| Sheet metal stamping(SMS)is a very important manufacturing technology,it has a wide range of applications in the automobile industry,chemical industry,aviation,machinery and electronics and national defense industry.Generally,the SMS process involves many factors such as material properties,structure of the blank and mould,friction,BHF,most of which are coupled together to determine the quality of the products.Therefore,SMS has long been a hot research filed in the world and how to find a reasonable ratio among the coupled parameters is one of the most challenging topics.In this paper,a pharmaceutical bottle cup made by aluminium alloy 8011 is taken as study object.To reduce the high thinning rate of bulge loop inner wall on the cap top,orthogonal experiment,stamping numerical simulation and Artificial Neural Network(ANN)are adopted.A comprehensive assessment about the influence of the main parameters is acquired.As a result,Ann is successfully applied to find the parameter optimized matching.At the very beginning,tensile test of aluminium alloy 8011 is conducted on an electron-tensile tester(BTM5104)to analysis its material properties.Then,by using orthogonal experiment,a FEM simulation scheme is designed to obtain the better proportional parameter series for SMS process.Further parameter optimization is made by ANN based on the result of orthogonal experiment optimization.The mathematical model of BP neural network is deduced.A three-layers BP networ with 5 input and 1 output is built in MATLAB.In this BP network,the result of orthogonal experiment is used as a high quality sample to obtain a better forecasting effect about minimum thickness.The results show that the network has a higher precision(below 2%).Finally,optimal ratio of process parameters is acquired through the BP network,the result is better than orthogonal experiment.Study results indicate that ANN has obvious advantages to deal multi-parameter coupling question,especially in pressing process. |
PDF Full Text Request