Distributed State Estimation for Linear Systems With Application to Full-Car Active Suspension Systems

Abstract

In this article, we treat the problem of distributed state estimation for medium- to large-scale linear systems. We consider a system, which consists of several subsystems, and assume that there already exist linear local estimators for the individual subsystems. These existing local estimators could be designed using algorithms of different types. We propose a systematic approach to connect the local estimators through information flows, such that a distributed state estimation scheme is formed for the linear system. The convergence and boundedness of the estimation error for the developed distributed scheme are proven. The proposed method is applied to a full-car active suspension system. The entire system is decomposed into five subsystems interacting with each other. A distributed estimation architecture is established based on the proposed method. The effectiveness of the proposed approach for the full-car suspension system is illustrated using simulation results, which confirm good estimation performance.