02796nam a2200241   450000500170000000800410001703500150005810000200007324500680009326000520016130000220021349000270023550200580026250001510032052017430047165000390221470000330225370000420228670000480232871000540237681000590243085600650248920260817163248.0030798         th                  eng    a.b114253251 aMohan, A. Madhu10aART1 neural networks for machine cell and part family formation  aBangkok :bAsian Institute of Technology,c1995  a71 leaves :bill.1 aThesis ;vno. IE-95-02  aThesis (M.Eng.) - Asian Institute of Technology, 1995  aA thesis report submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Engineering and Technology  aAn increasingly competitive and changing environment and ever-changing customer preferences have compelled manufacturers to look for ways to increase productivity. One area under examination is design of a production system based on Group Technology. Group Technology is a modern manufacturing philosophy that advocates the 'product organization' as against the 'production organization' in conventional systems. A more practical and effective approach for implementing GT is the Cluster Analysis approach. This study looks into the application of a Neural Network for forming machine and part family formation. Unlike the non-learning algorithms, the neural models are able to learn and store the learned patterns. For the classification problem, ARTl neural network was used as it is flexible and unsupervised. The input to the neural network is the machine-component incidence matrix is based on the production flow analysis. A computer program in TURBO C is used for executing the algorithm. The experiments are carried out on eight sets of data taken from literature. Initially the original similarity criteria designed for ARTl by its originators was used. Later, another criteria based on Hamming distance was incorporated. In most of the cases the results were same as that of the earlier reference works. From the results obtained, it was clear that ARTl is effective in cluster identification. The actual clustering depends on the way of presentation of input patterns. Even though different clusters were identified there was not significant change in the performance indices. Hence, it can be used as an effective tool in the Computer Integrated Manufacturing environment, as it gives flexibility for the machine cells formed.10aNeural networks (Computer science)1 aNagarur, N. N.,eChairperson1 aPandey, P. C.,eExamination committee1 aTabucanon, Mario T.,eExamination Committee2 aAsian Institute of Technology,eScholarship donor2 aAsian Institute of Technology.tThesis ;vno. IE-95-02  3Full-Textuhttp://203.159.5.9/ait-thesis/detail.php?q=B15202