Optimization Of Indirect Isothermal Extrusion Process For 7055 Aluminum Alloy Based On Machine Learning Prediction Model

High-strength aluminum alloy isothermal extrusion process is a precision aluminum plastic deformation processing technology.Traditionally,the extrusion and heat treatment processes are determined based on the combination of standardized alloy preparation systems and experiments,and production experience.However,when there are fluctuations in raw material quality or new alloys are developed,it often requires trial extrusion to adjust the process,resulting in low productivity,low efficiency,and high costs.To address these issues,researchers proposed to introduce artificial intelligence expert systems into extrusion processing technology.Artificial intelligence expert systems can form a knowledge base for extrusion processing systems and make automatic decisions through knowledge reasoning,thereby improving the intelligent level of extrusion processing.This technology has become the core driving mechanism for extrusion production,promoting the transformation and upgrading of traditional manufacturing towards intelligent manufacturing.In this context,the aim of this study is to combine numerical simulation and artificial intelligence expert systems,by incorporating extrusion processing experience knowledge,to construct an indirect isothermal extrusion process optimization method based on machine learning prediction models.The method aims to predict product quality through numerical simulation,optimize extrusion processing parameters,improve production efficiency,and reduce material waste and energy consumption,thereby achieving flexible customization of high-strength aluminum alloy extrusion processing.The main work of this paper is as follows:A unidirectional thermal compression test with a 60% deformation ratio was conducted on cast 7055 aluminum alloy,and true stress-strain curves were obtained at different temperatures and deformation rates.The thermal deformation behavior was analyzed,and a constitutive model for the aluminum alloy was established based on the Arrhenius equation.Subsequently,the constitutive equation was incorporated into finite element numerical calculations.A finite element simulation model was developed for the indirect extrusion process of 7055 aluminum alloy rods,and the material rheological and microstructural evolution behaviors during extrusion were analyzed.The velocity and stress-strain fields under different extrusion process parameters were obtained through finite element simulations,and the effects of extrusion parameters on extrusion temperature,deformation,microstructure,and mechanical behavior were studied.Furthermore,a digital process optimization model was constructed by combining the established finite element model and artificial neural network.By digitally modeling 216 different indirect extrusion cases,the mapping relationship between extrusion control process parameters,outlet temperature of the profile,and product performance was obtained.By using the generated indirect extrusion dataset,a backpropagation neural network model was established,which can accurately predict the relationship between indirect extrusion process parameters and material properties,with an average prediction error of 0.78%.Based on this,an indirect extrusion limit diagram for 7055 aluminum alloy was established in this study,providing a basis for rapid quality control of indirect extrusion.Based on the above model,for a specific extrusion profile,the isothermal extrusion velocity was determined by the extrusion limit diagram.A dataset containing18 outlet temperature variation sequences was constructed.Using the Temporal Fusion Transformer model,a predictive model was established for the outlet temperature sequence of the aluminum profile during isothermal extrusion.The constructed model has good generalization ability and performs well on the validation set,with an average error of 1.8%.The extrusion velocity-stroke control curve for isothermal extrusion of aluminum profiles was optimized while ensuring maximum efficiency and defect-free production.The temperature difference at the outlet area of the extruded profile obtained by isothermal extrusion did not exceed 5℃.By using machine learning,the finite element simulation or experiments in the optimization of the indirect isothermal extrusion process can be omitted,achieving rapid optimization of the indirect isothermal extrusion process.It has been found through the practical application of machine learning in extrusion processing optimization that machine learning models can discover hidden patterns and rules by learning from a large amount of historical data,and predict the profiles state under different process parameters.This data-driven modeling method can optimize process parameters more accurately and efficiently,thereby improving production efficiency and product quality.