrix; The calculation expression of the variational autoencoder is: ； in, is the variational posterior distribution, is a hidden variable, is the prior distribution, is the Kullback-Leibler divergence.

4. The method according to claim 1, wherein The expression of the maximum mean difference method is: ； in, is the feature map mapped to the reproducing kernel Hilbert space, is the reproducing kernel Hilbert space, is the source domain data distribution, is the target domain data distribution, is the source domain sample, is the target domain sample.

5. The method according to claim 1, wherein The process of optimizing the laser shock parameters based on the distribution of the laser shock peening area includes: Set different combinations of laser shock parameters, conduct experimental verification, and establish a mapping relationship between the effects of parameters on surface residual stress; Optimize parameters based on the mapping relationship, perform laser shock, record the plasma pressure changes and shock wave propagation characteristics in the shock area, and analyze the shock wave attenuation law under different laser parameters; The finite element simulation method is used to verify the distribution of residual stress, and corrections are made based on experimental measurement data to obtain optimized laser shock peening process parameters.

6. The method according to claim 1, wherein Based on the feedback data, the tool wear diagnosis and stress distribution prediction model is optimized through an incremental transfer learning method to obtain a multi-working condition adaptation model. The process also includes: in the process of updating the model parameters, a freeze-fine-tuning strategy is adopted to freeze some layers of the deep neural network, update the parameters of specific layers, and for the low-level feature extraction layer, keep the parameters unchanged so that the low-level feature extraction layer continues to extract common features; for the high-level decision layer, parameter fine-tuning is allowed to make the model adapt to the new working condition data.

7. The method according to claim 4, characterized in that In the incremental transfer learning method, the optimization objective is expressed as: ； in, is the source condition model loss, The deviation of the distribution calculated as the maximum mean difference, and is the weight coefficient, Train the model parameters for the source condition, Train the model parameters for the target operating conditions.

Images