03084nas|a2200253 a 450000500170000000800410001703500150005810000230007324501110009626000670020730000180027449000280029252018870032050001790220750200580238665000230244465000440246770000440251170000470255570000450260271000570264781000600270485600660276420260817161528.0110110s2009    th uu|m rtt   0|  a1eng d  a.b120883530 aSuntaree Sae-huere13aAn adaptive particle swarm optimization algorithm for a multicommodity distribution network design problem  aPathum Thani, Thailand :bAsian Institute of Technology,c2009  a54 p. :bill.1 aThesis ;vno. ISE-09-07  aThis thesis studies a multicommodity distribution network design problem (MDNP) in the supply   chain   that   involves   locations   of   plants   and   distribution   centers   (DCs),   and   determining the best strategy to distribute the products in a distribution network. The goal of  the  model  is  to  select  the  number,  size  and  the  location  of  plants  and  DCs  in  order  to  minimize  the  total  relevant  costs.  To  be  more  applicable  in  the  industry,  a  model  is  formulated with the distance limitation constraint and the multi-capacity level availability for  the  facilities  to  supply  each  type  of  the  products  in  each  candidate  plant  and  to  store  group products in each candidate DC.  An  adaptive  Particle  Swarm  Optimization  algorithm  is  applied  to  solve  the  problem.The parameters  of  particle  swarm  optimization  to  be  adapted  include  inertia  weight  and  acceleration  constants.  The  algorithm  is  evaluated  by  using  the  benchmark  problems  provided by Vinaipanit (2006) and some additional randomly generated test problems. The solutions  are  compared  with  the  solution  from  the  commercial  software  package  LINGO,  GA (Vinaipanit), and GLNPSO without adaptive feature in order to verify the performance of the proposed algorithm.  The  results  show  that  the  proposed  algorithm  can  solve  the  problem  and  performs  well  with  the  percentage  of  {u0BCC}{u0BEB}{u04A7}about  1.6  and  obtain  the  better  solution  than  GA  in  small  and  medium size. For large size problem, the solution is slightly inferior due to the condition of experiment  that  is  the  difference  of  population  and  iteration.  Moreover,  the  quality  of  results  from  adaptive  GLNPSO  is  better  than  non  adaptive  GLNPSO  with  the  same  parameters setting and significant level of 0.05.    aSubmitted in partial fulfillment of the requirements for the degree of Master of Engineering in Industrial and Manufacturing Engineering, School of Engineering and Technology  aThesis (M.Eng.) - Asian Institute of Technology, 2009 0aComputer algoritms 0aBusiness logisticsxMathematical models0 aVoratas Kachitvichyanukul,eChairperson1 aHuynh, Trung Luong,eExamination Committee1 aGong, Dah Chuan, eExamination committee2 aRoyal Thai Goverment Fellowship, eScholarship donor2 aAsian Institute of Technology.tThesis ;vno. ISE-09-07  3Full-Textuhttp://203.159.5.9/ait-thesis/detail.php?q=B03250