Application of NQG Control with Uncertain and Incomplete Measurements

Nonlinear Quadratic Gaussian control is a technique for extending the standard LQG theory to nonlinear systems. Under the assumption that the statistics of the state and measured variables can be adequately represented by a gaussian distribution, a set of necessary conditions for the minimization of a quadratic cost function can be derived for nonlinear regulator systems. This control synthesis technique is applied to a position servomechanism in this paper. In particular, the system under study consists of a dc motor driving a load through a compliant shaft. The effects of coulomb friction and backlash in the transmission are considered.