| Summary: | 碩士 === 國立清華大學 === 核子工程學系 === 83 === Generally, dynamic system's responses were described by the
differential equations .Therefore, to model the system dynamics
of a complicated system , it requires rather long time to set
up the system governing equations. In addition, the computation
effort is also very huge. In order to reduce the computing
time, the calaulation of dynamic response of balance of plant
is usually neglected in the large system codes. This research
is to set up the dynamic model of balance of plant using the
neural networks so that the computing time can be very short.
In this research, the diagonal recurrent neural networks were
adopted, which need rather few neurons and also achieve good
performance. In addition, it converged very fast and then
reduced the training time. The data used for training were
generated by the compact simulator of Mannashan Nuclear Power
Plant which was developed by Institute of Nuclear Energy
Research and Institute for Information Industry. The whole
system was divided into six subsystems:high pressure turbine,
moisture seperator and reheater, low pressure turbine,
condenser, low pressure feedwater heater, and high pressure
feedwater heater. The neural network models were trained for
such subsystems. When the training was finished, the neural
networks of subsystem were connected to simulate the dynamic
behavior of balance of plant.
|
|---|
