基于BP神经网络的磨料水射流铣削质量研究  

Study on Abrasive Water Jet Milling Quality by the BP Neural Network

作　　者：刘丽[1] 崔宣[1] 万庆丰[1] 

机构地区：[1]西华大学机械工程与自动化学院,成都610039

出　　处：《组合机床与自动化加工技术》2014年第5期130-132,共3页Modular Machine Tool & Automatic Manufacturing Technique

摘　　要：为提高磨料水射流铣削加工的表面质量,需深入研究磨料水射流工艺因素对加工质量的影响,以获得较好的表面质量.以射流压力、横移速度、磨料流量、靶距、横向进给量和工件表面粗糙度为研究对象,进行磨料水射流铣削加工试验.基于人工神经网络理论,建立磨料水射流铣削加工表面粗糙度的BP神经网络模型.借助大量试验数据,进行网络训练与验证.研究结果显示,用BP神经网络计算得到的横移速度来铣削工件,获得的工件表面粗糙度与期望值之间的相对误差为0.31％～3.09％,网络模型预测精度较高.In order to obtain a high surface quality of abrasive water jet milling, study on the influence of abrasive water jet process factors on machining quality were necessary. With the research of water jet pressure, traverse speed, abrasive flow rate, standoff distance and traverse feed, the abrasive water jet milling experiment was carded out. Based on artificial neural network theory, the BP neural network model of abrasive water jet milling surface roughness was established. Sample datas obtained through experiments, and applied to train and verify the model. The results indicated that the traverse speed which used for milling workpiece were calculated with the BP neural network, the relative error between the expected values and experimental measurements is from 0.31% to 3.09%, the accuracy of BP neural network is higher.

关 键 词：磨料水射流 BP神经网络 铣削加工 表面粗糙度 

分 类 号：TH16[机械工程—机械制造及自动化] TG65[金属学及工艺—金属切削加工及机床]