摘要
针对采煤机电机由于工作面环境恶劣且在各种激励源下产生的振动信号相互叠加耦合使得故障信号噪声大并且难以提取特征等问题,基于多元集合经验模态分解(MEEMD)和小波联合去噪算法,对截割电机的4种故障信号进行MEEMD分解;通过判定峭度,将峭度高的分量进行小波阈值去噪,将去噪分量和相关系数较高的相关分量进行重构,获取降噪信号;然后,基于变分模态分解(VMD)进行故障信号特征提取,建立基于长短时记忆(LSTM)的模型网络;对提取的VMD能量熵特征进行训练,完成故障分类训练。得到的故障分类准确性较高,实现了复杂工况下的采煤机电机故障诊断。
Aiming at the problem that the fault signal noise is large and it is difficult to extract features due to the harsh environment of the working face and the superposition and coupling of the vibration signals generated by various excitation sources,based on the joint denoising algorithm of multivariate ensemble empirical mode decomposition(MEEMD)and wavelet,decomposed the four fault signals of the cutting motor by MEEMD.Denoised the wavelet threshold of the component with high kurtosis by judging the kurtosis,and reconstructed the denoising component and the correlation component with high correlation coefficient to obtain the noise reduction signal.Then,the fault signal feature extraction was carried out based on variational mode decomposition(VMD),the model network based on long short-term memor(LSTM)was established.The extracted VMD energy entropy feature was trained,and the fault classification training was completed.The fault classification accuracy obtained was high,and the fault diagnosis of the shearer motor under complex working conditions was realized.
作者
吴昊阳
欧阳敏
赵江滨
Wu Haoyang;Ouyang Min;Zhao Jiangbin(CCTEG Coal Mining Research Institute Co.,Ltd.,Beijing 100013,China;Tiandi Science and Technology Co.,Ltd.,Beijing 100013,China;Xi’an University ofScience and Technology,Xi’an 710054,China)
出处
《煤矿机械》
2025年第9期186-190,共5页
Coal Mine Machinery
基金
中国博士后科学基金第71批面上资助“地区专项支持计划”项目(2022MD713793)。
关键词
信号去噪
特征提取
故障诊断
循环神经网络
signal denoising
feature extraction
fault diagnosis
recurrent neural network
