Calibration of flow and water quality modeling using genetic algorithm

Research output: Chapter in book / Conference proceedingConference article published in proceeding or bookAcademic researchpeer-review

20 Citations (Scopus)

Abstract

In mathematical simulation for flow prediction and water quality management, the inappropriate use of any model parameters, which cannot be directly acquired from measurements, may introduce large errors or result in numerical instability. In this paper, the use of a genetic algorithm for determining an appropriate combination of parameter values in flow and water quality modeling is presented. The percentage error of peak value, peak time, and total volume of flow and water quality constituents are important performance measures for model prediction. The parameter calibration is based on field data of tidal as well as water quality constituents collected over five year span from 1991 to 1995 in Pearl River. Another two-year records from 1996 to 1997 are utilized to verify these parameters. Sensitivity analysis on crossover probability, mutation probability, population size, and maximum number of generations is also performed to determine the most befitting algorithm parameters. The results demonstrate that the application of genetic algorithm is able to mimic the key features of the flow and water quality process and that the calibration of models is efficient and robust.
Original languageEnglish
Title of host publicationAI 2002
Subtitle of host publicationAdvances in Artificial Intelligence - 15th Australian Joint Conference on Artificial Intelligence, Proceedings
PublisherSpringer Verlag
Pages720
Number of pages1
ISBN (Print)3540001972, 9783540001973
Publication statusPublished - 1 Jan 2002
Event15th Australian Joint Conference on Artificial Intelligence, AI 2002 - Canberra, Australia
Duration: 2 Dec 20026 Dec 2002

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2557
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference15th Australian Joint Conference on Artificial Intelligence, AI 2002
CountryAustralia
CityCanberra
Period2/12/026/12/02

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this