铁路轨检车轴箱振动加速度消趋方法的比较

振动测量数据中往往包含了趋势项，这些趋势项对数据的分析会产生很大的影响，尤其在积分过程中，这种影响会进一步放大，所以消除趋势项是数据分析的基本前提。对包含在轴箱加速度中的趋势项进行消除，并进行双积分以达到与轨面高低不平顺进行比较的目的，采用了三种消趋方法进行对比分析的方式，三种消除趋势项的方法（多项式拟合消趋，滤波消趋及小波消趋）对实测数据的分析结果为：多项式拟合消趋不适于轴箱加速度的积分分析，而滤波消趋与小波消趋能适于这种分析，进一步，从波形包络来看，小波消趋与原波形最接近。最后提出了适合于轴箱加速度分析的消趋方法及其相关参数值。

Comparison of Methods for Removing Trend in Axlebox Acceleration Measured by Railway Track Inspection Car

There is always a trend in measured vibration data which has great influence upon the analysis of the data. This kind of influence will further amplify while the data is being integrated. So, it is the basic premise of data analysis to eliminate the trend. The trend is removed in axlebox acceleration data which are then double integrated in order to compare with the track longitudinal level irregularities in this paper. Three methods for removing the trend are applied to analyzing and contrasting. The analysis result for the three methods for removing the trend （by polynomial fitting, by filter and by wavelets） shows that detrend by polynomial fitting is not suitable for the integral analysis of axlebox acceleration while detrend by filter or by wavelet is either suitable, furthermore, waves using wavelet to remove the trend are most close to the original ones according to the wave envelopments. Finally, Methods suitable for removing the trend in axlebox acceleration and the relative parameters are put forward.