Co-Synthesis for Self-Awareness and Reconfiguration in Networked Systems

From u-t-autonomous.info
Jump to: navigation, search

Technical point of contact: Lynn Petersen, Office of Naval Research
Period of activity: 2019-2022


Overview of the Project

The objective of this project is to develop theory, algorithms and computational tools for the automated synthesis of control protocols for the autonomous reconfiguration of networked systems in dynamic, uncertain and possibly adversarial environments. The resulting algorithms will account for the effects of the self-awareness of the system on its performance and proactively improve the system’s self-awareness as a service to the control protocols under varying environmental conditions, mission requirements and health status of the sensing hardware.

We utilize dynamic reconfiguration of the electric power network---including the components, e.g., generators, transformers, rectifier units and energy storage units, and their connectivity---on a naval ship as a motivating application. Dynamic reconfiguration facilitates timely delivery of power with the desired characteristics to the right components despite varying demands, operational conditions, faults and damage to the vehicle. The main synthesis problem is the design of a reactive control protocol that dynamically decides, as the health conditions of the components of the network and the operator requests change during the mission, on the contactor openings and closings so that certain safety and performance requirements hold at steady-state and during transients.

The focus of the effort is on establishing a critical capability for autonomous, dynamic reconfiguration: joint self-awareness and reconfiguration in networked systems. More specifically, it will explicitly account for the run-time, network-wide knowledge the controller has to maintain in order to ensure safety and performance. The resulting algorithms will, at run time, make decisions to proactively establish such knowledge. At design time, they will help systematically determine the necessary sensor modalities, placement, quality, availability and redundancy.

We partition the proposed effort into three research thrusts:

  • Thrust I: Synthesis of partial-information control protocols for network reconfiguration
  • Thrust II: Design-time sensor design for partial-information control protocols
  • Thrust III: Run-time, robust sensor scheduling

Thrust I will establish a foundation for control synthesis with information limitations building on which Thrusts II and III will develop design- and run-time methods, respectively, for co- synthesis for self-awareness and network reconfiguration.