Data-based Robust Near-Optimal Decentralized Stabilization of Unknown Large-Scale Systems

Bo Zhao,Derong Liu,Yuanchun Li
DOI: https://doi.org/10.1109/ssci.2017.8280857
2017-01-01
Abstract:This paper investigates a data-based robust decentralized stabilizing control scheme for unknown large-scale systems via adaptive critic designs. The control consists of near-optimal stabilizing control and adaptive robustifying compensation. In order to avoid the common assumptions of boundedness and matched condition on interconnections, the actual states of interconnected subsystems are replaced with their desired ones. By using local input-output data, the subsystem dynamics can be obtained by neural network (NN) identifier. Then, with the help of local critic NN, the near-optimal decentralized control is derived via identifier-critic architecture for corresponding subsystems. Considering the replacement error, identification error and approximation error as overall error, an adaptive robustifying term is added to overcome it in real-time. Simulation example is given to verify the effectiveness of the present control scheme.
What problem does this paper attempt to address?