摘要
针对机械设备盲信号处理实践中,机械源信号数目大于多通道观测数目的欠定盲信号处理问题,将非高斯性的度量指标——负熵和经验模式分解引入机械设备的故障特征提取,提出基于负熵的旋转机械盲信号处理方法,从而解决欠定的单通道机械盲信号处理问题。首先将机械设备的单通道观测信号经验模式分解,得到本征模函数,然后计算本征模函数的负熵值,并将负熵值依序组成负熵基特征向量,最后采用最小二乘支持向量机进行机械设备的模式判别和故障诊断。液压齿轮泵的盲信号处理试验表明,该方法的故障识别率达到了93%以上,表明该方法是可以应用于机械盲信号处理实践的。
In blind signal processing of mechanical equipment,it is common that mechanical source number is more than that of multi-channel observation signals.This was called underdetermined blind signal processing.Then negentropy and empirical mode decomposition（EMD） was introduced into feature Abstraction of mechanical equipment to deal with their single channel blind signal processing.This algorithm was composed of three steps.Step 1 was to decompose single channel mechanical observation signals with empirical mode decomposition and get intrinsic mode functions（IMFs）.Step 2 was to compute negentropy values of IMFs and form a negentropy eigenvector according to IMFs sequence.Step 3 was to recognize different working patterns and diagnose different faults with LS-SVM.Its applications in blind signal processing of hydraulic gear pumps show that,its fault diagnosis rate is up to 93% on the whole.This indicates that this algorithm can be applied to mechanical blind signal processing.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2011年第10期1193-1197,共5页
China Mechanical Engineering
基金
总装备部预研重点基金资助项目
关键词
盲信号处理
负熵
经验模式分解
最小二乘支持向量机
blind signal processing
negentropy
empirical mode decomposition
least square support vector machine（LS-SVM）
