Abstract

    Open Access Research Article Article ID: TCSIT-1-101

    Optimizing the Cross Section of Cold-Rolled Steel Beams Using a Genetic Algorithm: Avoiding Local Optima Using Adaptive Mutation Control, Flexible Restriction Handling and Inbreed Avoiding Mating Strategies

    Werner Baumgartner*, Florian Esterhammer, Christoph Wolf, Anna Theresia Stadler

    In modern mechanical engineering and steelwork the use of cold-rolled steel sections is a standard method. These sections should be mechanically stable on the one hand and cost efficient on the other hand. To decide what profile suits for a certain case is a constrained optimization problem which is in general non convex, i.e. several local optima exist.

    To solve this non trivial problem we used genetic algorithms, search heuristics that mimic the process of natural evolution. For the specific application some additional problems had to be solved:First, an adaptive mutation control was implemented. Second, a mixed asexual and sexual reproduction was applied with an inbreed avoiding method based on the genetic distance of the individuals. Third, the restrictions were handled flexible, dependent on the mutation strength. This means that under the conditions of strong mutations (r-strategy), violations of the restrictions are allowed within some limits corresponding to reduced evolutionary pressure. Later on when approaching an optimum and the algorithm changes eventually to K-strategy, the restrictions become more severe corresponding to stabilising selection.

    The presented algorithm was tested on some cases; we found that significant improvement of cost efficiency was reached while mechanical stability was still granted. In comparison to hard restriction implementations like constant penalty functions or Lagrange-multipliers due to the flexible restrictions the algorithm tends significantly less to sustain in local optima. This approach could help to find cost efficient and light weight steel structures for mechanical engineering in the near future.

    Keywords: Cold forming; Evolution; Restricted optimization; Penalty function; Cold-formed steel; Steelwork

    Published on: Apr 26, 2016 Pages: 1-11

    Full Text PDF Full Text HTML DOI: 10.17352/tcsit.000001
    CrossMark Publons Harvard Library HOLLIS Search IT Semantic Scholar Get Citation Base Search Scilit OAI-PMH ResearchGate Academic Microsoft GrowKudos Universite de Paris UW Libraries SJSU King Library SJSU King Library NUS Library McGill DET KGL BIBLiOTEK JCU Discovery Universidad De Lima WorldCat VU on WorldCat

    Indexing/Archiving

    Pinterest on TCSIT