APPLICATION OF 2D-3D CONTINUOUS ANT COLONY APPROACH TO PIPE DISTRIBUTION NETWORK DESIGN OPTIMIZATION PROBLEMS

摘要

Recently [1] presented a new heuristic optimizationrnapproach, called 2D-3D Continuous Ant ColonyrnApproach (2D-3D-CACA), based on Ant ColonyrnOptimization Algorithm for solving 3D continuous realrnworld space problems. Ant colony algorithms are a subsetrnof swarm intelligence and consider the ability of simplernants to solve complex problems by cooperation. Thernproposed approach has been successfully used to optimizernthe path generation, cost estimation and road constructionrnproblem in 3D real world search space and has capturedrnsuccessfully the essence of the 3D problem structure. Thisrnpaper presents an application of 2D-3D-CACA to solvernthe pipe distribution network optimization problem. To bernclear, all previous attempts have been focused on discreternnature of pipe distribution network, where networkrnelements positions are fixed and known in prior. Thisrnpaper provides an initial study for the use of 2D-3DCACArnin solving pipe distribution network problems inrn3D continuous search spaces. Herein, the positions ofrnnetwork elements are not known in advance, except somerninitial starting points. The optimization of piperndistribution networks designs are complex, multi-modalrnproblems that cannot be easily solved with conventionalrnoptimization algorithms. The problem is formulated as arnconstrained multiple-objective optimization problem. Onernexample of Pipe network optimization is internal waterrndistribution network design (WDND) optimization. Waterrndistribution network design (WDNDs) is costlyrninfrastructure, and much attention has been given to thernapplication of optimization methods to minimize designrncosts. Herein, the application of the 2D-3D-CACA tornwater network design optimization is considered. In thisrnpaper, a mathematical formulation is developed whichrnenables 2D-3D-CACA to be used for the optimal designrnof WDN and thus to obtain the least cost design of thernnetwork. The 2D-3D-CACA has been used to determinernthe minimum cost of the water network system taking intornconsideration some design criteria (pipes lengths, no. ofrnrounds,….).rnAlthough, the paper presents an initial study andrninvestigation of the possibilities offered by 2D-3D-CACArnto optimize pipe distribution network design, preliminaryrnsimulation results on three 3D artificial models show thatrnthe proposed evolutionary optimization algorithm isrnefficient and suitable for any generalized pipe distributionrnnetwork design in 3D continuous search space.