Multireservoir flood-control optimization with neural-based linear channel level routing under tidal effects

In order to determine the optimal hourly releases from reservoirs under the estuary tidal effects during typhoon periods, this paper develops a generalized multipurpose multireservoir optimization model for basin-scale flood control. The model objectives include: preventing the reservoir dam and the downstream river embankment from overtopping, and meeting reservoir target storage at the end of flood. The model constraints include the reservoir operations and the neural-based linear channel level routing. The proposed channel level routing developed from the feed-forward back-propagation neural network is employed to estimate the downstream water levels. The developed optimization model has been applied to the Tanshui River Basin system in Taiwan. The results obtained by the optimization model, in contrast to historical records, demonstrate successfully the practicability in solving the problem of flood control operations.