基于神经网络和高阶统计量的滚动轴承故障分类

提出一种基于高阶统计量特征和BP神经网络相结合的滚动轴承故障分类方法。以滚动轴承的高阶统计量(双谱、三阶累积量)以及一些常见的无量纲指标作为轴承故障特征输入,以BP神经网络作为分类器,成功地对滚动轴承4种不同的故障进行了分类。对比RBF神经网络,尽管BP神经网络的训练速度不快,但分类效果良好。研究表明,高阶统计量和BP神经网络相结合的滚动轴承分类方法是有效的。

Bearing Fault Classifications Based on Nerve Network and High Order Statistics

The fault classification method for the rolling bearings is put forwards,based on combination of BP nervous networks with higher-order statistics.Taking the higher order statistics(such as bispectrum and three order cumulant)and some common dimensionless index as input values of bearing fault characteristics,the BP nervous networks as segregator,the classification of four different faults is successfully completed.The classification effect is better by comparison with RBF nervous networks,though the training speed of BP nervous networks is not so fast.The results show that it is effective to use the combination of higher-order statistics with BP nervous networks for classification of rolling bearing faults.