The aim of this study has been to predict the track geometry degradation rate using artificial neural network. Tack geometry measurements, asset information, and maintenance history for five line sections from the Swedish railway network were collected, processed, and prepared to develop the ANN model. The information of track was taken into account and different features of track sections were considered as model input variables. In addition, Garson method was applied to explore the relative importance of the variables affecting geometry degradation rate. By analysing the performance of the model, we found out that the ANN has an acceptable capability in explaining the variability of degradation rates in different locations of the track. In addition, it is found that the maintenance history, the degradation level after tamping, and the frequency of trains passing along the track have the strongest contributions among the considered set of features in prediction of degradation rate.

Railways are currently experiencing higher demands on safety, infrastructure performance, network capacity and service quality, etc. As a result, a high level of track availability, service quality and infrastructure robustness against unexpected events is required, at reduced cost, to meet the demands. The track geometry is one of the critical parameters influencing the performance of the infrastructure and the capacity and quality of service of the railway network. The track geometry continuously degrades and deviates from the designed level over time and under the influence of various factors. A poor track geometry negatively affects the ride quality and passenger comfort and increases the risk of train derailment. To ensure a safe and reliable track for the trains, the track geometry should be kept in an acceptable condition. This necessitates the development of an applicable and effective tamping regime, as the main maintenance action for track geometry rejuvenation. An effective tamping regime enables enhancement of the availability and safety performance of the track, to control the track degradation and to restore the damaged track to an operational state, at the lowest possible cost. 

The purpose of the research conducted for this thesis was to develop data-driven methodologies and optimization approaches for the development of a plan for predictive railway track geometry maintenance. First, a study was carried out to investigate the application of artificial neural networks to estimation of the track geometry degradation rate in spatial space by considering the operational and constructional covariates. In addition, the relative importance of the observed covariates for the track geometry degradation was also explored. The results showed that the maintenance history, the degradation value after tamping and the frequency of the trains passing along the track were the most important covariates affecting the track geometry degradation rate. Second, a case study was performed on a heavy haul railway line to analyse the isolated twist and longitudinal level defects. A data-driven model was developed to predict the occurrence of track geometry defects. A machine learning technique, namely the RUSBoost algorithm, was used to classify the track sections into healthy and unhealthy track sections. In order to capture information about the shape of the defects, first- and second-order derivatives of the track irregularities were used. The observed anomalies in the conducted case study in the pattern of the track geometry degradation were also explored. It was found that applying a combination of the kurtosis and the standard deviation to represent the track quality is beneficial for identifying the anomalies in the trend of geometry defects. Third, a simulation-based framework was developed with the aim of allocating an effective maintenance limit by assessing the effect of different maintenance limits on the total maintenance cost. The results showed that it is not one value only, but a range of values for the maintenance limits which can be selected to minimize the total maintenance costs. However, by considering the safety aspects of track geometry maintenance, it is suggested that one should select the lower bound of the range of cost-effective limits for maintenance planning. 

Finally, an optimization model was developed to schedule track tamping activities with the aim of minimizing both the track geometry maintenance costs and the number of unplanned maintenance actions. The proposed model enables infrastructure managers to examine the effect of different scenarios for the control and management of isolated defects on the track geometry condition and the maintenance cost. The results obtained in the present research are highly relevant for specific industrial challenges and are expected to enhance the capability of infrastructure managers to make effective and efficient decisions when developing their planning and scheduling of the tamping regime. The results have been verified through interaction with experienced practitioners working for a major railway infrastructure manager and railway maintenance subcontractors.

This paper presents an in-depth case study of a heavy-haul railway line in Sweden to analyze the twist and longitudinal level geometry defects. A linear model was applied to model the evolution of the amplitude of the longitudinal level defects and twist over time. Despite the effect of the defect shapes on the dynamic track loads, the amplitude of the defects still is the only criterion used for the assessment of geometry defect severity. The application of first- and second-order derivatives to capture information about the shape of defects was investigated in the case study. In addition, the RUSBoost algorithm was used to classify track sections into healthy and unhealthy sections using the imbalance class data set. In this algorithm, the standard deviation and the kurtosis of the geometry parameters were used as explanatory variables. Finally, the abnormal track geometry degradation patterns identified in the case study were explored in detail. The results of the analysis can be used directly in maintenance modeling and used for the purpose of maintenance scheduling. 

This study has been dedicated to the optimization of opportunistic tamping scheduling. The aim ofthis study has been to schedule tamping activities in such a way that the total maintenance costs andthe number of unplanned tamping activities are minimized. To achieve this, the track geometry tampingscheduling problem was defined and formulated as a mixed integer linear programming (MILP)model and a genetic algorithm was used to solve the problem. Both the standard deviation of thelongitudinal level and the extreme values of isolated defects were used to characterize the trackgeometry quality and to plan maintenance activities. The performance of the proposed model wastested on data collected from the Main Western Line in Sweden. The results show that different scenariosfor controlling and managing isolated defects will result in optimal scheduling plan. It is alsofound that to achieve more realistic results, the speed of the tamping machine and the unused life ofthe track sections should be considered in the model. Moreover, the results show that prediction ofgeometry condition without considering the destructive effect of tamping will lead to an underestimationof the maintenance needs by 2%.

