Decentralized state feedback robust H∞ control using an evolutionary optimization approach

This paper presents a new method for constructing a decentralized state feedback controller to solve a robust H^∞ control problem for an uncertain linear time-invariant system. The uncertainties in the system are described by Integral Quadratic Constraints. The proposed method involves solving an H^∞ Riccati inequality which is dependent on a set of parameters. The resulting decentralized controller gain matrix has a block diagonal form as the off-diagonal parts of the centralized controller are treated as uncertainties. This approach enables the controller to exploit the interconnections between subsystems. The optimization problem arising in the proposed method is solved using an evolutionary optimization approach, namely the Differential Evolution algorithm.