(393g) Integrated Data-Driven Monitoring & Explicit Fault-Aware Control of Chemical Processes: An Adaptive Approach for Smart Operation

The emergence of Industrie 4.0, and Smart Manufacturing initiatives along with the advances in computational power has significantly valorized process data for decision-making [1]. Today, industrial “Big Data” plays a significant role in process monitoring, and accordingly correction/prevention of process failures during operation. Process faults/failures arise when disturbances, that change process dynamics, occur in a chemical process and cannot be reversed by the existing controlling scheme. The common practice to handle this undesired condition in industry is to adopt different maintenance strategies including corrective actions. Corrective maintenance strategies aim rapid reaction and recovery of the process from faulty operation. One of the most common corrective actions is re-tuning of the existing controllers within the system while the process continues under faulty condition [2]. However, the major pitfall of this approach is the dead time spent under faulty operation during traditional controller re-tuning which may possibly harm the end-product quality, and/or cause process delays to achieve certain product quality.

In order to minimize the process delay between control action changes to correct any undesired changes during process operation, we propose an alternative approach that integrates a novel machine learning based process monitoring framework with multi-parametric Model Predictive Controller (mp-MPC) design. Central to the process monitoring framework are novel theoretical and algorithmic developments in SVM-based feature selection [3-5] which encapsulates highly nonlinear relationships between features, thus improves fault detection model accuracy. On the other hand, multi-parametric Model Predictive Controller (mp-MPC) design via the Parametric Optimization and Control (PAROC) framework [6] enables having the offline map of explicit fault-aware control strategies. Here, the control strategies are affine functions of the system states and the magnitude of the detected fault. By integrating a novel and accurate data-driven monitoring framework and explicit fault-aware control technology, we aim to produce an adaptive approach for smart operation where rapid switches between a priori mapped control action strategies are enabled by continuous monitoring information of the chemical processes. The results will be presented through the penicillin production benchmark model [7].

References

[1] Edgar, T. F., & Pistikopoulos, E. N. (2017). Smart manufacturing and energy systems. Computers & Chemical Engineering.

[2] Huang, B., & Shah, S. L. (2012). Performance assessment of control loops: theory and applications. Springer Science & Business Media.

[3] Onel, M., Kieslich, C. A., Guzman, Y. A., Floudas, C. A., & Pistikopoulos, E. N. (2018). Big Data Approach to Batch Process Monitoring: Simultaneous Fault Detection and Diagnosis Using Nonlinear Support Vector Machine-based Feature Selection. Computers & Chemical Engineering.

[4] Kieslich, C. A., Tamamis, P., Guzman, Y. A., Onel, M., & Floudas, C. A. (2016). Highly Accurate Structure-Based Prediction of HIV-1 Coreceptor Usage Suggests Intermolecular Interactions Driving Tropism. PloS one, 11(2), e0148974.

[5] Guzman, Y. A. (2016). Theoretical advances in robust optimization, feature selection, and biomarker discovery (Doctoral dissertation, Princeton University).

[6] Pistikopoulos, E. N., Diangelakis, N. A., Oberdieck, R., Papathanasiou, M. M., Nascu, I., & Sun, M. (2015). PAROC—An integrated framework and software platform for the optimisation and advanced model-based control of process systems. Chemical Engineering Science, 136, 115-138.

[7] Birol, G., Ãœndey, C., & Cinar, A. (2002). A modular simulation package for fed-batch fermentation: penicillin production. Computers & Chemical Engineering, 26(11), 1553-1565.