Scholarlink

<p>To address machining-induced deformation in aluminum alloy thin-walled structures after milling and the strong reliance of mechanical hammer peening straightening parameters on empirical experience, this study proposed a neural network-based prediction method using a 7075-T651 aluminum alloy T-shaped thin-walled component as the research object. A finite element model of the mechanical hammer peening straightening process was established in Abaqus to simulate the straightening value responses under different process parameters. By comparing simulation results with experimental data, systematic error characteristics were identified, and a Latin hypercube sampling method was employed to design experiments. A regression-based correction model was then developed to modify the simulation results, yielding a reliable dataset. The regression model evaluation showed high accuracy, with coefficients of determination (R²) greater than 0.95 and p-values less than 0.01 in both X and Y directions. Based on the corrected data, a backpropagation (BP) neural network was constructed to model the nonlinear relationship between hammer peening parameters and straightening value, and a particle swarm optimization (PSO) algorithm was introduced to optimize the network weights and thresholds. The predictive performances of the BP and PSO-BP models were evaluated using five-fold cross-validation, as well as the coefficient of determination (R²), root mean square error (RMSE), and mean absolute error (MAE). The results indicated that the PSO-BP model achieved higher prediction accuracy and stability than the conventional BP network. Finally, the PSO-BP model was applied to predict hammer peening parameters, and the predictions were validated through finite element simulations and experiments, confirming the effectiveness and accuracy of the proposed method. This approach provides a feasible technical route for achieving high-precision and controllable mechanical hammer peening straightening of aluminum alloy thin-walled components.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Prediction of mechanical hammer peening parameters for correcting distortions of 7075 aluminum alloy thin-walled parts based on PSO-BP model

  • 期刊论文

  • 发表日期：

    2026年7月2日

查看全文
  • Xiaohui Lin,
  • Na Fu,
  • Jianchun Liu,
  • Mingwei Chen,
  • Yukun Zhou

摘要

To address machining-induced deformation in aluminum alloy thin-walled structures after milling and the strong reliance of mechanical hammer peening straightening parameters on empirical experience, this study proposed a neural network-based prediction method using a 7075-T651 aluminum alloy T-shaped thin-walled component as the research object. A finite element model of the mechanical hammer peening straightening process was established in Abaqus to simulate the straightening value responses under different process parameters. By comparing simulation results with experimental data, systematic error characteristics were identified, and a Latin hypercube sampling method was employed to design experiments. A regression-based correction model was then developed to modify the simulation results, yielding a reliable dataset. The regression model evaluation showed high accuracy, with coefficients of determination (R²) greater than 0.95 and p-values less than 0.01 in both X and Y directions. Based on the corrected data, a backpropagation (BP) neural network was constructed to model the nonlinear relationship between hammer peening parameters and straightening value, and a particle swarm optimization (PSO) algorithm was introduced to optimize the network weights and thresholds. The predictive performances of the BP and PSO-BP models were evaluated using five-fold cross-validation, as well as the coefficient of determination (R²), root mean square error (RMSE), and mean absolute error (MAE). The results indicated that the PSO-BP model achieved higher prediction accuracy and stability than the conventional BP network. Finally, the PSO-BP model was applied to predict hammer peening parameters, and the predictions were validated through finite element simulations and experiments, confirming the effectiveness and accuracy of the proposed method. This approach provides a feasible technical route for achieving high-precision and controllable mechanical hammer peening straightening of aluminum alloy thin-walled components.