Laser shock peening monitoring method for digital twin-driven data-enhanced aviation landing gear

Abstract

The invention discloses a digital twin-driven data-enhanced aviation undercarriage laser shock peening monitoring method, and belongs to the field of aviation structural member remanufacturing, intelligent manufacturing and data driving regulation, and the method comprises the steps: building a fatigue evaluation system for an alternating load in an undercarriage service process, recognizing a high stress concentration region, and carrying out the recognition of a high stress concentration region; a multi-modal monitoring system is constructed, impact data is expanded by adopting a data enhancement method, an impact strengthening physical model is constructed in combination with finite element analysis, impact wave propagation and residual stress evolution are simulated, and collaborative optimization of data driving and mechanism modeling is realized in combination with an experimental data correction model. Impact quality is evaluated based on deep learning, multi-modal data is fused to analyze impact uniformity, and an enhanced abnormal region is identified. And the impact process parameters are optimized through reinforcement learning, and intelligent feedback regulation and control are achieved. According to the method, multi-modal monitoring, data enhancement, physical modeling and intelligent optimization are combined, the stability of laser shock peening is improved, and the fatigue life of laser shock peening is prolonged.

Description

Technical Field

[0001] The present invention belongs to the fields of aviation structural component remanufacturing, intelligent manufacturing and data-driven regulation, and in particular relates to a digital twin-driven data-enhanced aviation landing gear laser shock peening monitoring method. Background Art

[0002] As an important load-bearing structure of aircraft, aviation landing gear is subjected to complex alternating loads for a long time. Its fatigue damage and surface stress distribution directly affect its service safety and service life. Traditional landing gear strengthening mainly relies on methods such as shot peening, but the depth of the strengthening layer is limited, and the residual stress decays rapidly, making it difficult to meet long-term service requirements. Laser shock peening technology uses high-energy laser pulses to generate high-amplitude shock waves on the metal surface, increasing the residual stress level on the material surface and optimizing the microstructure, thereby significantly improving fatigue life. However, during the laser shock peening process, parameters such as impact energy, number of impacts, and spot diameter have an important influence on the strengthening effect. How to accurately monitor the impact quality and optimize process parameters remains an important technical problem in the field of strengthening and remanufacturing.

[0003] Currently, laser shock strengthening quality assessment relies primarily on post-experimental testing and lacks online monitoring methods, making it difficult to obtain real-time information on stress wave propagation, vibration response, and temperature changes during the shock process. Furthermore, the optimization of strengthening parameters under different working conditions still relies on empirical control, making adaptive optimization difficult to achieve. With the development of intelligent manufacturing and data-driven technologies, multimodal sensor monitoring, deep learning modeling, and reinforcement learning optimization have become important directions for improving shock strengthening quality. By combining generative data enhancement, physical mechanism modeling, and intelligent optimization strategies, the accuracy of shock quality assessment can be effectively improved, and intelligent adaptive control of the strengthening process can be achieved, providing technical support for the remanufacturing and service life extension of aircraft landing gear. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides a digital twin-driven data-enhanced aviation landing gear laser shock peening monitoring method, comprising:

[0005] Acquiring experimental data of the landing gear during service, and identifying high stress concentration areas based on the experimental data;

[0006] Constructing a multimodal monitoring system, and collecting data on a process of laser shocking the high stress concentration area based on the multimodal monitoring system to obtain laser shock data samples;

[0007] Expanding the laser shock data sample based on a generative data enhancement method to obtain an expanded data set;

[0008] Constructing a finite element mechanism model, performing numerical simulation analysis based on the finite element mechanism model to obtain numerical simulation data, and fusing the extended data set with the numerical simulation data to obtain a fused data set;

[0009] Constructing a shock quality assessment model based on a spatiotemporal attention graph neural network, inputting the fused data set into the shock quality assessment model for calculation, and obtaining a shock quality assessment result;

[0010] An intelligent feedback optimization model is constructed based on the impact quality evaluation results and the deep reinforcement learning method, and the laser shock process parameters are dynamically adjusted based on the intelligent feedback optimization model.

[0011] Preferably, the process of obtaining experimental data during the service of the landing gear includes:

[0012] Multi-channel force sensors are used to monitor the load evolution of landing gear during service;

[0013] Combined with strain gauges to measure the strain distribution under different working conditions, the material response characteristics under load are analyzed;

[0014] The experimental data are generated by obtaining residual stress data of key parts through X-ray diffraction technology and monitoring heat accumulation effects using infrared thermal imaging.

