Designed stable-plastic neural networks based on Perceptron capable to learn without losing information stored previously. The developed neural networks allow to predict the possibility of a process of slipping wheel sets in motion a traction rolling stock based on engine rotor speed signals.
stable plastic neural network, perceptron, ability to learn, forecasting slipping
"Stabylno-plastychnыe neironnыe sety na osnove pertseptrona v zadachakh prohnozyrovanyia buksovanyia" [Stable plastic neural network based perceptron in forecasting problems slippage],
Information Processing Systems,