電力線電流センシングによる設備の稼動状態推定技術の開発

抄録

This report describes the development of high-precision equipment state identification technology can be mounted easily to existing facilities. The purpose is a state estimation of machine tools in high-mix low-volume factory. Achieve the objectives, we propose technique a clustering of the feature by unsupervised learning, and a labeling scheme based on prior knowledge. Attempt several clustering technique, it found that spectral clustering is suitable to use a graph based on the similarity between samples. Also, we developed a labeling scheme based on prior knowledge, the accuracy rate of the operating state estimation has achieved 94% in the general-purpose lathe, 96.3% in the drilling machine.