Many real systems may be high-order, time-varying and/or non-linear to
the extent that conventional modeling and analysis techniques can no
longer be easily applied. The use of qualitative reasoning can avoid c
omplex mathematical operations, and can be employed to overcome some o
f the difficulties, Qualitative reasoning is attractive because of the
generality of the description of physical systems involved, which mea
ns that one particular qualitative model can be used to describe a lar
ge range of operating conditions (Abdulmajid and Wynne, 1989). Most ap
plications of qualitative reasoning in engineering have so far been co
ncerned with process monitoring and diagnosis. This paper illustrates
how to apply qualitative reasoning techniques to control problems, The
principles of qualitative reasoning are applied to a liquid level con
trol system. A qualitative controller is developed on the basis of cau
sal reasoning. The control rules are stored in the program as a look-l
ip table. When the program is executed, it starts with some initial co
nditions, and generates the required states of system variables. The c
ontrol action will be determined by the qualitative control rules to b
ring the output to the desired set-point, The simulation and experimen
tal results show that the qualitative controller works well for both l
inear and non-linear models, and also when rime delay is included in t
he system. (C) 1997 Published by Elsevier Science Ltd. All rights rese
rved.