[0015] Preferably, the process of identifying high stress concentration areas based on the experimental data includes: processing complex alternating loads based on the finite element method to obtain an elastic-plastic mechanical model, simulating stress distribution based on the elastic-plastic mechanical model, evaluating fatigue life in combination with the Miner linear cumulative damage method, and identifying high stress concentration areas.

[0016] Preferably, the process of obtaining the laser shock data sample includes:

[0017] A stress wave sensor is used to record the transient stress waveform induced by laser shock;

[0018] A triaxial accelerometer is used to collect vibration signals during the impact process and analyze the dynamic response of the impact area;

[0019] Monitor the temperature changes in the impact area using an infrared thermal imager to obtain the dynamic evolution of the heat-affected zone;

[0020] A spectral sensor is used to analyze the radiation characteristics of the plasma generated during the laser shock process;

[0021] Based on the dynamic time warping method, multimodal data alignment is performed on the transient stress waveform, the dynamic response of the impact area, the dynamic evolution of the heat-affected zone and the plasma radiation characteristics to obtain the laser impact data sample.

[0022] Preferably, the process of obtaining the extended data set includes:

[0023] Performing data enhancement on the laser shock data sample using a generative adversarial network to obtain a first data set;

[0024] Using a temporal generative adversarial network to enhance the impact process data that conforms to the time series characteristics in the laser impact data sample to obtain a second data set;

[0025] Performing probability modeling on the laser shock data samples based on a variational autoencoder to obtain a third data set;

[0026] Performing cross-modal data generation on the laser shock data samples based on a cross-modal transformation network to obtain a fourth data set;

[0027] Data screening and enhancement are performed on the first data set, the second data set, the third data set, and the fourth data set to obtain the extended data set.

[0028] Preferably, a finite element mechanism model is constructed, and a numerical simulation analysis is performed based on the finite element mechanism model to obtain numerical simulation data. The process includes:

[0029] Based on the theory of elastic-plastic dynamics, a finite element analysis model of laser shock is established to simulate the transient stress changes and shock wave propagation characteristics inside the material during the shock process;

[0030] The Johnson-Cook constitutive model is used to describe the strain rate-dependent plastic behavior of the material and simulate the residual stress evolution under different laser shock energies;

[0031] The residual stress distribution under different working conditions is obtained through laser shock experiments, and the residual stress distribution is compared with the transient stress change, shock wave propagation characteristics and residual stress evolution. The constitutive parameters are adjusted using the least squares optimization method to minimize the error between the numerically simulated stress distribution and the experimentally measured residual stress, thereby obtaining the numerical simulation data.

[0032] Preferably, the shock quality assessment model calculates the shock characteristics of each monitoring point using the expression:

[0033]

[0034] in, represents the impact characteristics of the llth layer, is the neighborhood set of node v, W (l) and b (l) is the trainable weight, and σ is the nonlinear activation function.

[0035] Preferably, the attention weight expression of the impact quality assessment model is:

[0036]

[0037] in, is the time correlation scoring function, is the attention weight.

[0038] Preferably, the intelligent feedback optimization model is generated based on a deep Q network;

[0039] The optimization objective function of the intelligent feedback optimization model is:

[0040]

[0041] Where γ is the discount factor, A t The current laser shock parameter adjustment strategy, R t is the reward function, S t The current system status, including laser shock parameters, workpiece status and monitoring data.

[0042] Compared with the prior art, the present invention has the following advantages and technical effects:

[0043] 1. This invention constructs a multimodal data acquisition and fatigue assessment system based on the complex alternating loads experienced by aircraft landing gear during service. Force sensors, strain gauges, and X-ray diffraction (XRD) technology are used to accurately monitor the stress distribution of the landing gear under different service conditions. The finite element method (FEM) is then used to calculate areas of high stress concentration and residual stress evolution trends. The Miner linear cumulative damage method is used to assess fatigue life, identify potential fatigue failure areas, and determine target locations for laser shock strengthening.

[0044] 2. The present invention combines a multimodal monitoring system with data enhancement technology to precisely control the laser shock process and ensure the stability of the strengthening effect. By integrating multi-source sensors such as stress wave sensing, vibration analysis, infrared thermal imaging, and spectral detection, the dynamic mechanical response of the impact area is monitored in real time, and the dynamic time warping (DTW) method is used to time-align the multimodal data. Furthermore, the impact dataset is expanded using generative adversarial networks (GAN) and temporal generative adversarial networks (TimeGAN), enhancing the diversity and physical consistency of the impact data and improving the training quality of the deep learning model and the accuracy of finite element simulation.

