(6bm) Applied Statistics and Data Analytics for Advanced Process Systems Engineering | AIChE

(6bm) Applied Statistics and Data Analytics for Advanced Process Systems Engineering

Authors 

Tulsyan, A. - Presenter, Massachusetts Institute of Technology

Since the formal inception of Process Systems Engineering (PSE) in 1961, PSE has rapidly evolved at the interface between chemical engineering, applied mathematics, statistics and computer science. Recent advancements in computing capabilities and big-data analytics, along with emerging trends in information technology have enabled PSE ideas to be extrapolated and applied to systems beyond chemical processes, such as molecular, health-care, neuroscience, financial, social, internet, and behavioral systems.

Today, process industries face unique challenges in terms of growing process complexity, shift towards process integration and miniaturization, dynamic operating conditions, hard economic constraints and stricter environmental regulations. In the face of rising uncertainty in process operations, data becomes an indispensable asset for smart economic decision-making and safe plant operations.  The use of process operation and process control computers along with process information systems have enabled process industries to collect and archive massive databases from equipment, process, operation and customer data [1]. Furthermore, advancements in process sensing technologies ranging from conventional process sensors to images, videos, and virtual-sensing technologies have made data acquisition (DAQ) not only accurate and cheap, but also contributed to the diversity and complexity of process data [1]. A recent AIChE Journal Perspective article [2] characterizes the massiveness of process data as “drowning in data.” Despite the richness of process data available at our disposal, PSE research has not been able to keep pace with developments in advanced statistical inferencing, machine-learning, data mining and analytics techniques.

My current research focus is on the use of statistics and data-analytics to develop novel data-based solutions for complex process systems engineering problems. The thrust of my research is on developing data-based process methods for inferential sensor development, process monitoring and control, product optimization, and safety and compliance management. With my strong background in chemical engineering, applied statistics, and computational sciences, we have been successful in developing data-based solutions for various applications in process, health-care, oil and gas, pharmaceutical and aerospace industries. Some of the key problems addressed in the current research plan includes:

• Mining and fusion of process information for improved process identification [8]

• Data quality and optimal experiment design for efficient non-linear identification [8, 9]

• Real-time adaptive Bayesian identification of non-linear processes with irregular data [7, 12]

• Performance assessment, diagnosis, and selection of non-linear filtering methods [4, 10, 11, 18]

• Inferential sensor for online monitoring of Type-II diabetic patients using clinical data [3]

• Model-based delay-timer alarm design for non-linear industrial processes [16]

• Reachability study for multivariate chemical systems subject to flow-rate disturbances [13, 14, 15]

• Global parameter estimation in chemical reaction networks using kinetic data [17]

• Multi-model approach to non-linear identification of distillation columns using EM algorithm [5]

• Theoretical statistical inferencing and filtering limits in ballistic target tracking problems [6]

As a faculty member, my research plan will be to continue to build on the fundamental and applied research in statistics and data analytics, and to provide improved data-based solutions to challenging problems in process, energy and health-care systems. Using my interdisciplinary research experience along with my expertise in applied statistics, some of the key research directions we would like to explore in future includes:

• Data Mining, Machine Learning and Big Data Analytics for Advanced Process Systems Engineering;

• Inferential Sensors for Process, Energy and Health-Care Systems; and

• Process Monitoring, Safe Operations, and Compliance Management in Process Industries.

References

1. Qin, S.J. "Process data analytics in the era of big data." AIChE Journal. Vol. 60, No. 9, pp. 3092-3100, 2014.

2. Venkatasubramanian, V. "DROWNING IN DATA: Informatics and modeling challenges in a data‐rich networked world." AIChE Journal. Vol. 55, No. 1, pp. 2-8, 2014

3. Barazandegan, M., Ekram, F., Kwok. E., Gopaluni, R.B., Tulsyan, A.“Assessment of type II diabetes mellitus using irregularly sampled measurements with missing data”. Bioprocess and Biosystems Engineering. Vol. 38, No. 4, pp. 615–629, 2015

4. Tulsyan, A., Huang, B., Gopaluni, R.B., Forbes, J.F. “Performance assessment, diagnosis, and optimal selection of non-linear state filters”. Journal of Process Control, Vol. 24, No. 2, pp. 460-478, 2014

5. Chen, L., Tulsyan, A., Huang, B., Liu, F. “Multiple model approach to non-linear system identification with uncertain scheduling variables using EM algorithm”. Journal of Process Control, Vol. 23, No. 10, pp. 1480-1496, 2013

6. Tulsyan, A., Huang, B., Gopaluni, R.B., Forbes, J.F. “A particle filter approach to approximate posterior Cram´er-Rao lower bound: The case of hidden states”. IEEE Transactions on Aerospace and Electronic Systems, Vol. 49, No. 4, pp. 2478-2495, 2013

7. Tulsyan, A., Huang, B., Gopaluni, R.B., Forbes, J.F. “On simultaneous on-line state and parameter estimation in non-linear state-space models”. Journal of Process Control, Vol. 23, No. 4, pp. 516-526, 2013.

8. Tulsyan, A., Huang, B., Forbes, J.F. “Designing prior for robust Bayesian optimal experimental design”. Journal of Process Control, Vol. 22, No. 2, pp. 450-462, 2012

9. Tulsyan, A., Huang, B., Gopaluni, R.B., Forbes, J.F. “Bayesian identification of non-linear state-space models: Part I- Input design”. IFAC 10th International Symposium on Dynamics and Control of Process Systems, December 18-20, 2013, Mumbai, India

10. Tulsyan, A., Khare, S.R., Huang, B., Gopaluni, R.B., Forbes, J.F. “Bayesian identification of nonlinear state-space models: Part II- Error Analysis”. IFAC 10th International Symposium on Dynamics and Control of Process Systems, December 18-20, 2013, Mumbai, India

11. Tulsyan, A., Huang, B., Gopaluni, R.B., Forbes, J.F. “Performance assessment of non-linear state filters”, IFAC International Symposium on Advanced Control of Chemical Processes, July 10-13, 2012, Singapore

12. Tulsyan, A., Gopaluni, R.B. “Sequential Monte-Carlo methods for inferencing in non-linear chemical systems- A tutorial”. Processes, 2015.

13. Tulsyan, A., Barton, P.I. “Interval enclosures for reachable set of chemical kinetic flow systems. Part 1: Developing sparse transformation”. Chemical Engineering Science, 2015.

14. Tulsyan, A., Barton, P.I. “Interval enclosures for reachable set of chemical kinetic flow systems. Part 2: Direct bounding method”. Chemical Engineering Science, 2015.

15. Tulsyan, A., Barton, P.I. “Interval enclosures for reachable set of chemical kinetic flow systems. Part 3: Indirect bounding method”. Chemical Engineering Science, 2015.

16. Tulsyan, A., Gopaluni, R.B. “Industrial alarm design for nonlinear dynamical systems”. Journal of Process Control. Working paper.

17. Tulsyan, A., Schaber, S., Barton, P.I. “PERKS: Software for parameter estimation in reaction kinetic systems”. AIChE Journal. Working paper.

18. Tulsyan, A., Khare, S.R., Huang, B., Gopaluni, R.B., Forbes, J.F. “An optimal filter switching strategy for on-line state and parameter estimation”. IEEE Transactions on Signal Processing. Working paper.