LPV Identification of a Heat Distribution System

This paper deals with incremental system identification of district
heating systems
to improve control performance. As long as various
parameters, e.g. valve settings, are kept fixed, the
dynamics of district heating systems can be approximated well by
linear models; however, the dynamics change significantly when these
parameters change. For this reason, we attempt to identify the system
using linear parameter varying models. We demonstrate how the
so-called ``Hansen Scheme," for linear time-invariant systems,
can be employed for incremental identification of linear
parameter varying systems as well. The approach is tested on a
laboratory setup emulating a district heating system, where local
controllers regulate pumps connected to a common
supply. Experiments show that cross-couplings in the system can indeed
be identified in closed-loop operation.