基于神经网络的电火花加工工艺选择模型研究  

Study of a Process Selection Model for EDM Based on Neural Network

作　　者：彭泽军[1] 王宝瑞[1] 陈辉[1] 

机构地区：[1]中国工程物理研究院,绵阳621900

出　　处：《机械科学与技术》2006年第4期394-397,共4页Mechanical Science and Technology for Aerospace Engineering

基　　金：中国工程物理研究院自然科学基金项目(20040322);双百人才基金项目(2005R0305)资助

摘　　要：针对电火花加工工艺数据的特点,研究了神经网络输入输出数据的预处理方法,提出了基于对数变换的数据预处理算法,测试表明效果较优。分析了电火花加工工艺的特点及其复杂性,提出了基于神经网络的电火花加工工艺选择模型,该模型能模拟熟练操作者的决策过程。测试结果表明,该模型能真实反映机床本身的工艺特点,模型值和实测值相差较小,能实现在给定加工要求下电加工参数的自动选择。In light of the characteristics of electric discharge machining （EDM） process data, the paper studies a neural network data input and output pretreatment method and presents data pretreatment algorithm based on logarithm transformation. Test results show that the data pretreatment method is effective. By analyzing traits and complexities of the EDM process, its selection model based on neural network is introduced. The model can be used for simulating decision-making processes of skillful operators and show the true characteristics of processes of machining tools. Errors between model values and experiment values are trivial. It can accomplish the auto-selection of electrical machining parameters according to a given machining requirement.

关 键 词：电火花加工 工艺选择模型 神经网络 数据预处理 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]