应用AR模型的多参数与多测点信息融合的故障分类
发布时间:2018-01-02 11:37
本文关键词:应用AR模型的多参数与多测点信息融合的故障分类 出处:《机械科学与技术》2017年06期 论文类型:期刊论文
更多相关文章: 行星齿轮传动系统 混沌特征参数 多测点 支持向量机 EMD AR模型
【摘要】:为了找到针对齿轮传动系统多类故障分类的有效方法,对行星齿轮传动系统进行故障实验,获取振动信号。采用EMD方法对该振动信号进行预处理,得到若干个IMF分量之和,对前4个有效的IMF分量分别建立AR模型,得到对应的自回归参数序列ф,进而对其分别计算关联维数、最大Lyapunov指数、样本熵这3个混沌特征参数,并将其作为辨识特征量。将不同测点对应的ф的不同混沌特征参数信息融合作为支持向量机的输入向量,建立6种不同故障状态的训练集,实现对故障类型进行分类。结果表明:对实验获取的振动信号进行EMD和AR模型处理后,能在很大程度上提高故障分类准确率。
[Abstract]:In order to find an effective method for the classification of many kinds of faults in gear transmission system, the fault experiment of planetary gear transmission system is carried out to obtain the vibration signal, and the vibration signal is preprocessed by EMD method. The sum of several IMF components is obtained. The AR model is established for the first four effective IMF components, and the corresponding autoregressive parameter sequences are obtained, and the correlation dimensions are calculated respectively. The maximum Lyapunov exponent and sample entropy are three chaotic characteristic parameters. The information of different chaotic characteristic parameters corresponding to different measuring points is fused as input vector of support vector machine, and six training sets of different fault states are established. The results show that the accuracy of fault classification can be improved to a great extent after the vibration signals obtained by experiments are processed by EMD and AR models.
【作者单位】: 天津工业大学机械工程学院;现代机电装备技术重点实验室;
【基金】:国家重大科技成果转化项目(2060403) 天津市自然科学基金重点项目(10JCZDJC23400)资助
【分类号】:TH132.41
【正文快照】: 在风力机的传动系统中行星齿轮组的作用极其重要。然而,在运行过程中行星齿轮中的各部件极易出现损坏[1],导致系统出现各种故障。行星齿轮的故障特征极其微弱,难以获取,但连带的故障特征非常明显。很难从时域和频域中获取有用的故障信息。因此,国内外学者针对以上难点开始寻找,
本文编号:1369031
本文链接:https://www.wllwen.com/jixiegongchenglunwen/1369031.html