Predictive Analytics Model for Power Consumption in Manufacturing

A Smart Manufacturing (SM) system should be capable of handling high volume data, processing high velocity data and manipulating high variety data. Big data analytics can enable timely and accurate insights using machine learning and predictive analytics to make better decisions. The objective of this paper is to present big data analytics modeling in the metal cutting industry. This paper includes: 1) identification of manufacturing data to be analyzed, 2) design of a functional architecture for deriving analytic models, and 3) design of an analytic model to predict a sustainability performance especially power consumption, using the big data infrastructure. A prototype system has been developed for this proof-of-concept, using open platform solutions including MapReduce, Hadoop Distributed File System (HDFS), and a machine-learning tool. To derive a cause-effect relationship of the analytic model, STEP-NC (a standard that enables the exchange of design- to-manufacturing data, especially machining) plan data and MTConnect machine monitoring data are used for a cause factor and an effect factor, respectively.