Modified ART1 neural networks for cell formation using production data

Abstract

In the present work, an attempt has been made to form disjoint machine cells using modified ART1 (Adaptive Resonance Theory) to handle the real valued workload matrix. The methodology first allocates the machines to various machine cells and then parts are assigned to those cells with the aid of degree of belongingness through a membership index. The proposed algorithm uses a supplementary procedure to effectively take care of the problem of generating cells with single machine that may be encountered at times. A modified grouping efficiency (MGE) is proposed to measure the performance of the clustering algorithm. The results of modified ART1 algorithm are compared with the results obtained from K-means clustering and genetic algorithm. The modified ART1 results are also compared with the literature results in terms of number of exceptional elements. The performance of the proposed algorithm is tested with genetic algorithm and K-means clustering algorithm. The results distinctly indicate that the proposed algorithm is quite flexible, fast and efficient in computation for cell formation problems and can be applied in industries with convenience.