Recommendations for implementing artificial intelligence (AI) technology to improve the fault diagnostic process for railcar systems and subsystems are provided for transit railcar maintenance professionals.

This research project was initiated in response to the desire of transit agencies to improve the effectiveness and efficiency of the railcar fault diagnostic process by reducing the extensive labor-hours devoted to fault diagnosis and the large number of errors made in the diagnostic process. The objectives of the project were to assess the potential application of AI techniques in diagnostic practices in the railcar maintenance environment and, where appropriate, to recommend steps to introduce such practices. The researchers (1) identified AI techniques that are applicable to the diagnosis or prediction of railcar failures; (2) identified the AI techniques with high probabilities of success; (3) estimated the magnitude of potential benefits from using these techniques; (4) identified in order of priority the railcar subsystems (e.g., propulsion, brakes, doors) that benefit most from application of each of these techniques; and (5) developed a research program for systematically evaluating and implementing these techniques. Seven AI techniques--expert systems, case-based reasoning, model-based reasoning, artificial neural networks, computer vision, fuzzy logic, and knowledge-based systems--were investigated to determine their potential for applications to the diagnosis of transit railcar systems and subsystems. The report concludes that AI technology is sufficiently mature for cost-effective application in the transit railcar diagnostic process and provides recommendations for implementation of the technology.

The final report for this project was published as TCRP Report No. 1, "Artificial Intelligence for Transit Railcar Diagnostics." Funding for a demonstration of AI technology as recommended in the final report has been approved as TCRP Project E-2A. The demonstration will use a combination of model-based reasoning and expert systems AI technologies to develop a "mechanics' assistant" tool that can be used to diagnose problems with the railcar propulsion system (including traction motors, gearboxes, power switchgear, and control logic units). It has been estimated that this AI technology could pay for itself in 1 year if it can provide a 7.2 percent reduction in the propulsion system mean time to repair.

TCRP Report 1 is also available in portable document format (PDF). Double-click on the files below to access this report. (A free copy of Adobe Acrobat Reader is available at https://www.adobe.com).

Front matter; Chapter 1: Introduction and Research Approach; Chapter 2: Findings; Chapter 3: Application; and Chapter 4: Conclusions and Suggested Research

Appendixes A-E