Page 289 - Distributed model predictive control for plant-wide systems

12

High-Speed Train Control with

Distributed Predictive Control

12.1 Introduction

In recent years, the Chinese railway systems have gone through a massive phase of upgrada-
tion and expansion. More and more the China Railway High (CRH)-speed has been used for
dedicated passenger lines. It is estimated that by the end of 2020, China will have 18,000 km
dedicated passenger lines, with an operating speed of 350 km/h, which will cover almost the
whole country.
The CRH consists of electric multiple units (EMUs), which include motor coaches and
trailer coaches where the driving force of the CRH is distributed.
In the existing work, most researchers assumed that the couplers between adjacent coaches
are stiff so that a complete train can be regarded as a rigid body [132]. In Ref. [133], the
dynamics of high-speed train is modeled by a cascade of coaches connected by flexible cou-
plers and subjected to rolling resistances, aerodynamic drag, and wind gust. However, this
chapter assumes that the aerodynamic resistance acts on the leading coach only and the rolling
resistance acts on every coach. It is more reasonable to assume that the aerodynamic resistance
acts on every coach [134], but there still is a drawback that all the empirical constants are the
same. In Ref. [135], robust adaptive control is proposed to track velocity and aerodynamic
drag is taken as the uncertain variable and acts on each coach equally.
In Ref. [136], the aerodynamic drag in the real situation acts on every coach, and it mostly
acts on the leading coach and the last coach. Since the aerodynamic drag is proportional to
the square of the speed [136], its influence on high-speed train dynamic behavior becomes
significant. Thus, it is very important to emend the dynamic model with real aerodynamics
effects.
In this chapter, we establish a new spring–mass model with accurate parameters and a more
real hypothesis. During the running of the EMUs, the in-train force is the most important
thing for the safe driving. The in-train force is constrained strictly, besides the traction and the
brake are limited by the real condition. Considering these constraints, we can use the model
predictive control (MPC) method to handle constraints effectively [3]. In the existing research,
the article in [137] proposes a cruise control of the longitudinal train longitudinal model based

Distributed Model Predictive Control for Plant-Wide Systems, First Edition. Shaoyuan Li and Yi Zheng.
© 2015 John Wiley & Sons (Asia) Pte Ltd. Published 2015 by John Wiley & Sons (Asia) Pte Ltd.