[0045] 3. This invention uses intelligent optimization methods to improve the quality of laser shock peening. It utilizes a spatiotemporal attention graph neural network (STAGNN) to fuse multimodal data such as stress waves, vibration, and temperature fields to accurately assess the quality of the strengthening process and identify areas of potential impact anomalies. Combined with deep reinforcement learning (DRL), the impact parameters are optimized based on real-time monitoring data to ensure uniform residual stress distribution. The impact energy, spot diameter, and number of impacts are intelligently adjusted to improve the consistency of the strengthening process. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] 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:

[0047] Figure 1 Schematic diagram of a method flow in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] 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.

[0049] 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.

[0050] Example 1

[0051] like Figure 1 As shown, this embodiment provides a digital twin-driven data-enhanced aviation landing gear laser shock peening monitoring method, including:

[0052] S1. In view of the complex alternating loads that aircraft landing gear is subjected to during takeoff, landing, and taxiing, a fatigue assessment system is established to identify areas of high stress concentration and provide a basis for target areas for subsequent laser shock strengthening. A multi-channel force sensor is used to monitor the load evolution during the service life of the landing gear, including axial force, tangential force, and normal force. In combination with strain gauges, the strain distribution under different working conditions is measured to analyze the material response characteristics under load. X-ray diffraction (XRD) technology is used to obtain residual stress data in key areas, and infrared thermal imaging is used to monitor the heat accumulation effect to evaluate the impact of alternating stress on material structure.

[0053] Based on the experimental data, an elastic-plastic mechanical model of the landing gear was constructed using the finite element method (FEM). Three-dimensional solid elements were used for meshing, and load boundary conditions were set under different service conditions. The stress distribution function of the landing gear under service conditions was assumed to be σ(x, y, z, t), which satisfies the momentum conservation equation:

[0054]

[0055] Where σ is the stress tensor, F is the external force, ρ is the material density, and vv is the velocity field. The equivalent stress distribution is calculated based on the VonMises criterion:

[0056]

[0057] Among them, σ1, σ2, and σ3 are the principal stress components. For areas of high stress concentration, a local adaptive mesh refinement method is used to improve the calculation accuracy in areas with large stress gradient changes, and the numerical model is corrected using experimental measurement data.

[0058] Based on the fatigue damage accumulation theory, the Miner linear cumulative damage method is used to calculate the fatigue life of the landing gear, and the damage factor is defined as:

[0059]

[0060] Among them, n i is the number of cycles under a certain stress level, N i is the corresponding fatigue life. f When ≥1, the material enters fatigue failure state. The fatigue limit is calculated by combining Goodman's modified criterion:

[0061]

[0062] Among them, σ f is the fatigue limit, σ -1 is the fatigue strength under symmetrical cyclic loading, σ m is the mean stress, σ u The ultimate strength of the material was determined by finite element stress distribution, combined with experimental data to analyze areas of high fatigue damage. This was further verified using surface residual stress distribution data. Ultimately, high stress concentration areas on the landing gear surface were identified as target locations for laser shock peening, providing a basis for subsequent optimization of shock parameters.

[0063] S2. To comprehensively monitor stress wave propagation, vibration response, temperature changes, and plasma characteristics during the laser shock peening process, a multimodal monitoring system was constructed to ensure quantifiable assessment of the strengthening effect in the impacted area. This system integrates multiple measurement methods, including stress wave sensing, vibration sensing, infrared thermal imaging, and spectral analysis. Combined with data synchronization and feature fusion technology, it enables high-precision online monitoring of the impact process.

[0064] First, a stress wave sensor is used to record the transient stress waveform caused by laser shock, including the shock pressure peak, rise time and attenuation characteristics, to evaluate the transmission efficiency of shock energy. The propagation of stress waves satisfies the wave equation:

[0065]

[0066] Where P is the shock wave pressure, c s The uniformity of the impact intensity and the energy absorption can be determined by analyzing the time-frequency characteristics of the shock wave.

[0067] Secondly, a triaxial accelerometer is used to collect vibration signals during the impact process and analyze the dynamic response of the impact area. The vibration data is converted to the frequency domain through fast Fourier transform (FFT) to extract the main vibration modes:

[0068]

[0069] Among them, A(ω) represents the vibration amplitude at different frequencies. Combined with the main frequency characteristics, it can be used to judge the transmission and attenuation of impact energy inside the material.

[0070] In addition, an infrared thermal imager is used to monitor the temperature changes in the impact area and obtain the dynamic evolution of the heat affected zone (HAZ). During the laser impact process, the local temperature rises, and its heat conduction is affected by the thermal conductivity k and specific heat capacity C of the material. p Influence, the temperature distribution satisfies:

[0071]

[0072] Where Q is the heat source term of the laser input. High-frame-rate thermal imaging records the temperature changes in the impact area to ensure that the strengthening process does not generate excessive thermal effects and avoid material performance degradation.

