Motivation

Stable operation of complex systems such as power networks, financial systems and transportation networks are crucial to any advanced society. When these systems become unstable or fail (for example, in the case of a large-scale power outage or financial crisis), there may be devastating humanitarian and economic consequences.

And yet the safety, stability and efficiency of these systems are affected by a great many uncertainties, presenting researchers with a whole range of complex, interrelated problems.

In the case of power networks, for example, there is a great deal of uncertainty stemming from the push to incorporate renewables and distributed generation into our grids.

How does uncertainty in supply affect the reliability of the system? How much power should be kept in reserve to ensure reliable operation at a reasonable cost? How should reserves, which may or may not be tapped at any given time, be priced?

As these systems become increasingly complex, they require increasingly sophisticated models and control methods to support their operation.

Researchers working on other classes of large-scale systems face similarly complex problems, and though each community has developed different tools and methods for dealing with them, there is much common ground.

Across many applications, stochastic hybrid systems are considered an ideal framework for studying complex, large-scale systems, and several distinct approaches have emerged from each community. Unfortunately, none of the models are presently able to deal with the complexities of real, large-scale applications.

And yet methodologies to address these concerns are viewed as an essential step toward the effective management of large-scale systems.

We are specifically motivated by the case of power systems, for example, where new methodologies are essential to the seamless introduction of renewable energy sources in the existing grid, in the development of smart grids, and toward the development of networks that are resilient to unforeseen natural or malicious events.

However, it is our hope that our methodologies can be applied to other classes of large-scale systems as well.

announcements

VeriSiMPL Toolbox
Verification via biSimulations of Max-Plus-Linear models. This toolbox is used to generate finite abstractions of autonomous and nonautonomous Max-Plus-Linear (MPL) models over R^n. Alessandro Abate & Dieky Adzkiya (TU Delft).

HSCC 2014
April 15-17, 2014
Berlin, Germany
Martin Fraenzle and John Lygeros chair the HSCC Program Committee.