ABSTRACT
Traditional control theory has the great control capability to clear system, but it is
hard to get a high control accuracy level to complex or hard to accurately described one.
However, fuzzy control is primarily through imitating people’s control experience
instead of depending on the plant model. So there has been a new method to use fuzzy
control to resolve the control difficulty of complex issues.
With the development of this field, intelligent control shows the greater charm due
to combination with fuzzy control and neural network. Although intelligent control
makes delightful progress, it also need enhance some side especially study for algorithm
adaptability. And the simulation technology is another side affects the intelligent control
research. At the present, whereas the variety of the algorithm and the complexity of the
plant model, it should depends on simulation to analysis the feasibility of the algorithm.
It can not only simulate the complex plant model but also analysis the control
characteristic much better to help, improve and optimize the algorithm.
This paper raises the fuzzy neural network control algorithm based on
Takagi-Sugeno mode. Aiming at the enhancement of the algorithm’s adaptability, it will
use conjugate gradient method instead of steepest descent method. And at the same time,
adjust the learning rate to optimize the fuzzy neural network control algorithm in order
to get the better adaptability, that’s named adaptive fuzzy control in this paper. To get
the better simulation result, this paper raises a simulation system that’s made up of two
industrial computers to simulate the plant mode and the controller. This is meaningful to
research complex control algorithm. The algorithm of plant mode and controller are
programmed by MATLAB language, and then use LabVIEW as the development
platform. Through simulation and experiment of fuzzy RBF neural network controller
and mixed Pi-Sigma neural network which proves adaptive fuzzy controller has the
much better adaptability to different plant mode. Make use of mixed programming
technology making them good integration, through experiment to validate the validity
of the simulation system. Through flux real-time control experiment, it shows adaptive
fuzzy controller based on LabVIEW has good real-time performance.
Key Word: Fuzzy Neural Network, Adaptive Fuzzy Control, Conjugate
Gradient Method, Simulation Technology