Non-linear control approaches for active railway suspensions

This thesis studies various linear and non-linear control approaches for active railway suspensions. The aim of the study is to improve the system performance of active secondary suspensions in response to different track features. The primary motivation for active suspension on railway vehicles is to improve suspension performance and thereby run faster or provide a better ride quality. The problem of discriminating between the random track and deterministic track input is a fundamental problem for the design of active secondary suspensions on railway vehicle. The basic requirement of an active suspension system is to improve the ride quality without increasing the suspension deflection unacceptably when the vehicle negotiates on both straight track and deterministic track features. This thesis presents and compares different control strategies of active suspension systems for railway vehicles. Firstly, a number of linear approaches for filtering the absolute velocity signal are theoretically examined in order to optimise the trade-off between the random and deterministic input requirements. What can be achieved with linear filters is initially determined. This is quantified by the degradation in the straight track ride quality needed to restrict the maximum deflection to an acceptable level as a vehicle traverses the transition to a typical railway gradient, and a range of filter types, frequencies and absolute damping rates are assessed in order to explore the boundary of what can be achieved through linear means. Secondly, some nonlinear Kalman-Filter methods are investigated to further improve the suspension performance. Finally, a comparison between linear and non-linear strategies is studied.