(346d) Community Detection for Distributed Control

Control of large-scale plants has been challenging, and centralized and distributed control of these plants have been studied extensively [1, 2, 3, 4]. Distributed control provides a control system with some form of modularity, which eases the task of controller installation, tuning, and maintenance. Also, efforts have been made to develop methods for the systematical decomposition of large-scale systems to subsystems to apply distributed control to the subsystems [5, 6]. In particular, community detection based on steady-state attributes of large-scale plants has been used to conduct the decomposition. For example, interactions have been quantified based on connectivity and response sensitivity at steady-state conditions, and community detection algorithms based on hierarchical clustering and modularity optimization have been proposed [5].

In this work, we present a community detection method that decomposes a system into smaller subsystems based on dynamic attributes and structural controllability of each subsystem with the local manipulated variables of the subsystem. The dynamic attributes include response times and time delays. The application and performance of the method are shown by applying it to the Tennessee Eastman process. The resulting subsystems are compared to those obtained using community detection based on steady-state attributes. A multi-objective whale optimization algorithm is used to solve the community detection problems.

References

[1] Christofides, Panagiotis D. "Control of nonlinear distributed process systems: Recent developments and challenges." AIChE Journal 47.3 (2001): 514-518.

[2] Liu, Jinfeng, David Munoz de la Pena, and Panagiotis D. Christofides. "Distributed model predictive control of nonlinear process systems." AIChE journal 55.5 (2009): 1171-1184.

[3] Andreasson, M., Dimarogonas, D. V., Johansson, K. H., & Sandberg, H. (2013, July). Distributed vs. centralized power systems frequency control. In 2013 European Control Conference (ECC) (pp. 3524-3529). IEEE.

[4] Dijkstra, E. W. (1982). Self-stabilization in spite of distributed control. In Selected writings on computing: a personal perspective (pp. 41-46). Springer, New York, NY.

[5] Daoutidis, Prodromos, Wentao Tang, and Sujit S. Jogwar. "Decomposing complex plants for distributed control: Perspectives from network theory." Computers & Chemical Engineering 114 (2018): 43-51.

[6] Tang, W., Allman, A., Pourkargar, D. B., & Daoutidis, P. "Optimal decomposition for distributed optimization in nonlinear model predictive control through community detection." Computers & Chemical Engineering 111 (2018): 43-54.