A Granular Genetic Algorithm for Machine Cell Formation

摘要

Cellular formation is a key issue of group technology for lot size production. Its purpose is to generate similar machines into a machine cell that can be employed to produce similar parts. In this research, the cellular formation problem is regarded as the clustering analysis for a two-dimensional data set on the part/machine incidence matrix. The problem itself is inherently associated with machine granules and machine granulation. In other words, it is to produce an appropriate number of granules that cover almost all of the data. Genetic algorithm (GA), a search method utilizing the principles of natural selection and genetics, is one of the most popular methods of evolutionary computation. Therefore, GA can be applied to searching a collection of boxes under the constraint of two limitations. The advantages of this technique are the simplicity of computation, the capability of handling a large-scale clustering analysis, and the lower possibility of falling into the local minimum. In this research, we attempt to solve the cellular formation problem using the integration of granule computation and genetic algorithm. The contributions may include the encoding of chromosome-like solutions and the definition of fitness function. Consequently, the test results show that our research method performs very well for the machine cell formation.