(509a) Online Monitoring and Robust, Reliable Fault Detection and Diagnosis of Chemical Process Systems

Nowadays, large amounts of data are continuously collected by sensors and monitored in chemical plants. Despite having access to large volumes of historical and online sensor data, industrial practitioners still face several challenges in effectively utilizing them to perform process monitoring and fault detection, because: 1) fault scenarios in chemical processes are naturally complex and cannot be exhaustively enumerated or predicted, 2) sensor measurements continuously produce massive arrays of high-dimensional big data streams that are often nonparametric and heterogeneous, and 3) the strict environmental, health, and safety requirements established in the facilities demand uncompromisingly high reliability and accuracy of any process monitoring and fault detection tool. To address these challenges, in this talk, we present a robust and reliable chemical process monitoring framework based on statistical process control (SPC) that can monitor nonparametric and heterogeneous high-dimensional data streams and detect process anomalies as early as possible while maintaining a pre-specified in-control average run length.

In addition, we present a novel integrated process monitoring and fault diagnosis framework that adopts the SPC algorithm for quick and reliable fault detection, as well as Principal Geodesic Analysis (PGA) for accurate fault diagnosis [1]. Thus, this integrated framework overcomes the drawbacks of conventional dimensionality reduction-based process monitoring techniques and achieves superior performance in the benchmark case study of the Tennessee Eastman Process (TEP). We also compare different choices of distance measures on the fault classification accuracy in the case study, such as Euclidian, Manhattan, Mahalanobis, and Cosine distance measures. We find that the cosine distance gives the highest classification accuracy among all candidates. These results and findings further improve the performance of our integrated framework.

References

[1] A. Smith, B. Laubach, I. Castillo, V.M. Zavala, 2022, Data analysis using Riemannian geometry and applications to chemical engineering, Computers & Chemical Engineering, 168, 108023.