01095768

Component

In this study, a Parallel Genetic Algorithm (PGA) is used in a simulation-based optimization of waterway project schedules. This PGA is intended to distribute a Genetic Algorithm application over several processors and thus speed up the solution procedure of a very large combinational problem, which may require several days of running time. The proposed PGA is based on a global parallel model, which is also called a master-slave model. A Message-Passing Interface (MPI) is used in developing the parallel computing program. The performance test results for a test problem show that adapting a simulation-based optimization algorithm to parallel computing is a promising way to reduce the computation time. Several additional techniques are found to further improve the PGA performance. These include: (1) choosing an appropriate task distribution method, (2) distributing simulation replications instead of different solutions, (3) avoiding the simulation of duplicate solutions, (4) avoiding running multiple simulations simultaneously in shared-memory processors, (5) avoiding using multiple processors which belong to different clusters (physical sub-networks).

01084478

08-2571

English

12p

2008

Transportation Research Board 87th Annual Meeting

DVD

Figures (5) ; References (28) ; Tables (4)

Genetic algorithms; Information management; Information processing; Network analysis (Planning); Optimization; Parallel processing; Scheduling; Simulation; Waterways

Administration and Management; Data and Information Technology; Marine Transportation

Transportation Research Board Annual Meeting 2008 Paper #08-2571

BTRIS, TRIS, TRB

Jan 29 2008 4:59PM