为了使信号在时频域内提高聚集性和减少相干项,重排方法在处理非平稳信号中得到广泛应用。然而在传统重排处理后仍存在一些相干项和低信噪比的问题。采用一种基于非参数Snakes的时频重排方法,用于对钢绞线中超声导波信号中的模态识别。结果表明,相对于短时傅里叶变换和传统重排方法,基于非参数Snakes的时频重排方法得到的谱图具有更高的聚集性,且有效提高了信噪比。并且与外围钢丝的群速度频散曲线对比显示,利用该方法得到时频谱图能更准确地反映出主要超声导波模态的信息和各模态群速度变化趋势。因此,利用基于非参数Snakes的时频重排方法,可以为提高不同波导结构中超声导波模态的识别能力奠定重要基础。
The reassignment method is a widespread approach in processing nonstationary signals for obtaining high aggregation and less overlapping components in time-frequency representations.Unfortunately,by using this traditional method,there exist some problems such as unavoidable overlapping components and low signal-to-noise ratio values.A novel time-frequency reassignment method based on nonparametric snakes is proposed for guided wave mode identification of steel strands.According to research results,it is clarified the proposed reassignment method based on nonparametric snakes can obtain high resolution time-frequency representations with significantly higher aggregation and improved signal-to-noise ratio compared with the short-time Fourier transform and traditional reassignment method.Furthermore,compared with group velocity dispersion curves in one of helical wires of the steel strand,the spectrogram provided by the proposed method can give the information of main ultrasonic guided wave modes and describe the trend of group velocity changes of these modes more accurately.Therefore,the proposed time-frequency reassignment method can provide a valuable basis for improving ultrasonic guided wave identification in different waveguides.