A model-free approach to linear quadratic (LQ) performance assessment
is presented in this paper. First a test for LQ optimality, based on c
losed-loop signals, is presented. Its result indicates whether the con
troller should be adjusted. From the same test it is possible to deter
mine the closed-loop pole positions which would be obtained by using t
he LQ optimal controller. These may then be used in a direct adaptive
control scheme to adjust the controller. Both state and output feedbac
k are addressed for discrete-time, single-input-single-output (SISO),
linear, time-invariant processes. (C) 1998 Elsevier Science Ltd. All r
ights reserved.