02-02-2013, 10:39 AM
NONLINEAR SIMULATION OF THE FRANCIS TURBINE NEURAL NETWORK MODEL
1NONLINEAR SIMULATION.pdf (Size: 277.18 KB / Downloads: 23)
Abstract:
Due to the difficulty in describing the nonlinear
characteristic of Francis turbine and the complex simulation
of the Francis turbine governing system (FTGS), this paper
takes advantage of the powerful nonlinear approximate
ability of the feed forward neural network to put up the
Francis turbine neural network model ("NM). The
Levenberg-Marquardt algorithm is used to train the
FTNNM which describes the flow characteristic and the
efficiency characteristic. The convergence speed of the
offline training is fast and the accuracy of the model is high.
The nonlinear model FTNNM and other models consist the
nonlinear simulation system under the environment of the
SIMULINK of MATALAB. The nonlinear simulation under
Merent operating situations can be implemented in the
system. The variability of the different inner parameters of
the system and the Francis turbine can be attained quickly
and truly. It provides a good base for the research of control
policy of the Francis turbine governing system (FTGS).
Introduction
Francis turbine governing system (FTGS) is a
complex control system, and the digital simulation is an
effective method to analyze its dynamic transition process
to design the best parameters of the controller. The
simulation of the FTGS is divided into big undulation and
small undulation according to the input signal at present.
When frequency disturbance is less than &8% rated
value or load disturbance is not bigger than 10% rated
value, small undulation appears. The system can be
regarded as a linear system under small undulation
condition.
Conclusions
Due to the complex nonlinear characteristic of the
FTGS, the nonlinear simulation should be considered no
matter weather it is under the big undulation or the small
undulation. The key of nonlinear simulation of FTGS is the
accuracy of the nonlinear model of the Francis Turbine. The
Francis Turbine neural network model FTNNM, which is
trained by the Levenberg-Marquardt algorithm, is put
forward in this paper. The model has high accuracy and
approximates the actual situation of the hydroelectric power
station.
The FTGS with the FTNNM is simulated using Simulink
Toolbox of MATLAB. Several simulation results under
different parameters and different operating modes can be
obtained rapidly and conveniently. This means, the
nonlinear simulation of the FTGS with the FTNNM is
satisfactory. It provides a good base for the research of
control policy of the Francis turbine governing system
(FTGS).