Analysis of automatically collected rail surface defect data for rail maintenance management

Fatigue-related defects on the surface of the rail cause functional and operational problems for railroads. Such problems include increased dynamic forces of the train wheels on the rail resulting in reduced life of rail and other components of the track. They are also precursors to internal fatigue defects and their presence prevents detection of internal defects by ultrasonic contact sensors. Concern over the frequency and consequences of these surface defects has prompted rail researchers to look more closely at their causes, their impacts and methods to remove them. However, there is no existing method to collect and process surface condition data quickly and inexpensively.;This thesis describes an automated video data collection system that can be used to acquire and process continuous images of the rail surface to determine the presence or absence of defects. The concepts of image processing and object recognition and their role in optical inspection systems are discussed in both the general applications and in rail surface defect detection. A system including hardware, software algorithms and procedures for laboratory-based automated defect detection is developed for continuous data processing and storage. The system is used to process real data.;The automatically collected and processed data can be used by railroad engineers and planners in making short-term and long-term maintenance decisions. However, the processed data is very disaggregate and needs to be organized and aggregated to be useful. Analytical techniques for aggregating the condition data for planning maintenance or modeling deterioration are described and the solutions to the aggregation problem are developed for both perfect and imperfect condition data for the purpose of rail grinding. The rail grinding problem is formulated as a set packing integer programming problem and heuristic solution algorithms are developed and applied to simulated and field data. Finally, the role of the research in overall rail and track maintenance management is discussed qualitatively.