東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Hybrid hard-soft computing approach ...

Zou, Rui.

FindBook

Google Book

Amazon

博客來

Hybrid hard-soft computing approach for data assimilation and uncertainty analysis of water quality modeling.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	Hybrid hard-soft computing approach for data assimilation and uncertainty analysis of water quality modeling./
作者:	Zou, Rui.
面頁冊數:	253 p.
附註:	Adviser: Wu-Seng Lung.
Contained By:	Dissertation Abstracts International63-06B.
標題:	Engineering, Civil. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3057516
ISBN:	0493728805

Hybrid hard-soft computing approach for data assimilation and uncertainty analysis of water quality modeling.
Zou, Rui.

Hybrid hard-soft computing approach for data assimilation and uncertainty analysis of water quality modeling. - 253 p.

Adviser: Wu-Seng Lung.

Thesis (Ph.D.)--University of Virginia, 2002.

This thesis presents a series of hybrid hard-soft computing approaches for robust data assimilation and parameter pattern uncertainty analysis of water quality models. The major components of the proposed methodologies are an alternating fitness genetic algorithm (AFGA) approach, a neural network (NN) embedded genetic algorithm (GA) approach, a multiple pattern projection analysis, and a compound margin of safety (MOS) method. The AFGA improves upon the standard GA in the capability of maintaining higher diversity of solution, providing a more reliable way to account for the parameter pattern uncertainty. A direct application of a GA to solve the inverse problem for a complicated water quality model, however, is infeasible due to the computational bottleneck. The NN embedded GA approach is proposed to avoid the computational bottleneck through using a fast NN functional evaluator to replace a time-consuming water quality model. A series of multiple layer feedforward NN models are developed as functional approximators for the input-output relationship underlying the water quality models, where the effects of network size, training parameters, and the insensitive parameters of water quality models are investigated. The existing NN embedded GA approach is found to be inapplicable in solving an inverse WQM problem, and as a remedy, an adaptive NN-GA approach is proposed based on a hypothesis regarding the role of overall generalization and specialized generalization in a NN guided GA search. The numerical examples show that the adaptive NN-GA approach is able to obtain suboptimal solutions for inverse problems of a complicated water quality model with reasonable computational efforts. Finally, a multiple pattern projection analysis method as well as a compound margin of safety method are presented to address the uncertainty and risk associated with a model-based decision making through utilizing the non-unique solutions of an inverse problem. The proposed methodologies are illustrated using two water quality models, a simple 1-D total phosphorus model, and a more complicated laterally averaged 2-D hydrodynamic and eutrophication model.

ISBN: 0493728805Subjects--Topical Terms:

783781
Engineering, Civil.

Hybrid hard-soft computing approach for data assimilation and uncertainty analysis of water quality modeling.
LDR:03094nam 2200289 a 45 001 935881
005 20110510
008 110510s2002 eng d
020 $a 0493728805
035 $a (UnM)AAI3057516
035 $a AAI3057516
040 $a UnM $c UnM
100 1 $a Zou, Rui. $3 1259583
245 1 0 $a Hybrid hard-soft computing approach for data assimilation and uncertainty analysis of water quality modeling.
300 $a 253 p.
500 $a Adviser: Wu-Seng Lung.
500 $a Source: Dissertation Abstracts International, Volume: 63-06, Section: B, page: 2940.
502 $a Thesis (Ph.D.)--University of Virginia, 2002.
520 $a This thesis presents a series of hybrid hard-soft computing approaches for robust data assimilation and parameter pattern uncertainty analysis of water quality models. The major components of the proposed methodologies are an alternating fitness genetic algorithm (AFGA) approach, a neural network (NN) embedded genetic algorithm (GA) approach, a multiple pattern projection analysis, and a compound margin of safety (MOS) method. The AFGA improves upon the standard GA in the capability of maintaining higher diversity of solution, providing a more reliable way to account for the parameter pattern uncertainty. A direct application of a GA to solve the inverse problem for a complicated water quality model, however, is infeasible due to the computational bottleneck. The NN embedded GA approach is proposed to avoid the computational bottleneck through using a fast NN functional evaluator to replace a time-consuming water quality model. A series of multiple layer feedforward NN models are developed as functional approximators for the input-output relationship underlying the water quality models, where the effects of network size, training parameters, and the insensitive parameters of water quality models are investigated. The existing NN embedded GA approach is found to be inapplicable in solving an inverse WQM problem, and as a remedy, an adaptive NN-GA approach is proposed based on a hypothesis regarding the role of overall generalization and specialized generalization in a NN guided GA search. The numerical examples show that the adaptive NN-GA approach is able to obtain suboptimal solutions for inverse problems of a complicated water quality model with reasonable computational efforts. Finally, a multiple pattern projection analysis method as well as a compound margin of safety method are presented to address the uncertainty and risk associated with a model-based decision making through utilizing the non-unique solutions of an inverse problem. The proposed methodologies are illustrated using two water quality models, a simple 1-D total phosphorus model, and a more complicated laterally averaged 2-D hydrodynamic and eutrophication model.
590 $a School code: 0246.
650 4 $a Engineering, Civil. $3 783781
650 4 $a Engineering, Environmental. $3 783782
650 4 $a Engineering, System Science. $3 1018128
690 $a 0543
690 $a 0775
690 $a 0790
710 2 0 $a University of Virginia. $3 645578
773 0 $t Dissertation Abstracts International $g 63-06B.
790 $a 0246
790 1 0 $a Lung, Wu-Seng, $e advisor
791 $a Ph.D.
792 $a 2002
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3057516