A Model Predictive Pattern Tracking Control Algorithm for Automatic Train Operation System.

Abstract

Automatic Train Operation (ATO) system is a system to control a train running automatically. It is required to operate a train to run satisfying some constraints concerned with distance, time, speed limit, and to run with little energy and riding comfort. The ATO system in this paper calculates how to run (running pattern) for a train concerned about these conditions before running as a reference, and controls a train to run along the running pattern while running.Amodel predictive pattern tracking control algorithm for this type of ATO system is described in this paper. It is an algorithm to control notch switching to cancel effects of condition variances and disturbances during running, , and to make a train run tracking a reference running pattern with riding comfort.This algorithm predicts how a train runs in future and selects times to switch notches leading smaller error between the train running and a reference running pattern. It realizes comfortable running with few times of notch switchings and accurate stopping of a train, a non-linear and discrete system, by adjusting notch switching times of the reference running pattern using Simulation-based Model Predictive Control.