Machine-component cell formation in group technology: a neural network approach

Abstract

This paper presents a neural network clustering method for the part-machine grouping problem in group technology. Among the several neural networks, a Carpenter-Grossberg network isselecteddue to the fact that this clustering method utilizesbinary-valued inputs and it can be trained without supervision. It is shown that this adaptive leader algorithm offersthe capability of handling large, industrysizedata sets due to the computational efficiency. The algorithm was tested on three data sets from prior literature, and solutions obtained werefound to result in block diagonal forms. Some solutions were also found to be identical to solutions presented by others. Experimentson larger data sets, involving 10000 parts by 100 machine types,revealedthat the method results in the identificationof clusters with fast execution times. If a block diagonal structure existed in the input data, it was identifiedto a good degreeof perfection. It was also found to be efficient with some imperfections in the data.

Genetic design of cellular manufacturing systems

“…Recently, the application of artificial intelligence has received considerable attention as an alternative to solving the problem. Some interesting papers in this area include the application of expert systems (Kusiak, 1988), simulated annealing (Boctor, 1991), fuzzy logic (Chu & Hayya, 1991), and neural network (Kaparthi & Suresh, 1992). In this article, an approach based on the concepts derived from genetic algorithms (Goldberg, 1989;Holland, 1975) is proposed as a general methodology to solve the machine-part cell formation problem.…”

An Integrated Approach of Art1 and Tabu Search to Solve Cell Formation Problems

“…Dagli and Huggahalli [12] suggested the input vectors be presented to ART1 in descending order of the number of l's in each vector. Kaparthi and Suresh [18] suggested that the binary entries in the machine-part matrix be reversed. The density of 1's in the machine-part matrix for a cell formation problem is generally low.…”

“…This process is repeated for all the input vectors. The ARTI algorithm is outlined below ( [11] and [18]). …”

Machine-part grouping and cluster analysis: similarities, distances and grouping criteria

“…Solving of the problems obtained with classical methods of optimisation, though, turned out to be too difficult. The next stage, still underway today, consisted in application of metaheuristics: tabu search, simulated annealing [15,16]; genetic algorithms [17][18][19]; simulated neural networks [8,[20][21][22], or [23] (fuzzy neural networks); or ant colony (see [24] for the facility layout problem). Regarding the methods used to solve the facility layout problem, mentioned before, it is worth noting that quite a similar reasoning occurred, see, e.g.…”