利用置信规则库的数控机床伺服系统故障诊断  

Fault Diagnosis of CNC Servo System Based on Belief Rule Base

作　　者：张邦成[1] 尹晓静[1] 王占礼[1] 周志杰[2] 张玉玲 吕康文[4] 

机构地区：[1]长春工业大学机电工程学院 [2]第二炮兵工程大学控制工程系 [3]徐州重型机械有限公司 [4]中国人民解放军96219部队

出　　处：《振动．测试与诊断》2013年第4期694-700,729,共7页Journal of Vibration,Measurement & Diagnosis

基　　金：国家自然科学基金资助项目(61004069);吉林省教育厅"十一五"项目(2009109)

摘　　要：为了诊断数控机床伺服系统故障,针对数控机床伺服系统故障原因和征兆之间存在的复杂非线性关系,提出利用置信规则库(belief rule base,简称BRB)对伺服系统进行故障诊断。该方法能够利用各种类型的信息,建立输入和输出之间的非线性模型,通过优化模型对系统进行故障诊断。为了解决专家给定的初始BRB中参数不精确的问题,结合可以获得的伺服系统发生故障时的信息类型,基于数值观测值的优化学习模型对由专家给定的初始BRB系统中的参数进行训练,对数控铣床伺服系统中工作台的故障进行诊断。实验结果表明,利用该方法训练后的BRB系统能够准确反映系统的行为,可充分利用各种不确定信息,提高故障诊断精度。In order to diagnose the fault of the CNC machine tools servo system,a belief rule base(BRB)is proposed to diagnose the fault of the servo system for the complex nonlinear relationship between fault causes and symptoms of CNC machine tools servo system.This method can utilize various types of information to establish a nonlinear model between the input and output,and also can diagnose the fault system using the optimization model.In order to solve the problem of inaccurate parameters of the initial BRB given by experts,and based on the optimization learning model of the observation numerical value,the parameters of the initial BRB given by experts are trained by combining the types of information received from the servo system failure.Then,the workbench of CNC milling machine servo system can be diagnosed.The experimental results show that this method after training BRB system can accurately reflect the behavior of the system and take full use of uncertain information to improve the accuracy of fault diagnosis.

关 键 词：置信规则库 专家系统 故障诊断 数控机床 伺服系统 

分 类 号：TG659[金属学及工艺—金属切削加工及机床] TP277[自动化与计算机技术—检测技术与自动化装置]