Warning of abnormal conditions of catalytic hydrogenation reaction in petroleum refining based on multi-level threshold

Abnormal conditions in petrochemical operations are often diagnosed manually or semi-manually, but this method is often inadequate and ineffective when facing increasingly complex operating environments. Therefore, a warning of abnormal conditions in the catalytic hydrogenation reaction process in petroleum refining based on multi-level threshold is proposed. For the stationary process, the multi-level threshold is refined, and the monitoring of a multi-level fixed threshold detection is established. For non-stationary processes, the time-varying threshold detection problem is transformed into a time-invariant threshold detection problem for stationary objects by calculating the residuals of the fault-tolerant filtering, and to realise the abnormal working condition detection. The proposed method can monitor the abnormal conditions of catalytic hydrogenation reaction in petroleum refining under unsupervised conditions, and provide a theoretical basis for warning of abnormal conditions in the field. © 2025 Elsevier B.V., All rights reserved.