Research And Application Of Rail Shortwave Matching Algorithm

The short waveform of rail surface as one kind of rail irregularity,which is an important factor to affect the train operation safety.So we'd be better to find the way to detect the short waveform irregularity of railway rapidly and efficiently to ensure the high-speed train moving quickly,safely and efficiently.And that is a great challenge for our country to develop high-speed railway and subway etc.The short waveform of rail irregularity detection can be divided into two major categories: dynamic and static.Using the continuous measuring trolley,such as BMY?RMF ? CAT etc,dynamic measurement can obtain the continuous measurement waveform of rail shortwave irregularity.These measuring trolley can obtain the continuous measuring waveform of rail shortwave irregularity,but it can not get the information of local rail flatness defects,such as hardness and bending,polishing irregularity etc.Rail shortwave measuring instrument is abbreviated as flatness of trolley,which is widely used in static detection of short waveform irregularity.And besides,there are also plenty of instruments,such as SED-RC?GEISMAR etc.These static measurement can obtain the information of the rail shortwave irregularity of segmentation measuring waveform,which can be get by the limit of measuring device.However,the ability of static testing is limited,which can obtain the frequency domain features of rail short wave irregularity continuously.In this article,the measuring data of rail shortwave irregularity can be integrated and matching by sectional and multiple measurement.In this way,a new measurement method can be created,which can be show time-domain and frequency domain characteristics of rail shortwave irregularity objectively.Therefore,Dynamic Time Warping which is abbreviated DTW,is applied in the field of pattern recognition,such as speech recognition.The algorithm of DTW being introduced in the field of rail short waveform irregularity matching,which can be solved the problem that the time axis can not be bent during waveform matching.First of all,waveform data being matched should be stored in time series.Using similarity search of time series,the optimum region in short waveform matching can be determined.Secondly,the experimental platform of rail shortwave irregularity measurement can be established.And besides,the overall structure of the measurement system,data acquisition andapplication software are designed.After the measurement system has been built,we can test and analyze the measurement accuracy and stability of experimental platform to ensure the accuracy and reliability of it.According to the measurement principle of platform,the measured data are not in the same coordinate system.We make use of two dimensional coordinates transformation formula to enable the being matched curves to be rotated and translated.In this way the being matched curves can be united in the same coordination.Taking this basic concept,we put forward a new matching algorithm based on DTW in the rail shortwave matching.Making the time axis to be bended and deformed,we are able to find the similar waveform and set up the cumulative cost matrix.And besides,using the recursive thinking we can calculate the shortest bending path between the two curves.The process of calculating the shortest bending path is essentially the process of matching the two time-series.However,the two time-series exist two kinds of states after matching.The one is that point-to-point matching has been achieved,and the other is a multi-point matching state.According to the multi-point matching,quadratic fit method is put forward.And then,we can establish point-to-point correspondence among the being matched curves to ensure the accuracy of the point-to-point correspondence.On the basis of the value of DTW,the matching effect can be evaluated.In this paper,we contrast the traditional algorithm based on the region with dynamic time warping.The experimental results show that we can make the short wavelength dynamic matching due to the dynamic time warping algorithm.And besides,the stitching precision of using the dynamic time warping is increased by 10 times than the traditional algorithm.What's more,the dynamic time warping has the advantages of fewer stitching errors and better stability.