|
The computer''s calculational ability has been much
improvement in the last year. Artificial Neural Network (ANN)
has an important breakthrough in theoretical, and it makes the
ANN become very popular method of research among the Artificial
Intelligence (AI). ANN basis on simulation of thinking and the
ability to memorize of a living thing. Back-Propagation Neural
Network (BPN) model is universality in exercise in this paper of
ANN. Three Rainfall-Runoff-Models harmonies with average
rainfall of catchment and direct rainfall of every raingages and
every average rainfall on geographical region in this paper. The
hourly rainfall data collected from 6 recordingraingauges over
TsengWen reservoir catchment are used for case study. Using
rainfall data and inflow data in input-layer and using objective
data in output-layer. After training, about property of
Rainfall-Runoff-Model can be separated and memorized in ANN.
Providing to analyze method of Rainfall-Runoff-Model for
catchment of large area in this paper. If learning direction is
considered by application direction in ANN, then using
''Observation Learning'' supply to operate in the scene and using
''Simulation Learning''supply to design. On the model to simplify,
network can be pruning referred figure of weight sensitivity
analysis.
|