[0073] Finally, a spectral sensor is used to analyze the radiation characteristics of the plasma generated during the laser shock process, and the coupling efficiency of the laser energy and the shock stability are evaluated by monitoring the emission spectrum of the plasma.

[0074] Because each sensor has a different time base, dynamic time warping (DTW) is used to align multimodal data to ensure consistency in data timing. Ultimately, through multimodal feature fusion, a comprehensive monitoring framework for the impact area is constructed, providing data support for accurate assessment of reinforcement effects.

[0075] S3. Due to the influence of multiple factors during the laser shock peening process, such as laser energy, number of shocks, spot diameter, and material properties, the actual data samples collected are limited, and the data distribution has complex nonlinear characteristics. Directly training the monitoring model may result in insufficient generalization ability. Therefore, a generative data augmentation method is introduced to expand the multimodal dataset of the shock process and improve the adaptability and robustness of the monitoring model under different working conditions.

[0076] The Generative Adversarial Network (GAN) is used to enhance the shock stress wave, vibration response, temperature field and plasma spectrum data to simulate the data distribution under different shock parameters and working conditions. Assume that the real data set is:

[0077] D={X stress ,X vib ,X temp ,X plasma}

[0078] Among them, X stress Represents stress wave data, X vib Represents vibration data, X temp Represents temperature data, X plasma Represents plasma spectrum data. GAN consists of a generator G(z) and a discriminator D(x), and the optimization objective is:

[0079]

[0080] Among them, P real represents the distribution of real shock data, P z is a random noise distribution. Through adversarial training, the generator is continuously optimized, and the generated data gradually approaches the true distribution, thereby expanding the data set and improving the monitoring model's adaptability to different laser shock parameters.

[0081] During the laser shock peening process, the data of various physical fields change dynamically over time. Simply generating independent data points based on GAN may ignore the time dependency. Therefore, a time-generative adversarial network (TimeGAN) is used to generate shock process data that conforms to the time series characteristics. Assume that the original time series data is:

[0082] X={X1,X2,…,X T}

[0083] TimeGAN combines the generative adversarial network (GAN) and recurrent neural network (RNN) structure, maps the time series to the latent variable space Z through the encoder-decoder, and models the time dependency in the latent variable space to optimize the objective function:

[0084]

[0085] Among them, D KL The Kullback-Leibler divergence is used to ensure the distribution consistency between the generated data and the real data. TimeGAN is trained to enhance the dynamic characteristics of the impact process, enabling the monitoring system to adapt to different laser impact intensities and working conditions, and improving prediction accuracy.

[0086] Directly using data-driven GAN to generate data may result in the generated data not satisfying the physical laws of impact. Therefore, physical constraint optimization (Physics-Informed Learning) is introduced in the data enhancement process to make the generated data conform to the physical model of laser impact. For example, stress waves satisfy the one-dimensional wave equation:

[0087]

[0088] Where P is the shock wave pressure, c s is the stress wave propagation velocity of the material. This equation is used as a regularization constraint for the generator, so that the data generated by GAN conforms to the energy transfer law of shock waves. In addition, the temperature field data satisfies the heat conduction equation:

[0089]

[0090] Among them, C p is the specific heat capacity, k is the thermal conductivity, and Q is the heat source term of the laser shock input. The data generated by TimeGAN is optimized with physical constraints to make it conform to the actual distribution of the heat-affected zone (HAZ) and improve the credibility of the data.

[0091] For nonlinear impact data distribution, a variational autoencoder (VAE) is used to perform probability modeling on impact data to enhance data diversity. Let the input data be X, VAE maps it to latent variables ZZ through the encoder, obeying a certain prior distribution p(Z), and then reconstructed by the decoder as Optimized Evidence Lower Bound (ELBO):

[0092]

[0093] Among them, q(Z|X) is the variational posterior distribution, D KLThe Kullback-Leibler divergence is calculated by VAE to generate impact data that conforms to the real distribution and is used for data enhancement to improve the generalization ability of the monitoring model.

[0094] During the laser shock peening process, complex nonlinear coupling relationships exist between various physical quantities (stress waves, vibrations, temperature, and plasma spectra). A cross-modal translation network (CMTN) is used to learn the mapping relationship between different modal data. For example, under limited stress wave data, the corresponding temperature field data can be predicted using the CMTN:

[0095] X temp =f(X stress )

[0096] At the same time, CycleGAN is used for cross-modal data generation to achieve the mapping from plasma spectrum data to stress wave data:

[0097] G(X plasma )=X stress ,F(X stress )=X plasma

[0098] Among them, GG and FF are two generators that compete with each other through cycle consistency loss:

[0099]

[0100] Ensure cross-modal consistency of generated data, improve joint modeling capabilities between different physical quantities, and enhance the integrity and consistency of impact process data.

