The successful prediction of the remaining useful life of rolling element bearings depends on the capability of early fault detection. A critical step in fault diagnosis is to use the correct signal processing techniques to extract the fault signal. This paper proposes a newly developed diagnostic model using a sparse-based empirical wavelet transform (EWT) to enhance the fault signal to noise ratio. The unprocessed signal is first analyzed using the kurtogram to locate the fault frequency band and filter out the system noise. Then, the preproc signal is filtered using the EWT. The l q -regularized sparse regression is implemented to obtain a sparse solution of the defect signal in the frequency domain. The proposed method demonstrates a significant improvement of the signal to noise ratio and is applicable for detection of cyclic fault, which includes the extraction of the fault signatures of bearings and gearboxes.

中文翻译：

---

使用稀疏经验小波变换对循环故障进行弱故障检测。

滚动轴承的剩余使用寿命的成功预测取决于早期故障检测的能力。故障诊断中的关键步骤是使用正确的信号处理技术来提取故障信号。本文提出了一种新的诊断模型，该模型使用基于稀疏的经验小波变换（EWT）来提高故障信噪比。首先使用曲线图分析未处理的信号，以定位故障频带并滤除系统噪声。然后，使用EWT对前置信号进行滤波。实现lq正则化的稀疏回归以获得频域中缺陷信号的稀疏解。该方法证明了信噪比的显着改善，适用于循环故障的检测，