Identification of Railway Ballasted Track Systems from Dynamic Responses of In-Service Trains

Abstract

Railway track is one of the most important part of the railway system, and its condition monitoring is essential to ensure the safety of trains and reduce maintenance cost. An adaptive regularization approach is adopted in this paper to identify the parameters of the railway ballasted track system (substructure) from dynamic measurements on the in-service vehicles. The vehicle-track interaction system is modelled as a discrete spring-mass model on Winkler elastic foundation. Damage is defined as the stiffness reduction of the track due to foundation settlement, loosening in the rail fastener and lack of compaction of the ballast. Accelerometers are installed on the underframe of the train to capture the dynamic responses from which the interaction forces between the vehicle and the railway track are determined. The damage of the railway track can be detected via changes in the interaction force. Numerical results show that the proposed approach can identify all stiffness parameters successfully at a low moving speed and at a high sampling rate when measurement noise is involved.

Experimental investigation on the use of multiple very low-cost inertial-based devices for comfort assessment and rail track monitoring

“…Direct track geometry assessment can be performed using accelerometers installed on axle boxes, applying double integration on the acceleration data acquired, in a simplified approach [3,25] or ideally considering the rail-wheel contact modelling in the transfer function [18,[29][30][31] to estimate irregularities. Accelerometers installed on the bogie and inside the car body are widely used for running behaviour and ride comfort assessment [3,7,32]; at any rate, they also can be used for geometry analysis if the suspension displacements are known [4] or if the dynamic properties and the operating conditions are well known [33,34] and considered in the train-track dynamic model.…”

“…• Sensor grade: relative low-cost, comprising tactical, industrial, and automotive-grade accelerometers [9,[11][12][13]18,19], and very low-cost, comprising consumer-grade sensors (smartphones and similar) [20,21,23,24,52]. • Signal processing approach, i.e., whether track profile is explicitly estimated by using detailed knowledge of the vehicle dynamic model and its parameters [31,[53][54][55], or it is implicitly evaluated based on the effect of track irregularities on vehicle vibrations, considering signal-derived features in time or frequency domain [9,15,35,56,57], or indexes related to comfort [19] or safety [58].…”

Extracting Information from Axle-Box Acceleration: On the Derivation of Rail Roughness Spectra in the Presence of Wheel Roughness

“…Cano et al [15] measured the frequency of the P2 resonance peak in the ABA spectrum and used a single mass-spring model to derive the associated track stiffness, knowing the combined mass of the unsprung vehicle components, rail and sleeper. [17] is initially near-perfect but is affected by measurement noise and the presence of a wheel flat. The stiffness identified by Quirke et al [16] is within 6% of the actual stiffness in a simulated environment with 3% measurement noise but without rail roughness.…”

Damage Detection and Condition Assessment of Civil Structures

“…Jiang et al (2018) investigate a structural stiffness identification method for assessing traditional Chinese mortise-tenon joints based on a novel index. Zhu et al (2018) investigate an adaptive regularization approach to identify the parameters of a railway ballasted track system (substructure) from dynamic measurements on in-service vehicles.…”