Combining ant colony optimization with 1-opt local search method for solving constrained forest transportation planning problems

Pengpeng Lin,Ruxin Dai,M. Contreras,Jun Zhang
DOI: https://doi.org/10.5430/air.v6n2p27
2017-04-04
Artificial Intelligence Research
Abstract:We developed a two-stage approach (ACOLS) combining the ant colony optimization (ACO) algorithm and a 1-opt local search to solve forest transportation planning problems (FTPPs) considering fixed and variables costs and sediment yields expected to erode from road surfaces as side constraints. The ACOLS was designed for improving ACO performance and ensure the applicability to real-world, large-scale FTPPs with multiple time periods. It consists of three major routines: i) least-cost route finding process from all timber sales simultaneously, ii) two stage search process developed to quickly find feasible (stage I) and high-quality (stage II) solutions and, iii) 1-opt local search solution refinement to further improve solution quality. The ACOLS was first applied to a medium-scale hypothetical FTPP on which four cases with increasing level of sediment constraint were considered. To test for robustness, the ACOLS was then applied to ten different problems instances created basing on the same topology of the hypothetical FTPP. Lastly, the ACOLS was applied to a real-world, large-scale FTPP considering thousands of roads segments, hundreds of timber sales, and multiple products and planning periods. Feasible solutions were found for all cases indicating the usefulness of our approach to provide managers with an efficient tool to address large-scale transportation problems.
What problem does this paper attempt to address?