Application of a deep neural network for the response characteristics of a hydrostatic buffer

초록

Recently, hydrostatic buffers have been applied as energy absorbers in railway vehicles. The reaction force of a hydrostatic buffer changes depending on the compressive displacement and speed. The European standard EN 15227 requires the speed-dependent characteristics of hydrostatic buffers to be accounted for in simulations. In this study, a response model that can consider the speed-dependent characteristics of the hydrostatic buffer was developed based on a deep neural network (DNN) by using impact test data. This model was validated by reproducing the reaction force of the buffer in the dynamic simulation. A comparison between the DNN model and other conventional approximation methods was also presented, and the results showed that the DNN-based model was effective and accurate.

키워드