变转速变载荷工况下的齿轮智能故障诊断北大核心

针对变转速变载荷工况下的齿轮故障检测、识别和分类问题,提出一种基于最大重叠离散小波包变换和人工神经网络的智能故障诊断新方法。研究自相关谱峭度图中的最大重叠离散小波包变换,并采用它将复杂的齿轮故障振动信号分解为频带和称为节点的中心频率。推导出每个节点的平方包络的自相关,以便计算每个节点在每个分解层次上的峭度,减少了非周期性脉冲和噪声的影响。将上一步得到的特征矩阵作为径向基函数神经网络的输入,从而实现齿轮状态的自动分类。并在变转速变载荷(16种)工况下对健康状态和5种不同类型齿轮故障的齿轮箱进行了具体测试分析。结果表明:该方法可以更好地提取特征信息,为齿轮故障诊断定位合适的解调频带,提高了所有工况下齿轮故障诊断的准确率。

Intelligent Fault Diagnosis of Gears under Variable Speed and Variable Load Conditions

Aiming at the problem of gear fault detection, identification and classification under variable speed and variable load conditions, a new intelligent fault diagnosis method based on maximum overlapping discrete wavelet packet transform and artificial neural network was proposed. The maximum overlapping discrete wavelet packet transform in the autocorrelation spectral kurtosis graph was studied, and the complex gear fault vibration signal was decomposed into frequency bands and central frequencies called nodes. Then, the autocorrelation of the square envelope of each node was derived to calculate the kurtosis of each node at each decomposition level, thus the influence of aperiodic pulse and noise was reduced. The feature matrix obtained in the previous step was used as the input of radial basis function neural network, so as to realize the automatic classification of gear status. At last, the gearboxes with healthy state and five different types of gear faults were tested and analyzed under variable speed and variable load (16 kinds) conditions. The results show that this method can be used to extract the feature information better, locate the appropriate demodulation frequency band for gear fault diagnosis, and the accuracy of gear fault diagnosis under all working conditions is improved.