关键词: 高速铁路, 无砟轨道, BP神经网络, 预防性维修, 监测

Abstract: The ballastless track systems of high-speed railways directly beared the high-frequency impacts and vibrations of high-speed trains, and the state changes of rail stretch deformation seriously affected trained safety. Traditional periodic maintenances were prone to overhaul or disrepair. A condition monitoring platform for ballastless track systems of high-speed railways was established by using fiber Bragg grating sensing technology, and the monitoring contents and monitoring point layout schemes were proposed. By analyzing the basic characteristics of monitoring data and combining with the track quality index, a back propagation(BP) neural network prediction model for ballastless track states was established， which was used to predict the service stats of rails. Prediction results guide the operation and maintenance departments to carry out timely and reasonable track maintenance and maintenance, so as to achieve preventive maintenance of high-speed rail systems.

Key words: high-speed railway, ballastless track, back propagation(BP) neural network, preventive maintenance, monitor