| dc.description.abstract | With industry 4.0, a new era of the industrial revolution with a focus on automation, inter-connectivity, machine learning, and real-time data collection and analysis are emerging. The smart digital technology which includes smart sensors, data acquisition, processing, and control based on big data, machine learning, and Artificial Intelligence (AI) provides boundless opportunities for the end-users to operate their plants under more optimized, reliable, and safer conditions. During an abnormal event in an industrial facility, operators are inundated with information to infer and act. Hence, there is a critical need to develop solutions that assist operators during such critical events. Also, because of the obsolescence challenges of typical industrial control systems, a new paradigm of Open Process Automation (OPA) is emerging. OPA requires a Real-time Operational Technology (OT) services to analyze the data generated by the sensors and control loops to assist the process plant operations by developing applications for advanced computing platforms in open source software platforms.
The aim of this research is to highlight the potential applications of big data analytics, machine learning, and AI methods and develop solutions for plant operation, maintenance, process safety and risk management for real industry problems. This research work includes:
1. an alarm management framework integrated with data-driven (Key Performance Indicators) KPIs bench-marking, and a visualization tool is developed to address alarm management challenges;
2. a deep learning-based data-driven process fault detection and diagnosis method on cloud computing to identify abnormal process conditions; and
3. applications such as predictive maintenance, dynamic risk mapping, incident database analysis, application of Natural Language Processing (NLP) for text classification, and barrier assessment for dynamic risk mapping,
A unified workflow approach is used to define the data-sources, applicable domains, and develop proposed applications. This work integrates data generated by field instrumentation, expert knowledge with data analytics and AI techniques to provide guidance to the operator or engineer to effectively take proactive decisions through “action-boards”.
The robustness of the developed methods and algorithms is validated using real and simulated data sets. The proposed methods and results provide a future road map for any organization to deal with data integration with such applications leading to productive, safer and more reliable operations. | en |
