Division of Informatics
Forrest Hill & 80 South Bridge

MSc Thesis #9538
Title:	Automated Pipe Route Generation Using Genetic Algorithms
Authors:	Kim,D
Date:	1995
Presented:
Keywords:
Abstract:	This study is about pipe route design using Genetic Algorithms (GAs). Pipe route design work, which is usually done manually in industry, was able to be solved by a computer program. The pipe route generation process is defined as a kind of optimisation problem, based on viewing it as a variant of the minimal rectilinear Steiner tree problem and four different search techniques were used for comparison. GAs showed a better performance than Simulated Annealing (SA), Random Search (RS), and Stochastic Hill Climbing (SHC). SA showed the second best performance of the four. GAs worked better with a large number of loosely interacting population sets. Using SA was not so simple because of the difficulty of parameter tuning, but even the best parameter set in this study did not show as good a result as GAs. Interestingly, on some problems, the performance of SHC was worse than of RS. Several reluctant topics are discussed: the multiple objective optimisation problem when allowing multiple pipe diameter in the system, extension of this work to three dimensions, and combining the pipe route generation program with existing CAD packages.
Download:	NO ONLINE COPY

Division of InformaticsForrest Hill & 80 South Bridge