Data-driven design of fractional order H∞ controller by convex optimization
摘要
A new method of the robust fractional order controller design based on the nonparametric model on the frequency domain is proposed. The robust performance criterion is given by the mixed sensitivity condition, and it is represented as the geometric constraints for the open-loop Nyquist curve. In order to determine the controller that satisfies the geometric constraints, the fractional order controller is selected as the linear structure with respect to parameters and a set of the admissible parameters is obtained by the solution of the finite number of linear inequalities numerically. Finally, the controller parameters are determined in the obtained candidate parameters space as such that optimization criterion is met. The performance of designed controller is validated through simulation for the fractional order proportional–integral–derivative controller. Proposed method provides the possibility to design a fractional order controller from input–output measurement data without the parametric model.