TY - JOUR
T1 - A new decentralised controller design method for a class of strongly interconnected systems
AU - Duan, Zhisheng
AU - Jiang, Zhong Ping
AU - Huang, Lin
N1 - Funding Information:
This work is supported by the National Natural Science Foundation of China [grant numbers 61225013 and 11332001], and partly by NSF [grant number ECCS-1230040].
Publisher Copyright:
© 2016 Informa UK Limited, trading as Taylor & Francis Group.
PY - 2017/2/1
Y1 - 2017/2/1
N2 - In this paper, two interconnected structures are first discussed, under which some closed-loop subsystems must be unstable to make the whole interconnected system stable, which can be viewed as a kind of strongly interconnected systems. Then, comparisons with small gain theorem are discussed and large gain interconnected characteristics are shown. A new approach for the design of decentralised controllers is presented by determining the Lyapunov function structure previously, which allows the existence of unstable subsystems. By fully utilising the orthogonal space information of input matrix, some new understandings are presented for the construction of Lyapunov matrix. This new method can deal with decentralised state feedback, static output feedback and dynamic output feedback controllers in a unified framework. Furthermore, in order to reduce the design conservativeness and deal with robustness, a new robust decentralised controller design method is given by combining with the parameter-dependent Lyapunov function method. Some basic rules are provided for the choice of initial variables in Lyapunov matrix or new introduced slack matrices. As byproducts, some linear matrix inequality based sufficient conditions are established for centralised static output feedback stabilisation. Effects of unstable subsystems in nonlinear Lur'e systems are further discussed. The corresponding decentralised controller design method is presented for absolute stability. The examples illustrate that the new method is significantly effective.
AB - In this paper, two interconnected structures are first discussed, under which some closed-loop subsystems must be unstable to make the whole interconnected system stable, which can be viewed as a kind of strongly interconnected systems. Then, comparisons with small gain theorem are discussed and large gain interconnected characteristics are shown. A new approach for the design of decentralised controllers is presented by determining the Lyapunov function structure previously, which allows the existence of unstable subsystems. By fully utilising the orthogonal space information of input matrix, some new understandings are presented for the construction of Lyapunov matrix. This new method can deal with decentralised state feedback, static output feedback and dynamic output feedback controllers in a unified framework. Furthermore, in order to reduce the design conservativeness and deal with robustness, a new robust decentralised controller design method is given by combining with the parameter-dependent Lyapunov function method. Some basic rules are provided for the choice of initial variables in Lyapunov matrix or new introduced slack matrices. As byproducts, some linear matrix inequality based sufficient conditions are established for centralised static output feedback stabilisation. Effects of unstable subsystems in nonlinear Lur'e systems are further discussed. The corresponding decentralised controller design method is presented for absolute stability. The examples illustrate that the new method is significantly effective.
KW - Decentralised control
KW - absolute stability
KW - large gain coupling, strict positive realness (SPR)
KW - unstable subsystems
UR - http://www.scopus.com/inward/record.url?scp=84969246744&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84969246744&partnerID=8YFLogxK
U2 - 10.1080/00207179.2016.1173230
DO - 10.1080/00207179.2016.1173230
M3 - Article
AN - SCOPUS:84969246744
SN - 0020-7179
VL - 90
SP - 217
EP - 233
JO - International Journal of Control
JF - International Journal of Control
IS - 2
ER -