基于神经网络的转向节热锻成形工艺优化

Hot Forging Forming Process Optimization of Automobile Steering Knuckle Based on Neural Network

建立了输入层为始锻温度、终锻温度、模具预热温度、锻压速度,输出层为抗拉强度、成形载荷、磨损量的广义回归神经网络(GNRR)模型,基于此模型对汽车转向节热锻成形工艺进行优化。结果表明:使用GRNN神经网络工艺参数优化的汽车转向节锻件,其抗拉强度由优化前的774 MPa提升到786 MPa,成形载荷、磨损量分别由优化前的25.6 MN、120μm降低到23.5MN、115μm,抗拉强度的提升率为1.55%,成形载荷、磨损量的降低率分别为8.2%、4.2%;最佳热锻工艺参数为始锻温度为1260℃、终锻温度为1140℃、模具预热温度为230℃、锻压速度为48 mm/s。

The GNRR neural network model with initial forging temperature, final forging temperature, die preheating temperature, forging speed as the input layer, and tensile strength, forming load and wear amount as the output layer was established. Based on this model, the hot forging process of the automobile steering knuckle was optimized. The results show that when the automotive steering knuckle forgings are optimized by using GRNN neural network process parameters, their tensile strength is increased from 774 MPa to 786 MPa, and the forming load and wear volume are reduced from 25.6MN and120 μm to 23.5MN and 115 μm, respectively, the improvement rate of tensile strength is 1.55%, and the reduction rate of forming load and wear amount is 8.2% and 4.2%, respectively. The optimal thermal forging process parameters are the initial forging temperature of 1260℃, the final forging temperature of 1140℃, the die preheating temperature of 230℃, and the forging speed of 48 mm/s.