Optimal low-sensitivity linear feedback systems

The paper considers the stochastic linear regulator and tracking problem for multivariable time-invariant systems. It is shown that in the limiting case, where the matrix weighting the input in the quadratic criterion is the zero matrix, the closed-loop system is insensitive to parameter variations in the sense of Cruz-Perkins, provided that the system to be controlled is minimum-phase. The weighting matrix in the Cruz-Perkins sensitivity criterion turns out to be the inverse of the covariance matrix of the measurement noise. A simple example illustrates the decrease of sensitivity obtained for a system with two inputs and two outputs.