基于LMD近似熵和PNN的轴承故障诊断
发布时间：2018-11-25 10:12
【摘要】：提出一种基于局部均值分解(LMD)近似熵和概率神经网络(PNN)的滚动轴承故障诊断方法。通过对信号LMD分解,非平稳信号能够转换成若干个平稳的乘积函数分量(PF)之和;轴承在发生不同故障时,产生频谱相异的信号,其近似熵不同,因此可通过提取原始信号的近似熵,来判别轴承的运行状态。实验表明,信号经过LMD分解得到若干PF分量,从中提取近似熵,组成N维特征向量,输入PNN模型,能够准确地判断故障类型;在小数据的情况下,相比于BP和RBF两种传统神经网络,PNN具有更优的故障分类能力。
[Abstract]:A fault diagnosis method for rolling bearings based on local mean decomposition (LMD) approximation entropy and probabilistic neural network (PNN) is proposed. By decomposing the signal LMD, the non-stationary signal can be converted into the sum of several stationary product function components (PF). When different faults occur in the bearing, the signal with different spectrum is produced, and its approximate entropy is different. Therefore, the operating state of the bearing can be judged by extracting the approximate entropy of the original signal. The experimental results show that the signal is decomposed by LMD to obtain some PF components, from which the approximate entropy is extracted, the N-dimensional eigenvector is formed and the PNN model is inputted, which can accurately judge the fault type. In the case of small data, compared with two traditional neural networks, BP and RBF, PNN has better fault classification ability.
【作者单位】： 中北大学机械与动力工程学院;
【分类号】：TH133.3
本文编号：2355728
[Abstract]:A fault diagnosis method for rolling bearings based on local mean decomposition (LMD) approximation entropy and probabilistic neural network (PNN) is proposed. By decomposing the signal LMD, the non-stationary signal can be converted into the sum of several stationary product function components (PF). When different faults occur in the bearing, the signal with different spectrum is produced, and its approximate entropy is different. Therefore, the operating state of the bearing can be judged by extracting the approximate entropy of the original signal. The experimental results show that the signal is decomposed by LMD to obtain some PF components, from which the approximate entropy is extracted, the N-dimensional eigenvector is formed and the PNN model is inputted, which can accurately judge the fault type. In the case of small data, compared with two traditional neural networks, BP and RBF, PNN has better fault classification ability.
【作者单位】： 中北大学机械与动力工程学院;
【分类号】：TH133.3
【相似文献】
相关期刊论文 前5条
1 徐光华;赵国强;印建安;姜阔胜;张虎;陶唐飞;;基于位移信号和近似熵的轴承加工质量动态检测技术[J];工业仪表与自动化装置;2014年03期
2 焦海锋;;基于小波—近似熵在风机故障诊断中的应用研究[J];电站系统工程;2007年04期
3 吕勇;徐金梧;李友荣;;递归图和近似熵在设备故障信号复杂度分析中的应用[J];机械强度;2006年03期
4 戴桂平;;基于EMD近似熵和LS-SVM的机械故障智能诊断[J];机械强度;2011年02期
5 ;[J];;年期
，本文编号：2355728
本文链接：https://www.wllwen.com/kejilunwen/jixiegongcheng/2355728.html
