Dangerous Rock Fall On High And Steep Slope Of Railway Identification And Location Technology

China’s complex terrain conditions,high degree of railway network coverage,its safe operation has always been the focus of the industry.Mountainous railway is an important part of railway in China.The intrusion limit of high and steep slope,including mountain collapse,is an important factor threatening the safety of mountain railway.At present,the common monitoring methods of dangerous rock fall in mountainous railway include video monitoring method and laser radar monitoring method,which can complete most of the monitoring tasks,but they also have shortcomings such as poor all-weather performance or high cost.Microseismic monitoring rules are commonly used in mine monitoring,focusing on the classification of mine vibration and source location,and the positioning accuracy is poor.In view of the above problems,this paper proposes a microseismic monitoring method which is suitable for the identification and location of dangerous rock and rock fall transgression limit of railway in mountainous area.In order to reduce the false alarm rate,the interference signals were filtered out first,and then the vibration events of mountain railway were divided into four categories: single rock fall vibration,rock fall bouncing vibration,multiple rock fall vibration and collapse vibration.The vibration events were classified and recognized by feature extraction and SVM.In order to reduce the amount of calculation and improve the response speed,a two-level classification and recognition model framework is constructed.The second is the classification and recognition model of rockfall and collapse vibration(Model 2).Model 1 takes [vibration number,duration of first vibration,total duration of vibration,duration of stationary phase,attenuation coefficient] as the basis of classification and recognition.Model 2 takes [signal vibration number,total duration of vibration,total energy after normalization,maximum value of energy spectrum and sub-maximum value of energy spectrum] as the basis of classification and recognition.The experimental data show that Model 1can recognize four kinds of seismic events with 88.11% accuracy,and the errors mainly focus on two kinds of seismic events: multiple rockfall and collapse.With the assistance of Model 2,the comprehensive recognition accuracy can be improved to99.12% to further identify the above two kinds of events.On the basis of the completion of classification and identification,in order to improve the positioning accuracy of falling rocks,according to the initial position of falling rocks,the falling rock vibration is subdivided into rail vibration,sleeper vibration and Dao Zha vibration.In this paper,a time difference location fingerprint matching localization method is proposed,that is,the time difference location fingerprint database is established offline,and the real-time matching location is established online,so as to quickly and accurately complete the positioning of seismic events.This method can ignore the complexity of the transmission medium of seismic wave,and does not need to calculate the velocity parameters in advance,so the positioning accuracy is high.At the same time,the fingerprint matching algorithm has a small amount of computation and fast response.The results show that the absolute positioning error of this method can be controlled within the range of 0.8m.And the response speed is extremely fast,the average time to complete a matching location is0.019 8 s.The microseismic monitoring method proposed in this paper is applied to the section K141+150-350 of Shuohuang Railway,which is prone to dangerous rocks and rockfalls.The test results show that the method has good performance.