Simulation-optimization with machine learning applied to production planning and scheduling - case: semiconductor manufacturing

Industry 4.0, which may eventually represent a fourth industrial revolution, is a complex technological system that has been widely discussed and researched, hav ing a great influence in industrial, since it introduces relevant advancements that are related with smart and future factories. Production Planning and Scheduling (PP&S) paradigms within industries are one of the main sectors influenced by Indus try 4.0 advancements. Undoubtedly, capacity allocation and production scheduling are important aspects in PP&S to be studied by researchers. From a globalized per spective, semiconductor manufacturing is one of the main contributors to support the Industry 4.0 era. Thus, in the first phase of this research, a new Mixed Integer Linear Programming (MILP) model for a Capacity Allocation Problem in a Pho tolithography Area (CAPPA) is proposed. To solve CAPPA, an improved Genetic Algorithm (GA) named Improved Reference Group GA (IRGGA) is used to solve CAPPA efficiently by improving the generation of the initial population. In the second phase, a novel metamodeling approach is proposed to metamodel a Discrete Event Simulation Model (DESM) of a Stochastic Job Shop Scheduling Problem (SJSSP). Finally, the designed metamodeling approach is integrated within an evo lutionary Simulation Optimization (SO) method called an Evolutionary Learning Based Simulation Optimization (ELBSO) algorithm. Using a comprehensive exper imental analysis this algorithm is examined and proof of its superiority is provided