On-Line Decentralized Control of Interconnected Systems with Unmodelled Nonlinearity and Interaction

This paper presents a completely decentralized control scheme for the control of interconnected systems with unmodelled nonlinearity and interaction. The interaction due to interconnection and the intrinsic nonlinearity associated with a subsystem are represented by aggregative deviations of state derivatives from their linearized nominal values of the decomposed subsystems. Then, based on a model following technique, the aggregative deviation are tracked with online improvement. The solution involves the design of the decentralized control system with performance close to the decomposed optimal design and the generation of corrective signals for the aggregative deviations of state derivatives. This approach completely decentralized and all the operations are subsystem based, therefore the burden of computations is reduced significantly, the global system is robust to modelling errors and structural perturbations. By the Lyapunov´s direct method, the sufficient condition for the stability of the global system is established. Computer simulations for the decentralized control of a two-link, ¿-rmanipulator are conducted.