RAILWAY MAINTENANCE AND SAFETY: ARTIFICIAL INTELLIGENCE LINKS

There are numerous variables interacting in a complex manner which due to the large amount of data available, cannot be explicitly described by an algorithm, a set of equations or a set of rules in the railway assessment. In any situation, there may be both a shortage of key information and an excess of other information. Neural network and approximate logic techniques have demonstrated its usefulness and accuracy in predicting accidents that would occur under different combinations of conditions in some fields in parallel and aviation industries. This paper presents the recent research result of the development of fuzzy linguistic risk levels using approximate logic approach to deal with uncertainty with our industrial partners. Expert and engineering judgements are then mapped and transferred to neural network models of an intelligent safety prediction system for railway infrastructure safety analysis. It will be evaluating the accuracy of risk predictions made by conventional (statistical) and artificial intelligence techniques. For the covering abstract see ITRD E123761.