On nonlinear controllers for persistent disturbance rejection

An l/sub 1/-optimal linear time-invariant (LTI) compensator may have an order significantly higher than that of the plant. In the state feedback case, there has been recent work exploring the use of nonlinear static feedback controllers which provide near optimal performance. Here we show that such controllers are noise tolerant and we derive a superposition-like bound on the closed loop performance in the event that the plant initial condition, the disturbance, and the noise are all non-zero.