"Warping-Strain" Coupling Prediction And Control Of Aluminum Alloy Plate During Hot Rolling

Aluminum alloys are widely used in various engineering structures due to their advantages of high strength,good toughness,and low cost.Serpentine rolling is a new type of technical method for rolling thick plates.It not only can fully deform the center of the thick plate,but also breaks up the coarse grains in the cast state,and completely transforms the plate from the cast structure to the processed structure.On the other hand,the reason for the technical characteristics of the serpentine hot rolling is that it will cause warpage of the plate,which will prevent the rolling stock from entering the next pass.In severe cases,it may cause the rolling stock to crash into the rolling mill,causing serious problems.Industrial accidents.Since its development is in the primary stage,most studies only explore the laws of a single process parameter.Therefore,comprehensive consideration of various process parameters,control of slab shape,and improvement of deformation in the thickness direction of the rolling stock,progresses in the development of snake-shaped rolling.necessary.In this paper,the 7075 aluminum alloy as the research goal,the use of ABAQUS simulation software to study the serpentine rolling numerical simulation technology,and the use of neural network nonlinear mapping ability and genetic algorithm optimization ability of the average thickness of the equivalent strain and plate warping Coupling prediction and control of factors.The results of this research are as follows:(1)Through the serpentine hot-rolling simulation simulation,the effects of initial thickness,allocating ratio and reduction on the warping factor were studied.(2)By studying the constitutive model of materials,friction type,thermal boundary properties and other key technologies,the finite element model of aluminum alloy thick plate serpentine rolling is constructed to obtain the training samples of neural network.The warping factor measurement experiment shows that the finite element The simulation values are in good agreement with the measurement data,with only about 13% error between them.(3)Selecting input samples using orthogonal experimental design methods.With the help of the neural network's nonlinear mapping ability,a coupled prediction model with average equivalent strain and warping factor is constructed from a finite number of training samples,and the predicted values are correspondingly limited.The meta simulation results show that the prediction error does not exceed 7.2%.(4)Under the condition that the warping factor does not exceed a certain threshold,an optimization model of the rolling process parameters is set up to maximize the average equivalent strain.The individual fitness function is constructed by the average equivalent strain of the plate and the rolling process parameters are optimized.The model of genetic algorithm solving technology.The proposed "finite element analysis-neural network prediction-genetic algorithm optimization" integrated control method not only can improve the calculation efficiency of plate rolling process simulation,but also can achieve reasonable design and selection of aluminum alloy plate rolling process parameters.