As the management and analysis of large and complex data sets is becoming an increasingly important part of TRL’s work across business areas, the TRL Academy is funding a programme of research that aims to develop and enhance machine learning. Machine learning can process large amounts of data more quickly and efficiently than manual analysis, and it may also reveal relationships between data sets that have not been considered. This report summarises the research carried out under the TRL Limited reinvestment scheme. The project considered three different types of potential applications of machine learning: Analysis of train driver behaviour (clustering), which successfully used clustering to analyse a small data set and still provide some useful conclusions; Condition forecasting of road pavements which showed that the resources of existing data sets is significant and though the final results were inconclusive the act of researching this data has built a stable framework on which to progress; and the crack detection study showed that some of the more mundane and labour intensive processes can be automated and useful results obtained. 

Want to know more about this project?