Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/5191
標題: 遺傳演算法在地下水復育系統的不確定性之應用
Genetic Algorithms
作者: 林妙貞
LIN, Maio-chen
關鍵字: 遺傳演算法
出版社: 環境工程學系
摘要: 
The most difficult job of applying the groundwater management
models to the contaminated aquifer remediation problems is the
consideration of the influence of the uncertainties according to
the unknown aquifer properties and parameters. In the past,
the parameters of the aquifer flow and contamination transport
models are often assumed as known already. However, in the
real aquifer conditions, the uncertainties of the modeling
parameters play a very important role and will significantly
affect the reliability of the modeling results. This
research develops an optimization model which combines genetic
algorithms with a groundwater simulation model to solve
groundwater remediation problems. In light of the importance
of the parameter uncertainty, the spacious uncertainty of the
conductivity of the aquifer is also taken into account for
modeling analysis to investigate its impact upon the
reliability of the groundwater remediation schemes. The
performances of the genetic operators and the parameters used
in the genetic algorithms model are also evaluated in this
research to find out the suitable operators and parameter
values for the case studied in this research. Furthermore, a
new penalty function is used to improve the handling of the
constraint violation. The objectives of the pump-and-
treat remediation systems are to minimize the total pumping
rates and the total number of pumping wells. There are two
different situations:(1) only pumping wells are used;(2) both
pumping and injection wells are considered. The optimal
remediation strategies are obtained from the optimization model
by considering a single conductivity realization and muiltiple
conductivity realizations. The reliability of each optimal
remediation strategy is verified by Monte Carlo Simulation.
The results show that the reliability increases as more
conductivity relizations were taken into account in the
optimization model. When 30 conductivity realizations were
used, the reliability of the remediation strategy is about
90%. This shows that using multiple conductivity realizations
as model constraints can actually get highly reliable
remediation strategies. This research also finds that
remediation strategies which use both pumping and injection
wells have higher reliabilities and lower pumping rates,
indicating that with complement of injection wells in the
groundwater remediation systems, the cleanup goals could be
reached more effectively. About the comparison of different
types of objective functions, the results show that minimizing
well numbers can more practically meet the requirement of
minimizing cost.
URI: http://hdl.handle.net/11455/5191
Appears in Collections:環境工程學系所

Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.