Data-driven CBM tool for risk-informed decision-making in an electric arc furnace

Abstract

Nowadays maintenance activities and safety management can be supported by a mature state of the art favouring the implementation of Condition Based Maintenance programme, which recommends maintenance decisions based on the information collected through asset life. The main idea, which grounds in the Industry 4.0 paradigm, is to utilize the asset degradation information, extracted and identified through different techniques, to reduce and eliminate costly, unscheduled downtimes and unexpected breakdowns and to avoid risky scenarios. This paper aims at developing and testing a data-driven CBM tool to provide fault diagnostics transforming raw data from the shop-floor into information, finally enabling risk-informed decision-making. The tool relies on a process of knowledge discovery that incorporates both prior knowledge and proper interpretation of data analytics results. Prior knowledge is extracted through a Process Hazard Analysis (PHA), while data analysis deals with Statistical Process Control and Novelty Detection. The model is proposed to integrate some Cyber-Physical System element in the extant plant automation, to exploit its computational capabilities through the continuous monitoring and data analytics. This enables a "watchdog agent" of risky scenario, allowing an on-line risk-assessment of safety-critical components, finally enhancing the intelligence in the industrial process.

A framework to integrate novelty detection and remaining useful life prediction in Industry 4.0-based manufacturing systems

“…Thus, every new RMS is labelled "healthy" if P(RMS) < yabnormal or "abnormal" otherwise, where yabnormal = P(RMS UP ) = P(μ + 3σ). This assumption is confirmed by previous work (Fumagalli et al 2019), and RMS UP can be described by Equation 5.…”

Exploring the impacts and contributions of maintenance function for sustainable manufacturing

“…Details on the environmental and social dimensions of maintenance complement extant approaches, such as lifecycle maintenance, emaintenance and intelligent prognostics tools (Takata et al 2004;Levrat et al 2008, Lee et al 2006. Our findings can be particularly relevant for the Prognostics and Health Management (PHM) body of knowledge in developing predictive maintenance approaches with overall benefits along the asset lifecycle (Sun et al 2012;Guillen et al 2016;Fumagalli et al 2019). More holistically, our findings can inspire smart / intelligent maintenance strategies to pursue TBL outcomes in an explicit manner, thus strengthening the sustainability agenda in advances towards smart production systems.…”

AI and BD in Process Industry: A Literature Review with an Operational Perspective

“…NLP is used for entity recognition for the steel product categories; while for OSR, vision-based automatic identification can track without embedding identification codes onto the steel product surfaces [11]. CPS and knowledge-based systems are used to detect functional failure to reduce the time of expertise acquisition and the cost of solving over-generalization and over-fitting problems with a data driven tool to provide fault diagnostics enabling risk-informed decision-making [12].…”