Track Slab Crack Identification Based On Power Flow Theory

Ballastless track is widely used in China’s railway construction due to the advantages of high stability,but it is prone to damage to the track structure under the long-term operation of the train vehicle,and once the track slab is damaged,it will lead to changes in the structural stress system,accelerate deterioration,affect the stability of train operation,the durability of the structure,and then affect the safety,so it is of great significance to study the new technology of track plate crack identification.At present,power flow theory is widely used in machinery,aerospace and other fields,and the recognition effect is very good,so it is necessary to apply power flow theory to the damage identification of orbital structures.In this thesis,the power flow characteristics of track slabs with crack damage are analyzed for CRTSII.type ballastless tracks,the characterization parameters of power flow damage degree are proposed,the feasibility of identifying track slab cracks based on power flow theory is studied,and two sets of track slab crack damage identification methods are proposed for transient impact conditions and vehicle-induced vibration conditions,and their noise resistance and robust performance are analyzed.The main research contents are as follows:(1)Study on power flow characteristics of track slab with cracks.On the theoretical basis,the feasibility of power flow theory to identify track slab cracks is verified.The time-domain,frequency-domain,time-frequency domain and correlation characteristics of the vibration response signals of the track structure under different damage degrees of cracks under the transient shock and vehicle-induced vibration were analyzed.The change law of characteristic indicators of different damage degrees under two loads was analyzed,and the sensitivity and applicability of damage characteristic indicators as identification indicators were judged.The feasibility of XGBoost for the selection of orbital structure response characteristic indicators is analyzed,the algorithm is used to reduce the dimensionality of the selected index system,and the optimal parameter combination in the actual detection is determined by comparing the relative contribution of the characteristic indicators.(2)Based on pattern recognition technology,the degree of damage to track slab cracks is identified.The pattern recognition method of BP neural network is used to analyze the damage degree of track plate cracks,and the results show that XGBoost combined with BP neural network can identify the damage degree of track slab cracks in ballastless tracks,and the recognition effect is good.(3)The noise resistance of the track slab crack damage degree identification method under transient shock and vehicle-induced vibration was analyzed.It is verified that the addition of 1% and 5% Gaussian white noise is insensitive to the influence of the two recognition methods,and the recognition effect is better.The robustness of the track slab crack damage identification method under the action of vehicle-induced vibration is verified,and the temperature changes of 5 °C,10 °C and 15 °C are not sensitive to the influence of the identification method,and the stability is good,which proves the reliability of the two methods.