Genetic Algorithms for the Design of Groundwater Remediation Systemsby Daene C. McKinney, (A.M.ASCE),
Abstract: The performance of genetic algorithms (GAs) to solve optimal aquifer remediation design problems is investigated. Some difficulties associated with GA models include premature convergence, model parameter selection, and algorithm approach. The appropriate GA model parameters and the handling of constraints are often problem specific. General guidelines for selecting suitable parameters and approaches are seldom available. Local hill climbing is not possible with conventional GAs and to obtain improved GA performance it may be necessary to use a hybrid combination of GA and gradient based approaches. The impact of GA parameters on model performance in solving optimal aquifer remediation problems was investigated. Results indicate that GA models with binary-coded genetic representations, large population sizes (150∼200), high multiple-point crossover probabilities (0.6∼0.8), small mutation probabilities (0.01∼0.1), and tournament selection, provide optimal performance for the example problems considered.
Subject Headings: Remediation | Algorithms | Groundwater | Hydraulic design | Parameters (statistics) | Aquifers | Probability | Convergence (mathematics) | Hybrid methods | North America | Georgia | United States
Services: Buy this book/Buy this article
Return to search