中文摘要：

提出了一种基于移频技术的短时傅里叶变换阶比分析算法.该算法利用傅里叶变换在频域的卷积性质,对原始信号在时域乘以e-j2πfit使fi的频谱能量搬迁到零频处,按一定的频率间隔改变fi就可以在零频处得到其他频率的频谱能量,以此来提高短时傅里叶变换在时频分析中的频率分辨率.然后在时频面上进行局部阈值降噪,同时跟踪转速的变化,最终应用到变速机械的阶比分析中.与短时傅里叶变换分析结果对比表明,本文方法可以更加准确地跟踪到实际的转速.实际降速过程中轴承信号利用本文方法进行阶比分析,成功提取到轴承的故障特征频率.

英文摘要：

An order analysis method was proposed which utilizes short time Fourier transforms（STFT） based on frequency shift（FS-STFT）.In the algorithm,by using the convolution properties of Fourier transforms in the frequency domain,the original signal is multiplied by e-j2πfit in the time domain,and the spectrum energy of fi is shifted to the position of zero frequency.Other spectrum energies can be got at the position of zero frequency when fi is changed according to a certain frequency interval.As a result,the frequency resolution of STFT is improved in the time-frequency analysis.Then local threshold de-noising was used in the time-frequency domain,and the change of speed signals was tracked.Eventually,the mothod was applied to the order analysis of varying speed machinery.A comparison of analysis results between STFT and FS-STFT shows that the method can track actual speed signals accurately.The order analysis for faulty bearings using the method in the deceleration process also demonstrates that the fault frequency can be extracted successfully.