中文摘要：

针对轴承故障振动信号的非平稳特征和本征时间尺度分解（intrinsictime—SCaledecomposition。ITD）方法的缺点，提出了基于三次多项式的本征时间尺度分解方法（cubicpolynomial—basedintrinsictime—scaledecomposition，CITD）和同态滤波的解调方法。首先采用CITD方法对轴承振动信号进行分解，将其分解为若干个合理旋转（properrotation，PR）分量之和，然后用相关系数筛选出最能表征故障信息的PR分量进行重构，最后对重构信号运用同态滤波解调来提取故障特征。仿真信号与轴承故障诊断工程实例的分析验证了该方法的有效性。

英文摘要：

Aiming at the nonstationary characteristics of a bearing fault vibration signal and the disadvantage trinsic Timescale Decomposition （ITD） method, an improved ITD method combining the Cubic polynomial of In based Intrinsic Timescale Decomposition （CITD） and the homomorphic filtering demodulation was proposed. In this ap proach, the CITD method was applied to decompose the vibration signal into a finite number of proper rotation com ponents, then the correlation coefficients were used to select the proper rotation（PR） components best representing the fault information and to reconstruct the signal. Finally, the fault characteristics was extracted from the recon structed signal by homomorphic filtering demodulation. The effectiveness of the present method was verified by anal ysis of simulation signals and engineering examples of bearing fault diagnosis.