Railway tracks deteriorate over time due to the combined effects of train loads and environment. To carry out appropriate maintenance, it is necessary to measure the condition of the track. There are two broad categories of condition measures namely: function and structural measures of conditions. Functional measures of condition assess the condition of the track from the point of view of the user. An example of a commonly used functional measure of condition is track geometry. In the UK, track geometry is measured by the track recording coach and it is carried frequently on the entire network. On the other hand, structural measures of condition assess the structural integrity of the track. Example of a commonly used structural measure of condition is track stiffness. In the UK, track stiffness is measured using the Falling Weight Deflectometer (FWD) and it is carried out less frequently on specific railway track sections as it is very slow, requires closure of the track and is expensive to operate. The aim of this research is to investigate the feasibility of using track geometry measurement to extract track stiffness information and ultimately develop a prototype automated system to achieve this.

This thesis presents work carried out over a number of years within the field of railway vehicle instrumentation. The railway industry is currently moving to be more heavily “data driven”. This means that railway organisations are putting policies into place whereby decisions have to be justified based on recorded and citable data. To achieve this, the railway industry is increasingly turning to greater and greater levels of instrumentation to deliver the data on which to base these decisions. This thesis considers not only this increased requirement for data, but the frameworks and systems that must be put into place in order first to obtain it, and then to extract useful information from it. In particular the author considers the issue of contextualisation of data, where multiple datastreams may be used to provide context for, or allow more accurate and beneficial interpretation of each other in order to support better decision making. In order to obtain this data, the thesis explores, through a series of case studies, a number of options for different instrumentation system architectures. This culminates in the development of a distributed system of embedded processors arranged in an extensible modular framework to provide a rich, coherent and integrated dataset which can then be processed contextually to yield a better understanding of the railway system.