Process Analysis Of Radial Forging Based On Neural Network

Radial forging is a special forging process for forging shafts and pipes.Due to its advantages of high material utilization,good mechanical properties and high production efficiency,it has been widely used in industrial production.However,the radial forging process has a lot of process parameters,when they are not properly set,the forgings will have various defects,such as: if forging speed is too fast,forgings are prone to distortion;friction between forging and hammers leds to non-uniform distribution of the metal axial flow in the cross section of forging and makes concave end surface,affecting dimensional accuracy and reducing material utilization rate;because the compression rate of radial forging process is generally small,uneven deformation distribution of the surface and the core is liable to cause axial tensile stress and may lead to rupture of forgings;and so on.Based on the above reasons,this paper studies the influence of process parameters on the forging load and the end surface during hot radial forging of circular cross-section forgings using flat hammers and arc hammers.In addition,the T-S fuzzy neural network is used for prediction and optimization using the finite element simulation results.The main research contents are as follows:Firstly,a four-hammer radial forging machine was used to carry out the hot radial forging experiments on 45 steel round cross-section specimens,and the data of core strain and depth of end surface were collected to verify finite element model.Secondly,two rational three-dimensional finite element models were established for flat hammers and arc hammers,respectively.Several geometrical parameters of hammers and dimensionless process parameters were selected for different models,respectively.Orthogonal experiments designed by Taguchi method were conducted to analyze the basic law of the forging load and the end surface of flat hammers and arc hammers.For radial forging process using arc hammers,a new process wass proposed with the bite ratio,adding a step of putting the forging into hammers and forging it in situ before the radial forging process.The results showed that the new process was conducive to reducing the depth of end surface.It was also found that under the conditions of similar process parameters(radial reduction rate,feed rate,etc.),the forging forged by the arc hammers had relatively small relative depth of end surface,relatively large core strain and better surface quality,but the forging load was larger.Finally,considering the finite element simulation is time-consuming and can't offer prediction and optimization for radial forging process in real time,the T-S fuzzy neural network which can predict the forging load and the depth of end surface is estiblished using the finite element simulation results,and then it put forward optimized plan for radial forging process.