Fault-tolerant Control Technology Research Of Automatic Train Operation

Because of its characteristics of safety, flexibility, economical efficiency and high-level service, train Fully Automatic Operation (FAO) is becoming the development trend of future urban rail transit system. Accurate stopping is one of the most important performance indexes of train automatic operation system, it require the train vehicle maintain uniform stopping accuracy to ensure the running safety. Therefore, realizing accurate stopping under any circumstance is the center of train fully automatic operation controller research. In addition, train vehicles have to run continuously for long time and their loads as well as external environment change over time. So it is inevitable that the train vehicles would have some faults. Therefore, conducting the research of developing train vehicle operation faults detection and fault-tolerant control technology is highly essential.Based on the practical requirement of automatic train operation, this thesis briefly describes the theoretical research of Fault Detection and Fault-tolerant Control and points out that these theoretical achievements pave way for the research of train fully automatic operation fault-tolerant controller.With the available urban rail vehicle dynamic braking system, this thesis first studies the problem of train sensors fault detection and propose a real-time detection algorithm base on the UIO state estimation theory.This thesis also studies the problem of train braking performance supervision. By comparing with the approach of adaptive control parameter estimation and IMM model estimation, this thesis finally propose a more advanced solution based on the detection filter state estimation theory.Based on the outcome of train sensors fault detection and train braking performance supervision, the thesis finally addressing the problem of train automatic fault-tolerant operation and propose an accurate control algorithm based on model following sliding mode control algorithm. It is verified that the proposed algorithm can perform online target curve generation and overcome the impact of train parameters drift and module faults.In order to show the effectiveness of the proposed performance supervision and control algorithms, this thesis conducts series of numerical simulations in laboratory. Even though considering many harsh conditions, the simulation results indicate that the proposed algorithms are highly effective.Theoretical analysis and verification simulations show the superiority of the proposed onboard controller frame. Therefore, the achievement of this thesis will be of help for the overall research of train fully automatic operation system.