Neural networks for railway hazard analysis

The ability of neural networks to learn by example instead of simply applying a set of formalised rules represents an innovative factor of great interest in terms of studying railway reliability. The proposed method overcomes the functional limits of traditional techniques of hazard rate analysis by taking into account several influencing factors. The primary purpose of this case study is to examine the applicability of neural-network methodology to reliability analysis and obtain an estimate of the failures occurring in a system, however complex it may be. Such an estimate is based on outline parameters (functional, environmental, maintenance, etc.) on which the probability of failure potentially depends. The approach thus allows evaluating the robustness of the design/product under consideration with the aim of making continuous improvement during its future application.