中文摘要：

单一的差分振子仅可实现对周期信号中某一频率成分进行检测，对于强噪声背景下的边频带，尽管可以利用多个差分振子组成差分阵列进行逐个检测，进而确定边频带的间隔，但这种做法无疑会带来巨大的计算量。在对调制信号进行Hilbert变换包络分析时，所得到的时域信号是原始调制信号中的低频分量，亦是调制波信号，若该低频分量仍然包含较强的噪声成分，传统的频谱分析将会失效。此时，我们可借助差分振子时间历程对含较强的噪声的包络进行检测。因此，提出基于差分振子时间历程的微弱调制信号检测方法，即首先对信号进行Hilbert包络解调，然后利用差分振子时间历程对含较强的噪声的幅值包络进行检测，并成功应用于风机早期故障检测中。

英文摘要：

Single difference oscillator can only detect the component with a certain frequency in periodic signas. In order to determine the interval of the side frequency-band of the signal in strong noise background, it is feasible to use the multiple difference oscillator arrays to detect the frequencies one by one. But this method is computer-time consuming and costly. Hilbert transform based on modulation signal envelope analysis can be used to obtain the low frequency component of the original modulation signal. But, if the low frequency component contains strong noise, the traditional spectrum analysis in this method will fail. To solve the problem, the time course of difference oscillator can be used to test the gained envelope signal which contains the strong noise. In this paper, a new weak modulation signal detection method based on time course of difference oscillator was proposed. First of all, Hilbert transform was used in the envelope demodulation to gain the envelope component of the signal. Then, envelope demodulation test on the gained component, which contains the strong noise, was done using the method of difference oscillator time course. The method was successfully applied in the early fan fault detection.