Parametric Linear Quadratic Control: Can We Avoid a State Estimator?

D.I. Wilson (New Zealand)

Keywords

LQ, output optimal, state estimation, Kalman filter

Abstract

Parametric linear quadratic (PLQ) control is an alternative to linear quadratic control that avoids the need to explicitly design and tune a state estimator. While the design com putation for PLQ is higher than that for LQG, the perfor mance is better for linear plants in the common case where the actual process noise is underestimated. For nonlinear plants, both LQG (using an EKF) and PLQ require very small sample times to maintain stability. It is postulated that the unscented transform (UT) could counter this dis advantage of the EKF, but it is not applicable in the PLQ case.

Important Links:



Go Back