A Configuration Space Evolutionary Algorithm For The Multi-Objective Dynamic And Static Facility Layout Problem

With n facilities and a bounded layout provided with give shape and size,the facility layout problem(FLP)is an NP-hard problem that is to research how best to place these facilities in the bounded layout without overlaps reasonably,and to meet the requirements of optimization objectives.In this paper,under the background of the layout in enterprise production,different constraints handing methods are proposed to optimize the layout according to the static and dynamic facility layout.We designed the intelligent optimization algorithm to solve multi-objective problem.The concrete research contents and results are as follows:(1)The configuration space evolution algorithm(CSE)for multi-objective static facility layout problem(MO-SFLP)is studied.Based on FBS,the problem is transformed into unconstrained optimization problem,and a mathematical optimization model is established.The CSE introduces the concept of "configuration bank",and constructsthe circular region with each configuration as the center and half of the average distance of all configurations in the configuration bank(dspace=davg/2).The convergence and diversity of solutions in the CSE is controlled by the space distance dspace,whose value is slowly reduced to narrow the search space.Also,a method of the nearest and farthest candidate solution based on the distance of objective function(NFCS)is combined with the non-dominated sorting to choose the Pareto-optimal solutions,which is able to get a good spread in the Pareto front and maintain diversity of the obtained solutions.Ten representative instances from the literature are tested,and the performance of the algorithm was analyzed.(2)The configuration space evolution algorithm for multi-objective dynamic facility layout problem(MO-DFLP)is studied.According to the characteristics of the dynamic facility layout problem is multi-stage and facilities length and width are fixed,facilities and the layout are supposed to be elastic objects.Introducing the elastic potential function,we put forward CSE-GM.we use adaptive gradient method of the embedded facilities to satisfy interferential constraints,and we redefine the encoding method of chromosome,the evolutionary strategy of population and the method of selecting optimal solution.To verify the validity of CSE-GM,two sets of dynamic classical examples are designed and an actual exampleis presented.Experimental results show that the proposed algorithm is an effective algorithm to solve MO-DFLP.