Toward big data analytics for smart manufacturing:

A case of machining experiment

抄録

Big data analytics, an essential constituent of smart manufacturing, fuels decision-making by extracting knowledge underlying big data. However, the existing analytics frameworks are heavily resource-dependent and computationally heavy. This creates a digital division in accessing and analyzing the data, known as big data inequality. As such, although large enterprises can utilize big data, small and medium-sized companies fall behind. Moreover, the frameworks often result in black box systems due to a lack of transparency, engagement of human intelligence, and integration of externally generated data. This study sheds some light on developing big data and relevant analytics, mitigating the abovementioned limitations. In particular, this study first articulates the limitations and alternatives in detail. It then elucidates how manufacturing big data can be constructed and made widely accessible by creating digital manufacturing commons. A big data analytics architecture is also proposed to extract knowledge from big data. The analytics consists of four integrated yet independent and modular sub-systems: data exploration system, data visualization system, data analysis system, and knowledge extraction system. The functionalities of each system are described. The sub-system-based modular architecture ensures transparency of the inner processes and engages human intelligence straightforwardly. Finally, considering a real-life machining experiment, this study demonstrates the big data construction process and knowledge extraction from the constructed big data using a developed data visualization tool.