A Data-Knowledge Hybrid Framework for Process Monitoring and Fault Detection with the Combined Use of Big Data

The main purpose of this study is the faster detection of faults. Advanced control knowledges has been adapted in chemical process monitoring during last decade. In this study, we attempted to apply state estimator (e.g., Kalman filter) design to non-steady state process to detect faults faster than statistical process control (SPC) only. Due to high nonlinearity in first principles of chemistry, the system is more complex and the reach time of the observer is late. Nevertheless, we expect to this framework to approximate small faults in complex systems. So, we introduce a hybrid methodology of state estimator design of nonlinear system and SPC. Then, we test the performance of the framework with comparison against the detection by check with given historical data of the process under presence of measurement noises.