Computationally efficient methods for digital control
The problem of designing a digital controller is considered with the novelty of explicitly taking into account the computation cost of the controller implementation. A class of controller emulation methods inspired by numerical analysis is proposed. Through various examples it is shown that these methods are capable of outperforming the conventional ones, in the sense of achieving a better performance/computation cost tradeoff. The performance indexes consider (i) the error between the continuous-time controller to be emulated and the digital controller; and (ii) the closed-loop induced norm.