(340by) Community Detection for Distributed State Estimation

The global competition for producing higher-quality products at lower costs has driven the development and implementation of advanced control, sensing, and monitoring methods [1-3]. Effective monitoring and control of a plant require adequate online information on the state variables of the plant, which may not be available. This unavailability can be compensated for using a state estimator. In the case of a large-scale plant, the use of a centralized state estimator is not computationally efficient [4], motivating the use of distributed state estimators. In a distributed implementation, a large-scale system is decomposed into subsystems with minimum inter-connection and maximum intra-connections. A distributed implementation offers advantages such as low computational burden, ease of implementation, and improved robustness to sensor failures [5].

As previous studies on distributed state estimation have mostly focused on a given distributed architecture [6, 7], in this work, we propose an algorithm based on community detection to capture the optimal architectures for distributed implementation of state estimators. To this end, a directed weighted graph is constructed based on a state-space process model. We formulate a community detection problem that maximizes the modularity index subject to the structural observability of each subsystem. We solve the multi-objective optimization problem using the Whale optimization algorithm [8], which takes advantage of the non-dominated sorting approach to generate all potential observable subsystems with a single run. This algorithm provides the user with all possible observable subsystems; for example, one may choose those configurations aligning with the physical topology of the system to reduce the computational burden associated with their implementation in real-time. The proposed method is implemented and validated on the Tennessee Eastman process [9], which has been extensively used as a benchmark.

REFERENCS

[1] Baldea, M. and P. Daoutidis, Dynamics and nonlinear control of integrated process systems. 2012: Cambridge University Press

[2] Soroush, M., et al., Model-predictive safety: A new evolution in functional safety, in Smart Manufacturing. 2020, Elsevier. p. 283-321.

[3] Christofides, P.D., J. Liu, and D.M. De La Pena, Networked and distributed predictive control: Methods and nonlinear process network applications. 2011: Springer Science & Business Media.

[4] Khan, U.A. and J.M. Moura, Distributing the Kalman filter for large-scale systems. IEEE Transactions on Signal Processing, 2008. 56(10): p. 4919-4935.

[5] Carli, R., et al., Distributed Kalman filtering based on consensus strategies. IEEE Journal on Selected Areas in communications, 2008. 26(4): p. 622-633.

[6] Zeng, J., et al., Distributed extended Kalman filtering for wastewater treatment processes. Industrial & Engineering Chemistry Research, 2016. 55(28): p. 7720-7729.

[7] Hu, Z., J. Hu, and G. Yang, A survey on distributed filtering, estimation and fusion for nonlinear systems with communication constraints: new advances and prospects. Systems Science & Control Engineering, 2020. 8(1): p. 189-205

[8] Mirjalili, S. and A. Lewis, The whale optimization algorithm. Advances in engineering software, 2016. 95: p. 51-67.

[9] Downs, J.J. and E.F. Vogel, A plant-wide industrial process control problem. Computers & chemical engineering, 1993. 17(3): p. 245-255.