Colonial Competitive Algorithm And Its Applications In Optimization Of Discrete Manufacturing System

This paper studies the Colonial Competitive Algorithm (CCA) and its applications inoptimization of discrete manufacturing systems. Colonial competitive clgorithm is a newlyemerged metaheuristic algorithm that draws it inspiration from the socio-political process ofimperialistic competition. It is a population based metaheuristic algorithm that consists ofsome main steps including empires construction, assimilation, imperialistic competition andelimination. Colonial competitive algorithm is applied to solve a number of key problems indiscrete manufacturing systems in this paper, including single row facility layout problem,process planning, integrated process planning and scheduling, and mixed-model U-linebalancing and sequencing problem.(1) Single row facility layout problem (SRFLP) is known as NP-hard and involvesarranging a number of facilities on a straight line to minimize transportation costs amongfacilities. CCA is employed to tackle this problem and details of applying CCA to SRFLP areelaborated. The quality of CCA is evaluated on a number of SRFLP instances (30small-sizeinstances and20large-size instances) from literature. For small-size instances with knownoptimal solutions, CCA is able to find all the optimal solutions. In addition, for large-sizeinstances with no proven optimal solutions, CCA succeeded in improving eight previouslyknown results.(2) In this paper, we investigate the optimization of process planning in which variousflexibilities are considered. The objective is to minimize the total weighted sum ofmanufacturing cost. Process planning is strongly NP-hard due to the existence of variousflexibilities as well as complex machining precedence constraints. To tackle this problem,CCA is employed to find promising solutions with reasonable computational cost.Computational experiments on three sets of process planning problems taken from literatureare carried out, and comparisons with some existing algorithms are presented. The resultsshow that the proposed algorithm outperforms existing algorithms like genetic algorithm,simulated annealing, tabu search and particle swarm optimization.(3) Effective performance of modern manufacturing systems requires integrating of processplanning and scheduling more tightly, which is consistently challenged by the intrinsicintractability of these two problems. Traditionally, these two problems are treated sequentiallyor separately. Integration of process planning and scheduling (IPPS) provides a valuableapproach to improve system performance. However, it is more complex than job shopscheduling or process planning. With respect to its NP-hardness, CCA is proposed to addressthe IPPS problem with an objective of makespan minimization. Performance of CCA is evaluated on four sets of experiments taken from literature. Computational results of CCA arecompared with that of some existing algorithms developed for IPPS, which validates theefficiency and effectiveness of the CCA in solving IPPS.(4) Implementation of mixed-model U-shaped assembly line is emerging and thriving inmodern manufacturing systems due to adaptation to changes in market demand andapplication of just-in-time production principles. In this paper, the line balancing and modelsequencing problems in mixed-model U-line are considered simultaneously, which results inthe NP-hard mixed-model U-line balancing and sequencing problem (MMUL/BS). CCA isdeveloped and modified to solve the MMUL/BS problem. The modified CCA (MCCA)improves performance of original CCA by introducing a third type of country, independentcountry, to the population of countries maintained by CCA. Implementation details of theproposed CCA and MCCA are elaborated using an illustrative example. Performance of theproposed algorithm is tested on a set of test-bed problems and compared with that of existingalgorithms such as co-evolutionary algorithm, endosymbiotic evolutionary algorithm,simulated annealing and genetic algorithm. Computational results and comparisons show thatthe proposed algorithms can improve the results obtained by existing algorithms developedfor MMUL/BS.This paper concludes with summarization of the above work...