A large part of the railway track geometry maintenance burden concerns local maintenance activates conducted to rectify isolated defects. Isolated defects are short irregularities in the track geometry that can dramatically increase the dynamic forces between the wheel and rail, which in turn will accelerate the growth or occurrence of internal rail defects. The dynamic force between the wheel and rail is dependent on the shape of the isolated geometry defects. However, the severity of isolated defects is mainly defined only by their amplitude. Therefore, in addition to amplitude, other characteristics of geometry defects must be considered to analyze severity of defects and to prioritize local maintenance actions. This study aims to use the first- and second-order derivatives of the longitudinal level defects to plan local maintenance activities. The derivatives of geometry defects provide useful information about the shape of defects. The information is used to categorize the isolated defects based on their severities and prioritize maintenance actions. In this regard, K-means clustering technique is applied. The results of this study will support the decision-making process regarding the planning of local maintenance activities. The foot-by-foot track geometry data collected from Main Western Line in Sweden is used to implement and test the model.

In order to evaluate the railway track geometry condition and plan maintenance activities, track inspection cars run over the track at specific times to monitor it and record geometry measurements. Applying an adequate inspection interval is vital to ensure the availability, safety and quality of the railway track, at the lowest possible cost. The aim of this study has been to investigate the effect of different inspection intervals on the track geometry condition. To achieve this, an integrated statistical model was developed to predict the track geometry condition given different inspection intervals. In order to model the evolution of the track geometry condition, a piecewise exponential model was used which considers break points at the maintenance times. Ordinal logistic regression was applied to model the probability of the occurrence of severe isolated defects. The Monte Carlo technique was used to simulate the track geometry behaviour given different inspection intervals. The results of the proposed model support the decision-making process regarding the selection of the most adequate inspection interval. The applicability of the model was tested in a case study on the Main Western Line in Sweden.

A great part of the cost of railway maintenance is related to the track geometry. Reliable and complete track geometry data form the foundation of an efficient and effective condition-based maintenance strategy. Generally, track geometry measurement data suffer from positional errors. Because the aim of track geometry degradation modeling has been to analyze the evolution of location-specific defects over time, accurate information on the position of defects is of crucial importance. Therefore, collected track geometry measurements in different inspection runs need to be pre-processed before they are used to model geometry degradation or implement a condition-based maintenance strategy. In this study, three relative position-based approaches are applied to align the positional measurement data obtained from different inspection runs. The two main types of data positional errors are shifted and stretched waveforms. In this regard, three time series alignment algorithms, i.e., crosscorrelation function, dynamic time warping, and dynamic time alignment kernel, are applied to align both shifted and stretched waveforms. Foot-by-foot track geometry data obtained from the main Western line in Sweden are used to implement and test the models. On the basis of the results, dynamic time warping outperforms the other two techniques to align shifted and stretched datasets.

The objective of this study has been to develop an approach to the allocation of an effective maintenancelimit for track geometry maintenance that leads to a minimisation of the total annual maintenancecost. A cost model was developed by considering the cost associated with inspection, preventivemaintenance, normal corrective maintenance and emergency corrective maintenance. The standarddeviation and extreme values of isolated defects of the longitudinal level were used as quality indicatorsfor preventive and corrective maintenance activities. The Monte Carlo technique was used tosimulate the track geometry behaviour under different maintenance limit scenarios and the effectivelimit was determined which minimises the total maintenance cost. The applicability of the model wastested in a case study on the Main Western Line in Sweden. Finally, a sensitivity analysis was carriedout on the inspection intervals, the emergency corrective maintenance cost and the maintenanceresponse time. The results show that there is an optimal region for selecting an effective limit.However, by considering the safety aspects in track geometry maintenance planning, it is suggestedthat the lower bound of the optimal region should be selected.

Track geometry degradation is a function of several interacting factors (e.g. traffic, environmental conditions, track structure, and maintenance history) and each interaction may follow several interrelating rules. Artificial Neural Network (ANN) showed a high capability in prediction of track geometry degradation, by analysing a set of features that explain the behaviour of geometry evolution. The aim of this study is to develop an approach using ANN to predict the degradation of track geometry along a track line. In the ANN model, maintenance history, curvature, annual traffic, presence of level cross, speed, frequency of trains passing along the track and material type and age are inputs to the model and track geometry degradation rate is the output of the model. A set of data from railway network in Sweden are used to train and validate the model. The results indicate that ANN is able to predict the degradation of track geometry.

Railway track geometry subjects to degradation with age and usage and highly affects track functionality. An accurate prediction of track geometry degradation is essential to enhance the performance of maintenance planning and scheduling. In this study, artificial neural network is used to predict track geometry degradation rate. The data of five lines from Swedish railway network are used to develop the model. The standard deviation of the longitudinal level is used as quality indicator for track geometry degradation. For all track sections, a set of features which may contribute in track geometry degradation are collected. Principal component analysis method is used to reduce the dimension of the features to facilitate the training of the ANN model. Then, the new principal features considered as inputs to ANN model. The results indicate that the proposed method can be used to predict track geometry degradation along the track line.