A Back-Analysis Technique for Condition Assessment of Ballasted Railway Tracks

Track substructure is a key component of railway transportation systems. Similar to the built environment of other surface transportation systems, track substructures are subjected to ageing and deterioration. This frequently leads to failure and collapse of systems and imposing costly repairs and maintenance. Further, limited knowledge about the substructure condition leads to employing inefficient, time-consuming, and expensive maintenance. As the importance of time and budget limitation, there is a need to develop more time and cost-efficient techniques for frequent condition assessment of the existing railway substructures. Falling weight deflectometer (FWD) is recognised as an effective non-destructive test (NDT) for surveying the ballasted railway substructures through the back-analysis process, including a forward analysis of track substructure and an optimisation method. This paper presents a novel hybrid back-analysis technique, including artificial neural network (ANN) and ant colony optimisation for the continuous domain (ACO_R) to backcalculate substructure layer moduli of railway track. To this aim, a dynamic finite element (FE) model is developed to generate a reliable dataset which is covering various layer moduli for ANN training. ACO_R is employed as an optimisation tool to optimise estimated layer moduli (ANN’s input). Furthermore, a validation study has been conducted using the developed FE model with back-analysed layer moduli values to evaluate the developed technique’s performance. The validation study results show that use of ANNs incorporates ACO_R results in excellent performance and robustness of the developed back-analysis technique. The hybrid ANN-ACO_R back-analysis technique is a computationally efficient method with no dependency on seed modulus values.