An improved neural network leader algorithm for part-machine grouping in group technology

Kaparthi, Shashidhar; Suresh, Nallan C.; Cerveny, Robert P.

doi:10.1016/0377-2217(93)90020-n

Cited by 49 publications

(14 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They found that their model could identify groups, even for large data sets, relatively quickly with a good degree of perfection. In their subsequent work, Kaparthi and Suresh (1993) showed that the performance of a basic ART model could be improved signi®cantly merely by reversing the zeroes and ones. Venugopal and Narendran (1994) applied a competitive learning model, a adaptive resonance theory (ART) model and a self-organizing feature map (SOFM) model for the machine cell formation problem and compared results with ZODIAC algorithm.…”

Section: Incremental Cell Formationmentioning

confidence: 99%

Incremental cell formation considering alternative machines

Mahesh¹,

Srinivasan²

2002

International Journal of Production Research

View full text Add to dashboard Cite

Many algorithms for cell formation have been developed for past three decades in cellular manufacturing. Some use binary data for cell formation and others use production data such as operation sequence, processing times, production volumes, etc. for cell formation. All these algorithms assume that the conversion of job shop to cellular manufacturing is performed comprehensively. (In other words, they assume that all the cells are formed at a time.) However, this is far from reality. In practice, cell formation is done incrementally, one after the other, rather than comprehensively. None of the algorithms developed so far addresses the issue of incremental cell formation. In this paper, the incremental cell formation problem is de®ned and various categories of problems are mentioned. One type of those categories is selected for solving. Two methods, namely the branch and bound technique and a heuristic based on a multistage programming approach, have been applied to solve the chosen problem. Data sets have been generated to compare these two methods in terms of quality of solution and demand on computational time. It has been found that the branch and bound technique gives a superior quality solution, but is computationally more demanding, where as heuristic based on a multistage programming approach is computationally far superior.

show abstract

Section: Incremental Cell Formationmentioning

confidence: 99%

Incremental cell formation considering alternative machines

Mahesh¹,

Srinivasan²

2002

International Journal of Production Research

View full text Add to dashboard Cite

show abstract

“…the zeros and ones were reversed in the part± machine incidence matrix. Dagli and Sen (1992) Kaparthi et al (1993) , the category proliferation of ART was addressed. It was shown that with a reversal of digits (zeros as ones, and vice versa), the solutions provided by ART were more robust.…”

Section: Sequence-dependent Clustering Of Parts and Machinesmentioning

confidence: 99%

“…In particular, ANNs have been shown to be e ective for clustering large, industry-size data sets (Dagli and Sen 1992, Kaparthi and Suresh 1992, Kaparthi et al 1993, and Suresh and Kaparthi 1994.…”

Section: Introductionmentioning

confidence: 99%

Sequence-dependent clustering of parts and machines: A Fuzzy ART neural network approach

Slomp¹,

Kaparthi²

1999

International Journal of Production Research

View full text Add to dashboard Cite

This study addresses the problem of identifying families of parts having a similar sequence of operations. This is a prerequisite for the implementation of cellular manufacturing, group technology, just-in-time manufacturing systems, and for streamlining material¯ows in general. A pattern recognition approach based on arti® cial neural networks is proposed, and it is shown that the Fuzzy ART neural network can be e ectively utilized for this application. First, a representation scheme for operation sequences is developed, followed by an illustrative example. A more comprehensive experimental veri® cation, based on the mixture-model approach is then performed to evaluate its performance. The experimental factors include size of the part± machine matrix, proportion of voids, proportion of exceptional elements, and vigilance threshold. It is shown that this neural network is e ective in identifying good clustering solutions, consistently and with relatively fast execution times.

show abstract

“…Kaparthi and Suresh [15] show that ART1 pattern recognition neural network has the capability of processing large size data. Kaparthi et al [16] proposed some modifications on ART1 to improve its performance. Suresh and Kaparthi [22] solved the cell formation problem using Fuzzy ART, and they compared the performance of the network with other ART family members.…”

Section: Introductionmentioning

confidence: 99%