作者: Bo Zhao , Derong Liu , Yuanchun Li
DOI: 10.1109/SSCI.2017.8280857
关键词: Approximation error 、 Adaptive system 、 Identifier 、 Computer science 、 Identification error 、 Robustness (computer science) 、 Artificial neural network 、 Decentralised system 、 Control theory
摘要: This paper investigates a data-based robust decentralized stabilizing control scheme for unknown large-scale systems via adaptive critic designs. The consists of near-optimal and robustifying compensation. In order to avoid the common assumptions boundedness matched condition on interconnections, actual states interconnected subsystems are replaced with their desired ones. By using local input-output data, subsystem dynamics can be obtained by neural network (NN) identifier. Then, help NN, is derived identifier-critic architecture corresponding subsystems. Considering replacement error, identification error approximation as overall an term added overcome it in real-time. Simulation example given verify effectiveness present scheme.