摘要
为了提高破碎机筒体容易产生多种非线性振动信号扛干扰能力,设计了一种基于多元经验模态分解(MEMD)和样本熵(SE)算法的破碎机负荷振动信号识别方法。采用MEMD算法分解噪声信号,计算不同IMF分量相关程度及有效IMF分量SE,完成破碎机负荷精确判断。研究结果表明:在初始波形中形成明显噪声,采用MEMD算法几乎所有高频噪声信号都被去除,获得的重构信号幅度跟总体趋势相近,具有优异的球破碎机振动信号预处理效果。提高相似容限时形成了明显的重叠现象,不同负荷状态获得了更高相似度。当设定25个隐含层神经元并且选择sin函数作为激活函数时识别模型达到了最高准确率,可以准确识别过负荷特征,明显高于EMD-样本熵和MEEMD-样本熵对于过负荷的识别准确率。
In order to improve the anti⁃interferenceability of the crusher cylinder to generate easily a variety of nonlinear vibration signals with an interference,a method based on MEMD and SE algorithm is designed to identify the crusher load vibration signals.The MEMD algorithm is used to decompose the noise signal,calculate the correlation degree of different IMF components and the effective IMF component SE,and complete the accurate judgment of the crusher load.The results show that the obvious noise is formed in the initial waveform,almost all high⁃frequency noise signals are removed by the MEMD algorithm,and the reconstructed signal amplitude is similar to the general trend,which has excellent effect on the pre⁃processing of ball mill vibration signals.When the similarity tolerance is increased,the obvious overlap is formed,and the higher similarity is obtained for different load states.When 25 hidden layers of neurons are set and the sine function is selected as the activation function,the identification model reaches the highest accuracy,which can identify overload characteristics with significantly higher accuracy than EMD⁃sample entropy and MEEMD⁃sample entropy methods.
作者
周喜
李道军
王会珍
李峰
ZHOU Xi;LI Daojun;WANG Huizhen;LI Feng(College of Automation and Internet of Things,Zhengzhou Polytechnic College,Zhengzhou 450100,China;School of Mechanical Engineering,Henan Polytechnic University,Jiaozuo Henan 453000,China)
出处
《机械设计与研究》
CSCD
北大核心
2024年第1期161-165,共5页
Machine Design And Research
基金
河南省高等学校重点科研项目(22B460032)。
关键词
振动信号
破碎机
状态识别
多元经验模态分解
样本熵
vibration signal
crusher
status recognition
multivariate empirical mode decomposition
sample entropy
