On a class of adaptive suboptimal Riccati-based controllers

In this paper a nonlinear modification is applied to the feedback law resulting from the standard linear optimal control problem formulated in state-space. This modification weights the value of the state feedback matrix with a gain factor whose derivative is proportional to the difference of a weighted inner product of the state vector and the logarithm of the gain factor. The result is completely general and therefore applies to multi-input multi-output systems. An example illustrates the extent to which the convergence rate is increased using the nonlinear feedback law for large values of the state