Identifying Dependency Relationships Between Events in Production Systems

Abstract

Industry 4.0 is directly related to the digital transformation of organizations. One of the main characteristics of this transformation is the ability to analyze and manage data to obtain relevant and previously unknown information from it. Specifically, in the context of operational risk management, an efficient analysis of the data that typically come from occurrence records, can support decision-making. This paper presents a new algorithm to identify dependencies between events in productive environments, such as equipment failures, defects, or safety and environmental incidents, supporting the decision-making process in the context of analyzing and managing operational risks. The algorithm was developed based on association rule mining algorithms, then tested with data from a Portuguese metalworking company, and finally validated. The test results show that the developed algorithm is capable of identifying dependencies, which have to be confirmed by specialized staff to conclude if they are causal relationships.