ENHANCING THE BIBMON PYTHON PACKAGE: EARLY FAULT DETECTION USING AN INTEGRATED FRAMEWORK

Industries are consistently affected by unexpected process faults, which may lead to plant shutdown and economic losses. Therefore, integrating multiple tools into a unified framework is an effective strategy for achieving reliable monitoring. In this study, we present a complete fault detection and monitoring framework built using existing tools and developed under the BibMon package paradigm. The framework's reliability was evaluated using the Tennessee Eastman Process (TEP) benchmark for two fault types: a step disturbance in the feed ratio and gradual kinetic deterioration. Results show that the framework consistently detects early fault stages, with all methods supporting each other’s findings. This highlights the framework’s effectiveness in providing reliable predictions, meaningful interpretations, and effective visualization capabilities within a single integrated platform.