水庫優養控制優選模式之發展與應用

This research first applied DDP, Dynamic
Differential Programming method, to solve the
problem of time-variant, nonlinear optimization for
reservoir eutrophication control. The Feitsui
Reservoir is the case for the research and the result
indicates the feasibility of solving the dynamics of
reservoir eutrophication optimal control problems.
The objective of the model is to minimize the
required treatment of pollutants. The conditions of
the water quality model are the equality constraints
and in addition there is the nonequality physical
restriction. This research adopted zero-dimensional
water quality model [1] and six water quality
variables has been taken into consideration. The
bio-chemical reaction is nonlinear; considering the
time-variant factor, the model is dynamic and
unsteady. The required treatment efficiency of
pollution source can then be derived from the result
of optimization. This enables the concentration of
chlorophyll-a in the lake to be under a certain level.
As the model is time-variant, the control variable
obtained is a function of time. Therefore the
control policy can be effectively adjusted
according to the change of time.