Design Of Partially Decentralized Controller And Its Application In Non-square Systems

Multivariable systems are common in industry. Decentralized control is generally used to deal with multivariable systems due to its simple structure and ease to tune. But for multivariable systems with strong coupling, decentralized control structure is often difficult to achieve high performance requirements. Partially decentralized control can take advantage of more plant information, with additional controllers on the basis of the decentralized control, thus can use a relatively simple structure to achieve comparable performance with the centralized control. Nonsquare systems are a kind of multivariable systems that the conventional control methods are often unable to get satisfactory performance, so applying partially decentralized control theory to the nonsquare system is of important theoretical and practical value.This paper proposes an independent partially decentralized controller design method, and applies it to the nonsquare systems. Partially decentralized control structure falls in between the decentralized control and multivariable centralized control. It can achieve similar performance to multivariate centralized control with a much simple structure. The method can be divided into two steps. The first step is to select partially decentralized control structure. This paper proposes a new method to select controller structure via the gap metric. This approach focuses on the robust performance of the system. It uses gap metric to measure the distance between the two systems; the smaller the distance the closer the selected partially decentralized control structure to the original system.The second step is to directly design the internal model controller viamodel inversion. Partially decentralized control system is extended to a nonsquare decentralized structure, and for the nonsquare systems, nonsquare partially decentralized control system is also extended to a nonsquare decentralized structure, and then obtain the generalized inverse matrix of the extended plant to design internal model controller., Finally by choosing a special kind of filter to ensure the stability of the system. Design examples are done for several typical nonsquare systems, the simulation results show that this method can be directly applied to the nonsquare systems and can achieve good control performance.