摘要
针对齿轮故障难提取和极限学习机(extreme learning machine,ELM)隐层节点数需要人为设定,致使齿轮故障分类模型准确度低、稳定性差的问题,提出基于核极限学习机(kernel extreme learning machine,K-ELM)的齿轮故障诊断方法。首先,将测得信号经经验模态分解(empirical mode decomposition,EMD)处理后得到一系列IMF本征模式分量,并提取各分量的排列熵(permutation entropy,PE)值组成高维特征向量集;然后利用高斯核函数的内积表达ELM输出函数,从而自适应确定隐层节点数;最后,将所得高维特征向量集作为K-ELM算法的输入建立核函数极限学习机齿轮故障分类模型,进行齿轮不同故障状态的分类辨识。实验结果表明:与SVM、ELM故障分类模型相比,核函数ELM滚动齿轮故障诊断分类模型具有更高的准确度和稳定性。
Due to the low accuracy and poor stability of gear fault classification model because of the hard extraction of gear fault and artificial settings for the number of hidden layer nodes of extreme learning machine （ELM）, a gear fault diagnosis method based on kernel extreme learning machine （K-ELM） is proposed. First, a series of IMF intrinsic mode component can be obtained after empirical mode decomposition （EMD） for measured signal,and permutation entropy value （PE） of various components should be extracted to form a vector set with high-dimension features. Second, the inner product of Gauss kernel function should be used to express the ELM output function to adaptively determine the number of the hidden layer nodes. After that, the high dimension feature vector set is used as the input of the K-ELM algorithm to establish the kernel function extreme learning machine gear fault classification model to achieve the classification and identification under different fault states of gears. The experimental results show that the K-ELM gear fault diagnosis classification model has higher accuracy and stability by comparing with the fault classification model of SVM and ELM.
出处
《中国测试》
北大核心
2017年第7期108-111,144,共5页
China Measurement & Test
基金
国家自然科学基金(51565046)
内蒙古自然科学基金(2015MS0512)
内蒙古科技大学创新基金(2015QDL12)
关键词
齿轮
故障诊断
排列熵
核函数
极限学习机
gear
fault diagnosis
permutation entropy
kernel function
extreme learning machine
