A novel self-learning optimal control approach for decentralized guaranteed cost control of a class of complex nonlinear systems

Abstract

In this paper, a novel self-learning optimal control approach is established to design the decentralized guaranteed cost control of a class of complex nonlinear systems under uncertain environment. By expressing the interconnected sub-systems as a whole system, establishing an appropriate bounded function, and defining a modified cost function, the decentralized guaranteed cost control problem is transformed into an optimal control problem. Then, the online policy iteration algorithm is employed to solve iteratively the modified Hamilton-Jacobi-Bellman equation corresponding to the nominal system. A critic neural network is constructed to obtain the optimal control approximately. At last, a simulation example is provided to verify the effectiveness of the present control approach.