Unplanned downtime in industrial processes costs an estimated 50 billion dollars per year, 42% of which are caused by equipment failure [1]. There have been substantial investments in the digital modeling of industrial assets to improve their efficiency and reduce the long-time maintenance costs. Another reason for these investments has been the increasing lack of personnel in-depth knowledge and awareness of asset condition across organizations in all sectors, resulting in a general lack awareness of optimal equipment maintenance, upgrade or replacement [2]. Accurate first-principle models are sometimes difficult to obtain for complex high-dimensional processes [3], which has driven interest in the development of robust data-driven frameworks capable of understanding process dynamics using only data samples gathered from its variables. Data-driven process monitoring in industrial processes is a multi-stage framework that involves modeling, fault detection and fault classification. Fault detection is a critical part of process monitoring, where the objective is to flag unexpected operating behavior quickly and accurately. In this paper, a novel extension of the Generalized Likelihood Ratio (GLR) chart is proposed, denoted as the Maximum Multivariate GLR (MMGLR) chart. Linear and nonlinear data-driven models, namely principal component analysis and Bayesian-optimized neural networks [4], are combined with different statistical charts towards the detection of multiple fault types in chemical processes. The results show that the MMGLR charts have a better detection accuracy than conventional charts, and that neural networks are more robust models and fault detectors than PCA and its extensions.
Keywords: Data-driven Fault Detection, Generalized Likelihood Ratio, Neural Networks.
References
[1] IndustryWeek in collaboration with Emerson,“How manufacturers can achieve top quartile performance,” 2023. [Online]. Available: https://partners.wsj.com.
[2] V. Bourne, Human error is more common cause of unplanned downtime in manufacturing than any other sector (Nov 2017). [Online]. Available: https://www.businesswire.com/news/home/ 20171106006370/en/Human-Erroris-More-Common-Cause-of-Unplanned-Downtime-in-Manufacturing-Than-Any-Other-Sector-According-to-New-Research
[3] O. J. Fisher, N. J. Watson, J. E. Escrig, R. Witt, L. Porcu, D. Bacon, M. Rigley, R. L. Gomes, Considerations, challenges and opportunities when developing data-driven models for process manufacturing systems, Computers & Chemical Engineering 140 (2020) 106881. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S0098135419308373
[4] N. Basha, G. Ibrahim, H.A. Choudhury, M. Challiwala, R. Fezai, B. Malluhi, H. Nounou, N. Elbashir, and M. Nounou, "Bayesian-optimized Neural Networks and their application to model gas-to-liquid plants," Gas Science and Engineering, Vol. 113, Article 204964, 2023, https://doi.org/10.1016/j.jgsce.2023.204964.
