Hybrid evolutionary system with a local optimizer for manufacturing cell loading

H&barbelow;ybrid E&barbelow;volutionary S&barbelow;ystem with a Local Optimizer for Manufacturing C&barbelow;ell L&barbelow;oading (HES-CL) is a full Prolog++ system for manufacturing cell loading developed based on the paradigm of the evolution. This artificial system assigns jobs to cells and then finds processing sequence of jobs on each cell. This thesis focuses on the investigation of an emerging method to assign a given number of jobs to a given number of manufacturing cells. The objectives of cell loading considered are either minimizes the number of tardy jobs or minimizes the total tardiness. As a result, when nt is used, different approaches are proposed: (1) In a two-phase approach, first evolutionary programming is used to generate a job sequence and second a classical scheduling rule is used to assign jobs to the cells. (2) In a three-phase approach, a local optimizer step is added. (3) In a with-learning approach, the new population is modified based on local optimizer results. Experimentation shows that, for two step approach only, the way that due dates are established affect the results and that most of the chromosomes have the best fitness within first 40 generations for the cases considered. The number of cells and population size also affects the solution quality. HES-CL allows users to perform experimentation with different sizes of manufacturing problems including demand and due dates with diverse "genetic" operators, and different evolutionary parameter set.