(161a) Sensor Fusion Applications in Industrial Chemical Process Monitoring | AIChE

(161a) Sensor Fusion Applications in Industrial Chemical Process Monitoring


Peng, Y. - Presenter, The Dow Chemical Co
Rendall, R. - Presenter, University of Coimbra
Sensor Fusion combines available information from both hardware and software sensors to provide more accurate and reliable inference of Key Performance Indicators (KPI) than the one based on individual sensors. Though the concept of sensor fusion is relatively new to Chemical Industry, it has been widely applied in areas like autonomous vehicles, smart healthcare, precision farming and smartphones for better decision making. Due to the complexity of chemical processes, an individual sensor or monitoring technique does not always provide comprehensive and reliable inference. Also, product quality variables of chemical processes, such as product compositions and purities, are often measured off-line using analytical techniques in a laboratory. While the advantage of these instrumental quantification is obvious as it usually produces high accuracy data, the disadvantages such as long turnaround time and low sampling frequency can often lead to inefficient process monitoring and control. Therefore, to obtain more frequent and faster estimations of product quality variables, software sensors (also known as soft sensors or inferential sensors), including partial least squares (PLS) models [1, 2] and other data-driven mathematical models like neural networks [3, 4], have been widely applied over the past three decades [5]. Hence, combining these available information from software and hardware sensors using sensor fusion is an attractive option to enhance the accuracy and reliability of process monitoring. With the use of sensor fusion, the assets of industrial chemical processes will be operated in a more efficient, reliable and profitable manner.

A number of applications have been carried out with the use of sensor fusion from Dow’s businesses, one related to critical chemical level monitoring in waste water streams and the other in a product manufacturing process. In the first case, both regression and classification type of software sensors are built and enhanced with decision and model fusion to achieve high accuracy on toxin level quantification. A more than 10 times faster turnaround is achieved than traditional GC/MS method with <5% false negative alarms of incorrectly predicting a low toxin concentration when actual level is high. In the second case, a multi-rate sensor fusion scheme based on Bayesian fusion is developed. The product impurity was estimated by fusing the measurements from a lab analyzer, an online analyzer and a PLS soft sensor with different sampling rates. The proposed approach is found to correctly identify impurity alarm without generating any false alarms [6].

Overall, applying sensor fusion in the proposed business/manufacturing processes is expected to achieve quick monitoring and bring increased product quality, on-time delivery, and product availability.


  1. Wold, S., M. Sjöström, and L. Eriksson, PLS-regression: a basic tool of chemometrics. Chemometrics and intelligent laboratory systems, 2001. 58(2): p. 109-130.
  2. Chiang, L.H., et al., Diagnosis of multiple and unknown faults using the causal map and multivariate statistics. Journal of Process Control, 2015. 28: p. 27-39.
  3. Bishop, C.M., Neural networks for pattern recognition. 1995: Oxford university press.
  4. Jang, J.-S.R., C.-T. Sun, and E. Mizutani, Neuro-fuzzy and soft computing: a computational approach to learning and machine intelligence. 1997.
  5. Kadlec, P., B. Gabrys, and S. Strandt, Data-driven Soft Sensors in the process industry. Computers & Chemical Engineering, 2009. 33(4): p. 795-814.
  6. Wang, Z.; Chiang, L., Monitoring Chemical Processes Using Judicious Fusion of Multi-Rate Sensor Data. Sensors 2019, 19 (10), 2240.