The contact force of pantograph is one of the most important indexes to evaluate the performance of current-collection. The quality of maintenance for contact wires will be improved by effective utilization of the contact force. However, methods to use the contact force in a concrete manner have not been proposed as of yet. This paper discusses whether it is possible to estimate the wear and the strain of contact wire from the contact force. For this purpose, the learning capability of an artificial neural network is used. Moreover, a complex number version of back-propagation algorithm, which uses complex numbers, is used for a learning weight. As the result of this study, we verified that this method was valid to estimate the wear and the strain of contact wire using the contact force.

Copyright (c) 2006 Railway Technical Research Institute