MULTI-CRITERIA OPTIMIZATION OF PROCESS PLANS FOR RECONFIGURABLE MANUFACTURING SYSTEMS: AN EVOLUTIONARY APPROACH

摘要

Production systems have developed over the years due to changing environment, external and internal drivers and conditions like new technologies, developed products and customer needs. These needs were the main drivers for integrated and evolved manufacturing systems which can be more responsive and customer focused. Prototypes of manufacturing industries which have been recently introduced as flexible and reconfigurable manufacturing systems are responding to these recent needs in peculiar ways focusing not only on product design level but also on integrated manufacturing systems and process planning level. In this work, a methodology will be presented to solve the NP-Hard problem of process planning through evolutionary optimization using Genetic Algorithms (GA) to generate and then find the optimized process plan for a part or a part family. For the creation of initial population without violating the logical or geometrical constraints, a ranked matrix based on precedence would be developed that will called as precedence group matrix (PGM). Fitness will be evaluated on the basis of setup and tool change matrices thus, making it a combinatorial optimization problem. Tool approach direction (TAD) will be assigned to each operation for generation of setup change matrix. Genetic operators like crossover, mutation and selection will be revised in order to maintain geometrical and logical constraints that come across in machining of the part. Position wise exchange will be used for crossover. A novel strategy has been proposed to check the conformance of new solution after mutation. To avoid the loss of good solutions with higher fitness value, elitist model has been proposed for selection purposes. Furthermore, a technique will be presented in order to achieve reconfigurability and responsiveness to accommodate new features in the already generated process plan, thus creating a hybrid between generative and variant process planning approach.