中文摘要：

在故障诊断中,针对当前各种时频分析方法存在的问题,作者提出一种基于自适应Chirplet变换的故障诊断方法.该方法在参数粗搜索的基础上,将多维参数的最优化过程转化为传统的曲线拟合问题,不仅解决了交叉干扰项和时频分辨率之间的冲突,而且还具有计算量小、运算速度快和参数估计精度高等优点.实验结果表明,该算法能够有效地提取故障轴承振动信号的时频特征,其诊断效果明显优于其他时频分析方法,因此,是一种有效的故障诊断方法.

英文摘要：

Aiming at the problems existed in all other TF（time-frequency） analysis methods in the fault diagnosis,a method based on the adaptive Chirplet transform was presented.Based on parameter coarse estimation,it convertsed multi-dimension optimization process to a traditional curve-fitting problem,not only solved the conflict between cross-term interference and TF resolution,but also had some merits,such as computational speed was fast and parameters estimation was more accurate.Experimental results showed that the algorithm could efficiently extract the TF characteristics of the fault signals,and the effect was better than other TF methods.Therefore,it was a efficient fault diagnosis method.