Business processes, e.g. economic or operational optimization, in large power systems employs heuristics based on solving a relaxation of the original problem due to the number of actors and components involved as well as the time-criticality for finding feasible solutions. In order to cope with the large number of conflicting objectives in the problem instance, multi-agent systems (MAS) are often chosen in order to find timely solutions while guaranteeing certain fairness aspects. However, utilizing MAS poses specific challenges achieving observability and other non-functional requirements in the context of safety-critical electrical energy systems. To cope with these systemic risks, we will sketch out methods to increase observability and dynamic reorganization in the sense of regulated autonomy.
The energy transition (Energiewende) introduces rapidly changing requirements into
historically grown, heterogeneous industry and application landscapes. Future energy
system solutions – so-called smart grids – are increasingly IT-based. The increasingly
decentralised and highly automated management of energy sources is changing the
existing communication and energy distribution structure as well as introduces various
new stakeholders to the system. In that sense it is an autonomous cyber-physical system
(ACPS) that is controlled or monitored by highly autonomous computer-based systems,
tightly integrated with the Internet and its users. Physical and software components are
deeply intertwined, each operating on different spatial and temporal scales, exhibiting
multiple and distinct behavioural modalities, and interacting with each other in
numerous ways that change with context [
Additionally, electrical energy systems exhibit special characteristics that set them
apart from other ACPS (e.g. transportation systems). Energy systems are considered a
critical infrastructure (German acronym: KRITIS1) and are indispensable lifelines of
modern societies. They have a size that transcends continents – from North Africa to
Scandinavia, from Ireland to Asia. Phenomena and their dynamics exhibit
instantaneous propagation speed – this is a special challenge with regard to the analysis and
timely reaction, i.e. containability of instabilities. Conflicts of objectives are
omnipresent – monetary, technical, political (national/international) interests meet in both
systems planning as well as operation. Last but not least, the energy system is undergoing
rapid and fundamental change following the energy transition [
From an ICT-perspective, future power systems are composed of billions of
components – producers and consumers of electrical energy, grid and operating equipment,
markets as well as (local) coordination platforms. This complexity together with
nonlinear system dynamics and almost immediate propagation velocities results in business
processes to be highly decentralized, multi-objective, dynamic, real-time optimization
problems [
One strategy to cope with this complexity is to follow distributed heuristical
optimization approaches on the basis of multi-agent systems (MAS) [
With multiple business process being executed at the same time and within the same technical (sub-)system, it is foreseeable that at any given time there is a set ordering regarding the criticality or priority over a certain number of processes, e.g. stabilityrelevant processes are more critical than economic dispatch, or large bulk-market contracts being of higher priority than small-scale decentralized contracts.
The goal of this work is to derive relevant solution parameters of distributed process
optimization and integrating them into existing supervisory automation and control
concepts. This not only allows for facilitating the observability of the most critical
processes but also the controllability of the heuristic solution process by assigning specific
ICT-resources (computationally or communication-specific) to individual processes.
Thus, event-driven reorganization – even temporary centralization of the optimization
process – may be achieved, in the sense of a Smart Grid dynamic regulated autonomy
approach [