摘要
针对强背景噪声下齿轮故障冲击特征提取问题,提出了一种基于自适应随机共振和稀疏编码收缩算法的齿轮故障诊断方法。该方法选用相关峭度作为随机共振检测周期性冲击分量的测度函数,借助遗传算法实现信号中周期性冲击特征的自适应提取;在此基础上,利用稀疏编码收缩算法对随机共振检测结果做进一步降噪处理,从而凸显冲击特征,提高故障识别精度。试验和工程实例分析结果表明,该方法可实现齿轮故障冲击特征的增强提取,为齿轮故障诊断提供依据。
Aiming at the impact feature extraction problem of gear faults under strong background noises, a gear fault diagnosis method was proposed based on adaptive stochastic resonance and sparse code shrinkage algorithm. In order to achieve the effective extraction of periodic impact features, cor- related kurtosis was adopted as the measurement index of stochastic resonance, which was used to construct the fitness function of genetic algorithm. According to the maximum of fitness function, the optimal stochastic resonance system parameters could be selected, so as to achieve the adaptive extrac- tion of the periodic impact features from the vibration signals submerged by strong noises. Then, sparse code shrinkage algorithm was applied to further reduce the noises from the detection results of stochastic resonance, thereby the impact features were highlighted to improve the identification accu- racy of gear faults. The experimental and engineering application results indicate that the proposed method can realize the enhancement and extraction of impact features of gear faults, which will pro- vide a basis for gear fault diagnosis.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2016年第13期1796-1801,1809,共7页
China Mechanical Engineering
基金
国家自然科学基金资助项目(51505415)
中国博士后科学基金资助项目(2015M571279)
秦皇岛市科技支撑计划资助项目(201502A008)
关键词
随机共振
相关峭度
稀疏编码收缩
冲击特征提取
stochastic resonance
correlated kurtosis
sparse code shrinkage
impact feature ex- traction