Fuzzy model-based control of a drum boiler-turbine system
Ahcène Habbi and Mimoun Zelmat
Applied Automation Laboratory, F.H.C., University of Boumerdès
Avenue de l’indépendance, 35000 Boumerdès, Algeria
Phone/Fax.: +213-24-816905, E-mail: habbi_hacene@hotmail.com
Abstract– This paper addresses the design of a fuzzy
control system with a fuzzy controller and a fuzzy
estimator for a fossil fuelled drum boiler-turbinegenerator
unit on the basis of a dynamic fuzzy model
that was developed in our recently published paper. In
the design procedure, a dynamics augmentation is
introduced and a dynamic fuzzy augmented system is
then determined for the nonlinear boiler-turbine
system to deal with its non-minimum phase behaviour.
The fuzzy control system is designed from a local
concept viewpoint on the basis of the optimal control
theory. The good performance of the designed fuzzy
control system is shown by simulations in various
conditions.
I. INTRODUCTION
Due to the deregulation in energy market, the power
companies are under ever increasing pressure to improve
the efficiency of their industrial equipment. For instance,
a combined cycle power plant used for cogeneration of
electric and thermal energy may need to provide a large
amount of steam on demand while at the same time
maintain balance in power generation. The drum boilerturbine-
generator unit is a critical part to the power plant.
It is difficult to change production configurations in a
boiler-turbine-generator unit because of the resulting
major disturbances in energy and material balance. When
the process losses this balance, it becomes much more
difficult to control due to the changes in process
dynamics. Because of the complicated dynamics of the
boiler-turbine system many modeling efforts have been
made [Bell and Åström, 1987; Flynn and O’Malley, 1999;
Åström and Bell, 2000]. Models that are suitable for
control design have also been investigated in many
papers, among them the more recent ones are [Åström and
Bell, 2000] and [Mayne et al., 2000]. In our recently
published paper [Habbi et al., 2003], we developed a
dynamic fuzzy model for a 160MW drum boiler-turbine
system. The principal model components of the steam
power plant are shown in Fig. 1. We demonstrated that
the proposed fuzzy model captures well the key
dynamical properties of the physical plant over a wide
operating range, and is suitable for model-based control
design. Control design for the boiler-turbine system is still
of substantial interest. One of the key difficulties is the
control of the water level in the drum boiler. Water level
dynamics are non-minimum phase, because of the shrinkand-
swell effect, and vary very much with the load. In
addition, both the dynamics and the material properties of
the fluid flowing through different types of tubes are
highly nonlinear and key quantities like the dry-out point
are difficult to measure. The boiler-turbine control
system, which controls the electrical output, the drum
steam pressure and the drum water level is necessary for
the stable load following, the safety of the power plant
and fuel saving. Most of the commercial power plant
controllers are synthesized based on the conventional
control theory. Due to the complexity of the drum level
process, three-element control system has been
developed. In such control scheme, PID controllers are
used. The problem is that the three-element control
system can only work in a limited operating range. When
there is a large disturbance of feedwater and steam
outflow, the control system is difficult to maintain in the
automatic mode. The need for simultaneous controlling of
the strongly interacting variables of the boiler-turbine
system makes the control of this complex system an ideal
application for advanced control. Without good control of
these variables, flexible production of energy to meet
demands will be difficult. On the issue of singularly
perturbed model based control, Patre et al. developed a
periodic output feedback controller for a steam power
plant, see [Patre et al., 1999]. However, the proposed
controller does not have much practical implications since
the design procedure is achieved based on simplified
dynamical equations of the steam power plant. The fifthorder
linear model used for the control design does not
capture well the key dynamical properties of the physical
plant. In [Kwon et al., 1989], Kwon and co-workers
investigated the use of the robust control theory for
designing a multivariable robust controller for the boilerturbine
system. The LQG method has been employed as
one of the robust control design techniques with loop
shaping procedure. The basic limitation is that the
controller is designed based on the linearized model of the
nonlinear plant. It may be inadequate for a wide operating
range and is sensitive to parameter variations and high
load changes.
The goal of this work is to propose a scheme for
designing a global fuzzy controller to control the boilerturbine
system represented by the dynamic fuzzy model
we developed in [Habbi et al., 2003]. The fuzzy controller
design method is achieved from a local viewpoint. First, a
fuzzy augmented system is suggested to deal with the
non-minimum phase behaviour of the plant and to meet a
desired loop shape. Then, quadratic optimization problem
is solved for each local fuzzy augmented system and a
global fuzzy controller is deduced for the global fuzzy
system. Finally, a fuzzy estimator is built upon classical
estimation theory using a local concept approach. The
.....