中文摘要：

针对快速谱峭度在低信噪比情况下分析效果差的问题,提出应用粒子滤波的前处理降噪方法来提高信噪比,从而解决谱峭度受噪声干扰效果差的问题,进而提高滚动轴承故障诊断的成功率。建立振动信号的状态方程,提取原始信号的背景噪声,将其与状态方程之和作为观测方程。联立状态方程与观测方程来建立状态空间模型。采用粒子滤波对信号重新估计,得到新序列即是降噪之后的信号,结合快速谱峭度方法,获取最佳分析频带。并结合频谱分析得出故障频率。对比快速谱峭度与经验模式分解（Empirical mode decomposition,EMD）方法降噪的谱峭度分析诊断结果,证明所提方法的有效性。

英文摘要：

Fast Kurtogram has a low performance for the signal with low signal-to-noise ratio（SNR）.Particle filter is used as preprocessing method to improve the SNR.It can be used to solve the problem about low performance and improve the fault diagnosis accuracy.State function of the vibration signal can be established.The background noise of origin signal can be extracted.State function and background noise are combined together as the observation function.State function together with observation function is used to construct the state space model.Particle filter is used to estimate new sequence for the noise-reduced signal.The optimal analysis band is obtained by using fast Kurtogram for the noise-reduced signal.Fault characteristic frequency can be obtained based on spectrum analysis.Comparison with fast Kurtogram and empirical mode decomposition filter as preprocessing method for fast Kurtogram,it can be concluded that the proposed method has good performance for rolling element bearing pattern recognition.