Research On Prediction Method Of TC4 Non-isothermal Sheet Metal Hot Bending Forming Accuracy

In this paper,TC4 sheet metal is taken as the research object,and a forming accuracy prediction method based on finite element analysis and machine learning algorithms is proposed by using self-resistive heating technology for forming.The purpose is to improve the accuracy of prediction forming.The main research contents of the paper are as follows:(1)Established a Johnson Cook constitutive model based on genetic algorithm optimization.Based on the hot stretching experimental data of TC4 material,the Johnson Cook constitutive model parameters were fitted and optimized using genetic algorithm.The optimized constitutive model is in good agreement with experimental data and can accurately describe the stress-strain relationship of TC4 material at different temperatures,providing an accurate material model for finite element simulation.(2)The finite element numerical simulation of self-resistance heating forming process was conducted.The temperature field distribution of the titanium plate was obtained by simulating the self-resistance heating temperature field of the TC4 plate.By simulating TC4 non isothermal sheet metal hot bending forming,the rebound situation under different temperature parameters was analyzed,and sample data was established for machine learning rebound prediction.(3)A machine learning based prediction model for hot bending forming accuracy has been established.The data obtained through finite element simulation was used as the sample database,and a machine learning model was constructed using gradient lifting regression algorithm.The node and corresponding temperature information were used as inputs,and the deformation amount was used as output.The machine learning model was used for prediction.By analyzing the corresponding relationship between predicted deformation and node coordinates and 3D shapes,a node data mapping model was established to achieve visualization of machine learning prediction results.Error analysis was conducted by extracting curvature radius and forming angle,and the results showed that the machine learning prediction model has high accuracy and meets the usage requirements.(4)TC4 non isothermal sheet metal hot bending process experiments were conducted to verify and analyze the accuracy of machine learning prediction forming under different temperature parameters.The results indicate that the machine learning prediction results are highly consistent with the experimental results,demonstrating its prediction accuracy and reliability,and verifying the feasibility of using machine learning as a method for predicting the accuracy of TC4 non isothermal sheet metal hot bending forming.