诊断重载铁路钢轨波磨WPD-ASSTFT时-频边际谱方法

WPD-ASSTFT Time-frequency Marginal Spectrum Method for Diagnosing Track Corrugation of Heavy Haul Railway

重载铁路钢轨波磨会加剧轮轨间振动,导致轨道-车辆系统零部件损坏,危及运行安全性,在实际养护维修中需及时发现波磨区段并进行打磨。提出一种基于轴箱加速度的WPD-ASSTFT时-频边际谱方法用以诊断重载铁路钢轨波磨。首先,对轴箱垂向加速度数据进行小波包分解得到若干个子信号,对各个子信号做自适应同步压缩短时傅里叶变换,叠加各子信号时频分布获得原始信号的时频分布。然后根据波磨的周期性特征对由WPD-ASSTFT得到的时频分布进行分频带滤波,计算时-频边际谱。最后,根据时-频边际谱分别计算波磨强度因子和能量因子,当两者都超过阈值时判定区段存在波磨,并提取波磨波长信息。用该方法对纯波磨区段和伴有焊接接头不平顺的波磨区段进行诊断,其结果与现场复核结果高度一致,证明用该方法能有效诊断重载铁路波磨。并对同线路不同月份检测数据对比分析,证明了该方法具有较好稳定性。

Corrugation of heavy haul railway will aggravate the vibration between wheel and rail,cause damage to the track-vehicle system components,and affect operation safety.Timely detection and grinding of corrugation is required during maintenance.A WPD-ASSTFT time-frequency marginal spectrum method based on axle box acceleration was proposed to diagnose rail corrugation of heavy haul railway.Firstly,several sub signals were obtained by performing wavelet packet decomposition(WPD)on the axle box vertical acceleration data,and adaptive synchro-squeezed short time Fourier transform(ASSTFT)was carried out on each sub signal.Time frequency distribution(TFD)of the original signal was obtained by superimposing each sub signal’s TFD.Then,the TFD obtained by WPD-ASSTFT was filtered by frequency division band according to the periodic characteristics of corrugation,to calculate the time-frequency marginal spectrum.Finally,the corrugation intensity factor(CIF)and energy factor(EF)were calculated respectively according to time-frequency marginal spectrum.When both exceeded the threshold,corrugation was determined to exist in the section,and the information on the wavelength of corrugation was extracted.When this method was used to diagnose pure corrugation section and corrugation section with welded joint irregularity,the results were highly consistent with the results of field recheck,which proves the effectiveness of this method in diagnosing the corrugation of heavy haul railway.The comparative analysis of the detection data of the same line in different months proves good stability of the method.