A method based on non-linear regression has been proposed by Toh and Devana
than (1996) to perform closed-loop process identification under naturally o
ccurring load disturbances. The significance of this method includes relaxa
tion of an important assumption on the knowledge of the input excitation so
urce, given a stable controller which is to be improved. While not aiming t
o identify a precise model as in classical design, this method uses an appr
oximate process model for on-line identification. This serves as a basis fo
r controller refinement, without having to interrupt the process operation
for identification test. In this paper, we present a refined formulation an
d then perform convexity analysis and quasiconvexity analysis on this refin
ed formulation. We show that these results can be utilized to obtain conver
gence independent of the initial estimates. Applicability of the identifica
tion for controller tuning is also demonstrated using a process with an unk
nown model.