Optimization for multi-part flow-line configurations of reconfigurable manufacturing systems based on genetic algorithm

Obtaining the optimal and K-1 near-optimal (K-best) Multi-Part Flow-Line (MPFL) configurations as candidates for each demand period was an important optimization problem for reconfigurable manufacturing system in the operational phases. Given the operation precedence graph for each part, relationship between sequence and Operation Setups (OSs) as well as machine options for each OS, the problem was to determine the MPFL configuration's parameters in order to minimize capital cost of MPFL configurations. The parameters included number of workstations, number of parallel machines and machine type as well as assigned OSs for each workstation. To generate K-best MPFL configurations, firstly a generic 0-1 NonLinear Programming (NLP) model which widened the solution space was developed by relaxing the limitation of the assignment of OSs in existing models. Then, a Feasible OS Assignment Oriented Generation Algorithm (FOAOGA) was proposed to efficiently find K-best MPFL configurations from the solution space of the 0-1 NLP model. A case study showed that the optimum found by FOAOGA was better than the optimum obtained by existing approach, and also demonstrated the effectiveness of the proposed model.