电火花线切割工艺模型优化与加工参数优选的研究

Research on optimizing WEDM process model and processing parameters

针对高速走丝电火花线切割加工中的电参数选择问题,将加工电参数和工艺指标参数分别作为神经网络的输入和输出,利用神经网络建立电火花线切割工艺模型,使用遗传算法对神经网络模型的权值和阈值进行优化,以加快网络的训练速度和避免网络陷入局部最小值。结合训练好的工艺模型,利用遗传算法以神经网络的误差传递函数作为适应度函数,对电火花线切割加工参数进行多目标优化,以达到对加工参数的优化选取。

In view of the choice of electrical parameters during high-speed wire cut electrical discharge machining （WEDM）,the electrical processing parameters and process index parameters were taken as the input and output of neural network.Neural network was used to establish WEDM process model and the weights and thresholds of neural network model were optimized by genetic algorithm,which speeded up the network training velocity and kept the network from sinking into a local minimum.In combination with the trained process model and taking the error transfer function of neural network as fitness function,the WEDM processing parameters were multi-objectively optimized with genetic algorithm,so as to optimize and select processing parameters.