[0101] By integrating the data generated by GAN, TimeGAN, PI-GAN and VAE, we adopt data screening and enhancement strategies to select the data set that best conforms to physical laws and covers different working conditions. real , generate data as X gen , define the data enhancement loss:

[0102]

[0103] Here, α and β are regularization parameters that optimize the data distribution to align with actual operating conditions. Ultimately, the enhanced data is used to train the laser shock peening monitoring model, improving the prediction accuracy for different impact parameters, material types, and fatigue damage states. This ensures the stability and adaptability of the monitoring system, providing high-quality data support for subsequent strengthening effect evaluations.

[0104] During laser shock peening (S4), stress evolution within the material is influenced by multiple factors, including laser energy, number of shocks, spot diameter, and material properties. While experimental data can provide some information about the shock process, its distribution is limited and cannot directly resolve the internal stress state. Therefore, a finite element mechanism model was introduced to analyze shock wave propagation, residual stress redistribution, and material hardening effects through numerical simulation. Corrections were then made based on experimental data to improve the physical consistency of the data enhancement and achieve collaborative optimization of data-driven and physical modeling.

[0105] First, a finite element analysis (FEA) model of laser shock was established based on the elastic-plastic dynamics theory to simulate the transient stress changes and shock wave propagation characteristics of the material during the impact process. Assuming that the material obeys the isotropic constitutive relationship, its dynamic stress distribution satisfies the momentum conservation equation:

[0106]

[0107] Where σ is the stress tensor, F is the impact force, ρ is the material density, and v is the velocity vector. The transient propagation of the shock wave is described by the wave equation:

[0108]

[0109] Where P is the impact pressure, c s is the propagation speed of the shock wave in the material. Adaptive mesh refinement technology is used to improve the calculation accuracy of high gradient stress areas and ensure accurate simulation of residual stress distribution.

[0110] Under the action of impact, a plastic deformation layer is formed on the surface of the material, and the residual stress redistribution is mainly affected by the strengthening and strain hardening effects of the material. The Johnson-Cook constitutive model is used to describe the strain rate-dependent plastic behavior of the material, and its flow stress expression is:

[0111]

[0112] Among them, A, B, C, n, m are material parameters, ε is the equivalent plastic strain, is the normalized strain rate, T * By simulating the residual stress evolution under different laser shock energies, the depth of the strengthening layer and the stress gradient distribution are analyzed, and the shock parameter selection is optimized.

[0113] The experimental data are used to modify the finite element calculation results and improve the physical consistency of the model. First, the residual stress distribution under different working conditions is obtained through laser shock experiments and compared with the stress results of numerical simulation. The constitutive parameters are adjusted using the least squares optimization method to make the stress distribution σ of the numerical simulation sim The residual stress σ expThe error between them is minimal:

[0114] Here, θ\theta represents the material parameter set of the finite element model, including the strain hardening parameter, dynamic strengthening coefficient, and thermal softening coefficient. Through iterative optimization, we ensure that the simulated data accurately reflects the stress evolution characteristics during the actual impact process, allowing the finite element calculation results to be used to assist in data enhancement and improve the predictive accuracy of the data-driven model.

[0115] Ultimately, through the collaborative enhancement of the finite element mechanism model, the fusion of experimental data and numerical simulation data is achieved, which supplements the deficiencies of experimental data and makes the data enhancement method not only rely on the generative model but also be constrained by the physical mechanism, thereby improving the credibility and adaptability of the impact strengthening monitoring system.

[0116] S5. The quality of laser shock peening directly impacts the fatigue life and service stability of landing gear. Therefore, a comprehensive assessment of post-impact stress distribution, reinforcement layer depth, and impact uniformity is required. This assessment should be combined with anomaly detection technology to identify areas of insufficient reinforcement or uneven impact. Because impact quality involves multimodal data such as stress waves, vibration signals, temperature fields, and plasma spectra, a comprehensive assessment model using data-driven deep learning methods is developed to improve the stability and robustness of the monitoring system.

