A method for lift coefficient estimation of a pantograph using sparse modelling

Abstract

It is necessary for pantographs of Japanese high-speed trains to satisfy both low noise characteristics and insensitive lift force characteristics toward the variation in wind direction and in the shape of pantograph head due to wear. Taking particular note of the profile of the pantograph head, streamline-wise profile generally leads to low noise but sensitive lift force characteristics, and rectangular profile leads to less sensitive lift force characteristics but generates larger noise. It is therefore difficult to achieve the two characteristics at the same time in a passive way. The authors propose a solution for this trade-off by applying the pantograph equipped with smoothed profile pantograph head, which gives better performance in terms of aerodynamic noise, and actuator to actively control the lift force. One of the most difficult problem to employ active control techniques on a pantograph is the measurement of its lift force. This paper proposes a method for lift coefficient estimation with realistic measurement points. The basic idea of this method is measurement of the surface pressure on the pantograph head with the use of static holes. By applying LASSO regression which is known as one of the sparse modelling technique, only 5 static holes enables to estimate the lift force with sufficient accuracy just as the lift force estimated by using 34 static holes data. This means that the proposed method allows us to materialize a simpler measurement system.

Keywords

• pantograph
• high-speed
• low-noise
• lift force
• lift coefficient
• sparse modelling
• static hole
• lasso
• l1 regularization