一种基于改进的傅里叶变换的弱故障信号特征提取新方法研究

New feature extraction method for weak fault signal based on improved fourier transform

弱信号特征提取一直是故障诊断领域的难点,文章结合传统傅里叶变换,提出一种将时间序列变换为频域,再从频域转换到时域的复数域,并对该复数域进行幅值计算的方法对弱信号进行特征提取。通过仿真计算发现该方法突出了信号的局部特征信息,不仅能对夹杂在信号中的微弱冲击成分进行较好的再现,而且也能在强背景噪声环境下提取微弱故障信息。最后通过齿轮齿面接触型故障实验验证了该方法的有效性。

In view of the traditional difficulty of feature extraction for weak signal in the realm of fault diagnosis, a new method was presented to calculate the complex magnitude and then to extract the feature of weak signal, in which the original time series was transfered to frequency domain, and then to complex number domain, based on Fourier transform. The algorithm gives prominence to the local characteristic information of signal. In the simulation of an example, the weak impulsive components in mixed-signal can well reappear and the information of weak fault signal can be drawn efficiently under the condition of powerful background noise. The gear fault diagnosis experiment states the effectiveness of the method proposed.