[0117] To comprehensively analyze the dynamic changes of the impact process, an impact quality assessment model based on the Spatio-Temporal Attention Graph Neural Network (STAGNN) is constructed. This model combines the spatial information extraction capability of the Graph Neural Network (GNN) with the dynamic feature capture capability of the Attention Mechanism to improve the modeling capability of impact quality. Assume that the monitoring dataset in the impact area is:

[0118] D={X stress ,X vib ,X temp ,X plasma}

[0119] Among them, X stress Represents stress wave data, X vib Represents vibration data, X temp Represents temperature data, X plasma Represents plasma spectral data. First, a graph neural network is used to establish the spatial topology of the impact area. The graph consisting of monitoring points is set as G = (V, E), where V is the set of monitoring points and E is the dynamic relationship between physical quantities. The impact characteristics of each monitoring point are calculated through graph convolution:

[0120]

[0121] in, represents the impact characteristics of the llth layer, is the neighborhood set of node v, W (l) and b (l) is the trainable weight, and σ is the nonlinear activation function.

[0122] In the time dimension, the attention mechanism is introduced to calculate the importance of different time steps to ensure that the model focuses on the feature changes at the key impact moments. Define the attention weight:

[0123]

[0124] in, A temporal correlation scoring function is used, using a learnable feedforward network to calculate the feature importance at different time steps. Finally, the spatial features of the graph neural network are combined with the temporal dynamics of the attention mechanism to predict the reinforcement quality of the impact area and evaluate the impact uniformity.

[0125] In the anomaly detection part, STAGNN is used to calculate the residual stress distribution in the impact area and compare it with the experimental measurement data to calculate the impact uniformity index:

[0126]

[0127] in, is the average residual stress in the impact area, and N is the number of measurement points. σ When the set threshold is exceeded, it is considered that there is uneven impact in the area.

[0128] Finally, a complete impact quality assessment system was constructed based on STAGNN to realize the fusion analysis of multimodal data, improve the accuracy of impact monitoring, provide data support for optimizing laser shock strengthening process parameters, and ensure the impact uniformity and stability of the strengthening effect.

[0129] S6. The process parameters of laser shock peening directly affect the residual stress distribution and fatigue performance of the landing gear surface. Therefore, it is necessary to optimize process parameters such as laser energy, spot diameter, pulse width, and number of shocks based on the shock quality assessment results, and construct an adaptive optimization model to improve the shock peening effect. Because the strengthening process involves complex nonlinear dynamic mechanisms, traditional experience-based parameter control methods are difficult to adapt to different working conditions. Therefore, Deep Reinforcement Learning (DRL) is used to build an intelligent feedback optimization system, which enables the process parameters to be dynamically adjusted based on historical strengthening data to achieve the optimal laser shock peening strategy.

[0130] The parameter space of laser shock peening is set as:

[0131] P={E,d,τ,N}

[0132] Where E is the laser energy, d is the spot diameter, τ is the pulse width, and N is the number of impacts. σ , Depth of Strengthening Layer D h and fatigue life improvement rate R f As the optimization goal of reinforcement learning, a mapping relationship between process parameters and reinforcement quality is constructed. Deep Q-Network (DQN) is used for parameter optimization, defining the state S, action A, and reward function RR. Assume that the state of the impact process is:

[0133] S t ={X stress ,X vib ,X temp ,X plasma ,P t}

[0134] Among them, X stress 、X vib 、X temp 、X plasma Represents the current impact monitoring data, P t is the current process parameter. The reinforcement learning strategy is represented by the Q function, and the optimization goal is:

[0135]

[0136] Where γ is the discount factor, A t The current laser shock parameter adjustment strategy, R t is the reward function. The reward function is defined as:

[0137] R t =αS r -βU σ -λΔT

[0138] Among them, S r is the improvement rate of residual stress after impact, U σ is the stress uniformity, ΔT is the temperature change in the heat affected zone, and α, β, and λ are weight parameters.

[0139] An Experience Replay mechanism is used to store monitoring data from different impact strategies and train the DQN model, enabling it to learn the optimal parameter adjustment strategy under different impact conditions. After multiple iterations of reinforcement training, the model can automatically adjust parameters such as laser energy and number of impacts to achieve optimal residual stress uniformity, avoid local stress concentration, and improve the quality of the reinforcement layer.

[0140] Ultimately, the intelligent feedback optimization system based on reinforcement learning realizes adaptive control of the laser shock peening process, improves shock stability and optimization efficiency, ensures the best strengthening effect for different landing gear models, materials and service conditions, and improves the fatigue life and service reliability of the landing gear.

