-
n the work described in this article, controlling the strip
- 上一篇：高速冲击下无网格模拟英文文献和翻译
- 下一篇：直立式风力发电机英文文献和中文翻译
I thickness of a steel rolling mill provides a real-world applica-
tion of nonlinear neural control. Different control structures
based on neural models of the simulated plant are proposed. The
results of the neural controllers, among them internal model
control and model predictive control, are compared with the
performance of a conventional PI controller. Exploiting the ad-
vantage of the nonlinear modeling technique, all neural ap-
proaches increase the control precision. In this case, the
combinalion of a neural model as a feedforward controller with
a feedback controller of integral type gives the best results.
Artificial Neural Networks 6778
Artificial neural networks have been the focus of a great deal
of attention during the last decade, due to their capabilities in
solving nonlinear problems by learning. Such networks provide
a paralle. structure with very simple processing elements. This
paradigm has been applied to solve different technological prob-
lems, such as signal classification and, more recently, nonlinear
and adaptive control problems [ 1 1.
Controlling the strip thickness of a cold rolling mill is an
application of the Daimler-Benz company AEG. The steel rolling
mill has :several characteristics, such as varying time delays and
varying gains, which make it appropriate to be tackled with
neural approaches. This is because neural networks offer some
distinct advantages over conventional approaches, providing a
general framework for modeling and control of nonlinear sys-
tems. Some work on using neural nets for modeling a rolling mill
has alreajy been done (see [2] and [3]). In particular, both papers
describe the use of neural network models to predict process
parameters for a certain rolling task. By incorporating a priori
knowledge about the rolling process into the training procedure
the authors of [2] achieved a better generalization in regions in
which training data was insufficient. The predicted values were
used to !.et up the controller parameters of a conventional con-
trolling xheme, which is a type of supervisory control. On the
other hand, the use of neural models for direct control has no
been investigated. Here, we focus on the generation of neura
models of the plant and their online use in different contro
schemes. such as model predictive control. Neural net based structures can be applied even if no exact
mathematical representation of the process exists; that is, models
can be derived from observations of the input-output behavior of
the plant. This leads to a significantly reduced controller design
effort in comparison with design techniques needed for setting
up the parameters of conventional controllers. Neural network
based controllers generally depend less on apriori knowledge of
the process to be controlled.
The next section of this article defines the control problem
and briefly introduces the model used for simulation of the plant.
The third section describes the traditional approach to solve this
control problem by using a linear controller. The fourth part deals
with modeling aspects using neural networks, and section five
compares control solutions using these neural network models.
Finally, conclusions are drawn including possible directions of