放電加工状態分類システムの開発 : ニューラルネットワークを利用した認識システムの最適化

概要

論文の詳細を見る
There have been recent reports made on a pattern recognition system using a neural network. However, difficult problems, such as how big a neural network should be adopted and which feature values for input of the neural network should be selected, have existed in the development of the recognition system. In this study, a new optimizing procedure for the neural network recognition system is proposed. Each output value for the nodes of the neural network after learning is used in the optimizing procedure. An optimum recognition system using a neural network for electrical discharge machining is developed as a good example of its application. The effectiveness of the optimum recognition system is proven to be practical in the electrical discharge machining process.
社団法人日本機械学会の論文
1993-05-25