Research On Algorithms And Applications For Two Kinds Of Facility Layout Problems By Using Modified Estimation Of Distribution Algorithm

With the increasing concern on the internal logistics,optimization for the internal logistics facility layout problem(FLP)is crucial to enhance operation efficiency so as to induce higher market competitiveness for enterprises.As a classical combinatorial optimization problem,decision making for FLP depends on the decision objectives,characteristics of facilities,material handling processes,physical conditions of production space,leading to the large-scale,nonlinear,and strong-constraint characteristics.As already proved,FLP can be classified as a NP-hard problem in the combinatorial optimization and its optimization process is highly difficult.Therefore,it possesses both theoretical significances and practical values to establish mathematical models and develop optimization algorithms for FLPs according to the layout type and features.Estimation of distribution algorithm(EDA)is one of the widely used probability-model-based intelligent algorithms,which has excellent global search ability.Based on these mentioned above,this paper focuses on two kinds of FLPs,i.e.,quadratic assignment problem(QAP)and fixed-positioning layout problem(FPLP),which are widely used in practical environments.Then,the modified EDAs are designed to solve the problems.The main works in this paper are listed as follows:(1)The researches on the QAP.Firstly,the basic theories and mathematical models of QAP have been reviewed,and then a hybrid EDA(HEDA)is presented to solving it.Secondly,based on the distributed features of both logistics matrix and distance matrix of QAP,a heuristic rule is designed with the aim to generate the initial population of HEDA.Thirdly,a local search operator based on interchange neighborhood structure and speed-up evaluation is raised to enhance the local exploitation capa’bility.Consequently,the computation complexity can be decreased by reducing some invalid and roundabout search during iteration of HEDA.(2)The researches on the FPLP.According to literature review,relevant studies concerned with FPLP are still unexplored,and this is the first work that takes FPLP as an independent type of FLPs.To solve it,an improved EDA(IEDA)is presented.Firstly,the basic nonlinear integer programming model of FPLP is established.Secondly,based on the proposed basic mathematical model,a penalty-function-based improved mathematical model is established via relaxing the capacity constraints.Thereafter,the new objective function is utilized to guide the search process of IEDA.(3)The effectiveness validation of mathematical model and solution methods.With the aim to demonstrate the performance and feasibility of the algorithms and models,simulation experiments and comparisons are conducted.All algorithms are encoded and executed in the integrated development environment(IDE)of Microsoft Visual Studio 2008(C++),and numerical results show the superiority of the proposed algorithms to the existing well-known algorithms in both solution quality and feasibility criterion.(4)The researches in potential application values.The proposed mathematical models as well as solution methods(HEDA and IEDA)are successfully applied to deal with the logistics facility QAP for a logistics park and the FPLP for a large construction site respectively,which reveal the potential practical values of the presented mathematical model and algorithms.From the researches in this paper,we can get the conclusions as follows:(1)Through the experiment results,we can find that the ED As have an excellent performance and robustness in optimizing the FLPs,which indicate the iterative process based on the probability matrix for EDA matches the problem structure of FLPs well.Furthermore,it demonstrates the importance of utilization for the problem characteristics information(i.e.knowledge).(2)The original frameworks of intelligence algorithm are unable to deal with the hard constraints,so embracing proper constraint handling technologies plays importantly to improve solution quality.(3)Although EDA is one of the novel intelligence algorithms in evolutionary computing area,researches on its framework are limited.Compared with other intelligence algorithms,EDA has better searching ability and sensitiveness in some certain problems and can deal with optimizations under a complex real-life system,so EDA has a broad application prospect.