[0141] Furthermore, the landing gear service load analysis and fatigue assessment in step S1 specifically includes the following methods:

[0142] During the service life of aircraft landing gear, the complex alternating loads it bears lead to local high stress concentration areas. In order to optimize laser shock strengthening (S4), service load analysis and fatigue assessment must be performed first. A multi-channel force sensor is used to collect axial force, tangential force, and normal force data. Strain gauges are used to measure the strain distribution under different working conditions, and X-ray diffraction (XRD) is used to obtain residual stress in key areas. Based on the finite element method (FEM), an elastic-plastic mechanical model is constructed to simulate the stress distribution of the landing gear under typical load conditions and calculate the equivalent stress:

[0143]

[0144] Among them, σ1, σ2, and σ3 are the main stress components. Combined with Miner's linear cumulative damage theory to calculate fatigue life, the damage factor is defined as:

[0145]

[0146] Among them, n i is the number of cycles under a certain stress level, N i is the corresponding fatigue life. f When ≥1, the material enters a fatigue failure state and needs to be laser shock strengthened in S4. The stress gradient data calculated in S1 provides benchmark data for the S2 monitoring system and serves as a real working condition sample for S3 data enhancement.

[0147] Furthermore, the multimodal monitoring system in step S2 specifically includes the following method:

[0148] To monitor stress wave propagation, vibration response, temperature changes, and plasma characteristics during laser shock peening in real time, S2 uses multimodal sensors (stress wave, triaxial accelerometer, infrared thermal imaging, and spectral analysis) to monitor the shock process online. Due to the time series deviation of the data collected by different sensors, dynamic time warping (DTW) is used for time series alignment. The matching error is set as:

[0149]

[0150] Among them, X, Y are different modal data sequences, π is the matching path, d(x i ,yj ) is the distance function. The aligned data is used for data enhancement in S3 to improve the stability of cross-operation monitoring. Data such as the impact energy distribution and temperature field evolution monitored by S2 are fed into S5 for quality assessment and provide input for feedback optimization in S6.

[0151] Furthermore, the data enhancement method in step S3 specifically includes the following methods:

[0152] Since the actual impact data collected is limited, S3 uses generative adversarial networks (GANs) and temporal generative adversarial networks (TimeGANs) to expand the multimodal dataset to enhance the adaptability of the monitoring model. GAN generates impact stress waves, temperature fields, and vibration data through adversarial training, and the optimization objective function is as follows:

[0153]

[0154] G(z) generates synthetic data, while D(x) distinguishes the data from authenticity. TimeGAN further incorporates time-dependent modeling to imbue the generated data with realistic dynamic characteristics, enhancing the reliability of S4's mechanism modeling. Furthermore, the generated data is used to train the S5 impact quality assessment model, improving its adaptability across various operating conditions.

[0155] Furthermore, the laser shock peening optimization in step S4 specifically includes the following methods:

[0156] S4 constructs a finite element analysis (FEM) model based on the high stress areas identified by S1 and the data generated by S3 to simulate the laser shock wave propagation and residual stress redistribution. r The relationship with laser shock parameters (laser energy P, spot diameter d, pulse width τ, number of shocks N) is:

[0157] S r =f(P,d,τ,N)

[0158] Experimental data is used to optimize this relationship, improving the consistency and stability of strengthening. After impact strengthening, residual stress distribution data is input into S5 for quality assessment and provides optimization reference for S6, allowing the strengthening process to be adaptively adjusted.

[0159] Furthermore, the impact quality assessment method in step S5 specifically includes the following method:

[0160] S5 uses the spatiotemporal attention graph neural network (STAGNN) to fuse stress wave, vibration, temperature and plasma spectrum data to predict the strengthening quality of the impact area and identify the abnormal strengthening area. The impact quality uniformity index is:

[0161]

[0162] in, is the average residual stress in the impact area, and N is the number of measurement points. σ If the threshold is exceeded, the reinforcement parameters in S6 need to be optimized. The reinforcement depth data evaluated in S5 is used to train the reinforcement learning (DRL) model in S6 to optimize the impact strategy and improve the consistency of reinforcement.

[0163] Furthermore, the laser shock process optimization method in step S6 specifically includes the following method:

[0164] S6 uses deep reinforcement learning (DRL) to optimize impact parameters, making the residual stress uniformly distributed and improving the fatigue life of the landing gear. t is the current impact parameter and reinforcement quality, action A t For parameter adjustment, the optimization goal is:

[0165]

[0166] Where γ is the discount factor, R t is the reward function, which is calculated as:

[0167] R t =αS r -βU σ -λΔT

[0168] Among them, S r is the improvement rate of residual stress after impact, U σ is stress uniformity, and ΔT is the temperature change in the heat-affected zone. S6 optimizes strengthening parameters to improve impact uniformity and provides feedback to S5 for quality assessment to ensure that the optimized strengthening strategy maintains high reliability under different working conditions.

