Genetic algorithm stopping criteria for optimization of construction resource scheduling problems
Kim, Jin-Lee
Genetic algorithm stopping criteria for optimization of construction resource scheduling problems created by by Jin-Lee Kim - Construction Management and Economics Volume 31, number 1-3 .
Genetic algorithms (GAs) have been widely applied in the civil and construction engineering management research domain to solve difficult and complex problems such as resource-constrained project scheduling problems (RCPSPs). Generally, a trial-and-error calibration approach is used to identify values for the GA parameters. Unlike with other parameters, few studies have been done, theoretically or experimentally, for determining when to terminate GA for optimization of the RCPSP. Two genetic algorithm stopping conditions are compared to demonstrate their suitability for application in the RCPSP and to assess their ability in searching optimal solutions efficiently. The extensive computational results show that the Elitist GA, when using the unique schedule method, provides 10% more optimum values than those obtained from the Elitist GA when using the iteration method with 24% less computational time. The unique schedule stopping approach can be valuable for GA users to design their purpose driven GA for optimization of the RCPSP as it provides a better near-optimal solution with reduced computational time.
01446193
Comparative studies--Resource allocation--Genetic algorithms heuristics
HD9715.A1 CON
Genetic algorithm stopping criteria for optimization of construction resource scheduling problems created by by Jin-Lee Kim - Construction Management and Economics Volume 31, number 1-3 .
Genetic algorithms (GAs) have been widely applied in the civil and construction engineering management research domain to solve difficult and complex problems such as resource-constrained project scheduling problems (RCPSPs). Generally, a trial-and-error calibration approach is used to identify values for the GA parameters. Unlike with other parameters, few studies have been done, theoretically or experimentally, for determining when to terminate GA for optimization of the RCPSP. Two genetic algorithm stopping conditions are compared to demonstrate their suitability for application in the RCPSP and to assess their ability in searching optimal solutions efficiently. The extensive computational results show that the Elitist GA, when using the unique schedule method, provides 10% more optimum values than those obtained from the Elitist GA when using the iteration method with 24% less computational time. The unique schedule stopping approach can be valuable for GA users to design their purpose driven GA for optimization of the RCPSP as it provides a better near-optimal solution with reduced computational time.
01446193
Comparative studies--Resource allocation--Genetic algorithms heuristics
HD9715.A1 CON