Papers »

Distributed robust control for physical infrastructure networks

The ability of physical infrastructure networks to withstand disruptions is a key consideration for sustainability. Due to their increasing scale and interconnectedness, these systems tend to exhibit complex behaviors that pose several new challenges in their design and operation. For example, local disruptions may give rise to cascading failures with potentially systemic effects, and local actions aimed at mitigating disruptions can increase vulnerability of the other parts of the system. Motivated by this, we consider distributed robust control problems for infrastructure networks within a physics-based network flow framework that also models cascading failures. The control actions correspond to, e.g., routing and inflow control at nodes in transportation networks, and to load shedding by load-generator node pairs in power networks. These control actions determine flow distributions in the network. A given flow distribution is called feasible if it satisfies link-wise capacity constraints. We quantify a disruption in terms of the aggregate loss in flow capacities on the links. The network resilience for a given disruption and a given control policy is defined as the minimum reduction of exogeneous inflows under which the resulting flow distribution is feasible. Our objective is to compute the maximal network resilience over the class of all distributed control policies. We propose algorithms to computer lower and constructive upper bounds on the maximal network resilience, and identify sufficient conditions under which the bounds are tight. These bounds provide a rigorous method to quantify resilience of infrastructure systems in the presence of real-time control actions. These bounds are also useful in quantifying loss in resilience due to the distributed nature of control actions. We illustrate our findings through extensive simulations on real-world networks.

Author(s):

Qin Ba    
University of Southern California
United States

Ketan Savla    
University of Southern California
United States