Fuzzy Approach to Car Retarding Adaptation on Hump Yards

The paper presents some theoretical basis of the approach to knowledge representation for calculation of retarding strategy on railway hump yards. The common approach based on strict algorithms is shown. The limitations of such algorithm are proved based on the real data collected from objects of Russian railways. It is also shown that periodically some “artifacts” are arisen. Such “artifacts” are caused by NE-factors, which cannot be formalized by the accurate calculations. To decide this problem, the linguistic variables describing retarding process are developed and fuzzy model based on these variables is proposed. As the future work, it is proposed to develop fuzzy-neural network, which will allow to generate new fuzzy rules unobserved by human-experts as initial.