[0169] 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 digital twin-driven data-enhanced aviation landing gear laser shock peening monitoring method, characterized in that: include: Acquiring experimental data of the landing gear during service, and identifying high stress concentration areas based on the experimental data; Constructing a multimodal monitoring system, and collecting data on a process of laser shocking the high stress concentration area based on the multimodal monitoring system to obtain laser shock data samples; Expanding the laser shock data sample based on a generative data enhancement method to obtain an expanded data set; Constructing a finite element mechanism model, performing numerical simulation analysis based on the finite element mechanism model to obtain numerical simulation data, and fusing the extended data set with the numerical simulation data to obtain a fused data set; Constructing a shock quality assessment model based on a spatiotemporal attention graph neural network, inputting the fused data set into the shock quality assessment model for calculation, and obtaining a shock quality assessment result; An intelligent feedback optimization model is constructed based on the impact quality evaluation results and the deep reinforcement learning method, and the laser shock process parameters are dynamically adjusted based on the intelligent feedback optimization model.

2. The method according to claim 1, characterized in that The process of obtaining experimental data during the service of the landing gear includes: Multi-channel force sensors are used to monitor the load evolution of landing gear during service; Combined with strain gauges to measure the strain distribution under different working conditions, the material response characteristics under load are analyzed; The experimental data are generated by obtaining residual stress data of key parts through X-ray diffraction technology and monitoring heat accumulation effects using infrared thermal imaging.

3. The method according to claim 1, characterized in that The process of identifying high stress concentration areas based on the experimental data includes: processing complex alternating loads based on the finite element method to obtain an elastic-plastic mechanical model, simulating stress distribution based on the elastic-plastic mechanical model, evaluating fatigue life in combination with the Miner linear cumulative damage method, and identifying high stress concentration areas.

4. The method according to claim 1, wherein The process of obtaining the laser shock data sample includes: A stress wave sensor is used to record the transient stress waveform induced by laser shock; A triaxial accelerometer is used to collect vibration signals during the impact process and analyze the dynamic response of the impact area; Monitor the temperature changes in the impact area using an infrared thermal imager to obtain the dynamic evolution of the heat-affected zone; A spectral sensor is used to analyze the radiation characteristics of the plasma generated during the laser shock process; Based on the dynamic time warping method, multimodal data alignment is performed on the transient stress waveform, the dynamic response of the impact area, the dynamic evolution of the heat-affected zone and the plasma radiation characteristics to obtain the laser impact data sample.

5. The method according to claim 1, characterized in that The process of obtaining the extended data set includes: Performing data enhancement on the laser shock data sample using a generative adversarial network to obtain a first data set; Using a temporal generative adversarial network to enhance the impact process data that conforms to the time series characteristics in the laser impact data sample to obtain a second data set; Performing probability modeling on the laser shock data samples based on a variational autoencoder to obtain a third data set; Performing cross-modal data generation on the laser shock data samples based on a cross-modal transformation network to obtain a fourth data set; Data screening and enhancement are performed on the first data set, the second data set, the third data set, and the fourth data set to obtain the extended data set.

6. The method according to claim 1, characterized in that The process of constructing a finite element mechanism model, performing numerical simulation analysis based on the finite element mechanism model, and obtaining numerical simulation data includes: Based on the theory of elastic-plastic dynamics, a finite element analysis model of laser shock is established to simulate the transient stress changes and shock wave propagation characteristics inside the material during the shock process; The Johnson-Cook constitutive model is used to describe the strain rate-dependent plastic behavior of the material and simulate the residual stress evolution under different laser shock energies; The residual stress distribution under different working conditions is obtained through laser shock experiments, and the residual stress distribution is compared with the transient stress change, shock wave propagation characteristics and residual stress evolution. The constitutive parameters are adjusted using the least squares optimization method to minimize the error between the numerically simulated stress distribution and the experimentally measured residual stress, thereby obtaining the numerical simulation data.

7. The method according to claim 1, characterized in that The shock quality assessment model calculates the shock characteristics of each monitoring point using the following expression: in, represents the impact characteristics of the llth layer, is the neighborhood set of node v, W (l) and b (l) is the trainable weight, and σ is the nonlinear activation function.

8. The method according to claim 1, characterized in that The attention weight expression of the impact quality assessment model is: in, is the time correlation scoring function, is the attention weight.

9. The method according to claim 1, characterized in that The intelligent feedback optimization model is generated based on a deep Q network; The optimization objective function of the intelligent feedback optimization model is: Where γ is the discount factor, A t The current laser shock parameter adjustment strategy, R t is the reward function, S t The current system status, including laser shock parameters, workpiece status and monitoring data.

Images