Data-driven approach to predict the sequence of component failures: a framework and a case study on a process industry

Purpose The research approach is based on the concept that a failure event rarely random and often generated by chain of previous events connected sort domino effect. Thus, purpose this study optimal selection components to predictively maintain basis their probability, under budget time constraints. Design/methodology/approach Assets maintenance major challenge for any process industry. Thanks development Big Data Analytics techniques tools, data produced such systems can be analyzed in order predict behavior. Considering asset as social system composed several interacting components, work, framework developed identify relationships between component failures avoid them through predictive replacement critical ones: are identified Association Rule Mining (ARM), while interaction studied Social Network Analysis (SNA). Findings A case example industry presented explain test proposed model discuss its applicability. provides an expand upon work areas prediction fault monitoring strategy components. Originality/value novel combined adoption ARM SNA hidden among define nature interactions communities nodes analyze local global paths most influential entities.