A Time-Series Process Event Log Preprocessing Approach for Data-Intensive and Predictive Operationalization of Smart Factories

Minjae Park, Sohyang Park, Thanh-Hai Nguyen, Dinh-Lam Pham, Dongkeun Oh, Kyoung-Sook Kim, In-Kyoo Chun, Kwanghoon Pio Kim

Abstract: This paper proposes a time-series process event log preprocessing approach applied to realize data-intensive and predictive operationalization of IoT-supported smart factories. The advanced conceptual approach of the data-intensive and predictive operationalization processes is newly proposed for effectively and predictively operating very large-scale smart factories that are supported by the Internet of Things platforms with temporal operationalization processes and their time-series event log datasets. The authors’ research group is trying to develop a predictive smart factory operationalization system, which is characterized by the advanced conceptual approach, and the starting point of which is at acquiring the time-series process event log datasets with a proper data preprocessing approach. Hence, this paper devises a time-series process event log preprocessing approach with its algorithmic details to arrange the proper datasets to be used for concretizing the data-intensive and predictive operationalization in the very large-scale smart factories. Finally, we develop an XES-formatted event log preprocessing system based upon the devised preprocessing algorithms and apply to a real and nation-wide production smart factory in order to validate the operationality and feasibility of the developed system.