基于容错理论无级变速器故障诊断分类  

Faults diagnosis and classification based on fault-tolerant theory for continuously variable transmission

作　　者：韩玲[1] 卢延辉[2] 安颖[3] 田丽媛[1] 

机构地区：[1]长春工业大学机电工程学院,吉林长春130021 [2]吉林大学汽车工程学院,吉林长春130022 [3]湖南大学汽车工程学院,湖南长沙410000

出　　处：《浙江大学学报（工学版）》2016年第10期1927-1936,共10页Journal of Zhejiang University：Engineering Science

基　　金：国家自然科学基金资助项目(50775094)

摘　　要：为了实现无级变速器(CVT)在线故障诊断及分类需求,基于改进人工蜂群算法结合最小二乘向量机,提取无级变速器故障特征向量,建立夹紧力与速比故障预测模型,按照预测模型进行故障定位、分类辨识.针对从动油缸压力、速比、执行机构等故障的特征向量进行Matlab仿真,分析经容错控制分类后的故障特征向量仿真曲线变化趋势.通过容错控制在Matlab/Simulink中建立无级变速器硬件在环仿真动态模型.结果说明,采用该方法,无级变速器故障诊断经容错可继续工作概率达98.728%,满足无级变速器在线故障诊断可靠性及稳定性的需求.The fault vectors were extracted by feature extraction of continuously variable transmission(CVT)based on the improved artificial colony algorithm combining with least squares support vector machine in order to fulfill the requirements about on-line fault diagnosis and classification of CVT.The fault predicting models of clamping force and ratio were established.The faults location and identification classification were based on predicting models.Parts of fault feature vectors for second cylinder pressure,ratio,and actuator were extracted to simulation by Matlab.The classification of feature vectors after the fault-tolerant controlled was analyzed by simulation curve change trend.CVT was built hardware-in-theloop simulation by fault-tolerant with Matlab/Simulink.Results showed that the probability of continue working reached 98.728% by using the method for fault-CVT,and the needs of CVT reliability and stability for the online fault diagnosis were meeted.

关 键 词：无级变速器 主动容错 改进人工蜂群算法 故障诊断 

分 类 号：U472[机械工程—车辆工程]