| Wrinkle phenomenon has become a key factor affecting the forming of thin-walled parts,especially in the rapid development of automobile,national defense,aviation and other industries,the quality and strength of thin-walled parts are more stringent.Therefore,the research on the accuracy and regularity of wrinkle instability has become an important subject in the field of forming manufacturing.In order to more accurately predict the wrinkling and instability of the sheet metal,our research group designed different test molds,finished wrinkling and instability tests under different working conditions(YBT test,wedge tensile wrinkling test,and cone deep drawing sidewall wrinkling test).After summarized and analyzed the critical wrinkling stress ratio and strain ratio data under three different working conditions,as well as highprecision fitting,not only has a unified critical wrinkling judgment line been established for different forming conditions of metal sheet metal,but also the prediction scope of wrinkling instability in sheet metal forming has been further expanded.To further simplify the test process,our research group successfully achieved the goal of passing a set of shearing dies,only test pieces of different shapes are changed(rectangular test pieces,upper and lower test pieces,circular test pieces,rectangular perforated test pieces).Finally,the unified critical wrinkling judgment line is obtained.This study established the unified wrinkling determination line of sheet material under four different working conditions as an example.With the help of the secondary development function provided by ABAQUS,the python scripting language was used to write the plug-in registration script and plug-in GUI script,established the user interface,and developed the kernel execution script.Parameterization,automatic modeling and automatic output of wrinkling judgment line are realized.The BP neural network prediction model for the uniform wrinkling of thin plate plates is established using the drawn decision line data.In order to verify the accuracy of neural network prediction results,different methods are adopted for verification.Compared the judgement line predicted by neural network with that by finite element numerical simulation,the results showed that the predictions of the two methods are basically the same.On this basis,the judgment line predicted by neural network is innovatively proposed as the judgment basis.Based on the wedge test,the stress ratio and strain ratio coordinates corresponding to the integral points on the thickness of all nodes of the finite element simulation results are extracted,and the positions are compared with the predicted judgment line,which are distinguished by different colors,and the results meet the expectations.The above research contents and results indicate that: neural network has more efficient forecasting ability,it can achieve fast output of uniform wrinkle detection line and has important practical application value for production practice.Simultaneously,the uniform wrinkle judgment line established in this subject can mark the wrinkle area of the specimen and obtain the cloud picture of the wrinkle area of the specimen. |
