【摘要】 针对滚动轴承振动监测信号的非平稳、非线性、非高斯等复杂特点,提出一种基于排列熵和集成支持向量机的退化状态评估方法。通过自适应噪声完备集合经验模态分解算法分解得到振动信号的本征模态函数,再以重构相空间分析本征模态函数的排序模式、提取排列熵作为滚动轴承状态特征,最后利用集成支持向量机来实现不同退化状态的智能评估。滚动轴承正常、内圈和滚动体不同退化程度下的实验数据分析结果表明,与样本熵特征、支持向量机模型相比,基于排列熵的集成支持向量机获得了更高的评估准确率,该文方法可以有效用于滚动轴承的退化评估。更多还原

【Abstract】 Aiming at the intricate non-stationary, non-linear and non-Gaussian characteristics of the vibration monitoring signal for rolling element bearings, a degradation status assessment method based on permutation entropy and ensemble support vector machine is proposed in this paper. The vibration signal was firstly decomposed by the complete ensemble empirical mode decomposition with adaptive noise to obtain intrinsic mode functions, then the order patterns of intrinsic mode functions were analyzed in the reconstructed phase space to extract permutation entropy as the condition features for rolling element bearings, and finally the ensemble support vector machine was utilized for intelligent evaluation of different degradation status. The experimental results of rolling element bearings under normal condition and different degrading severities with inner raceway and rolling element validates that, in comparison with the sample entropy feature and the support vector machine model, the proposed method achieves higher assessment accuracy. Thus the proposed method can be effectively utilized to evaluate degradation status of rolling element bearings.更多还原