Photovoltaic and wind energy plants that produce in a network are exposed to a variety of risks. Even a minor malfunction in IT or systems engineering can have far-reaching consequences for the ability to function. In the SecDER research project, six partners from science and industry are investigating how the power system can become more resilient with virtual power plants.
The background to the need for research is the restructuring of the energy system: In a decentralized energy system, many small and distributed plants work together and are jointly operated by a control system — the virtual power plant. The decentralized units communicate via a smart grid in real time and can become the target of cyber attacks in the process. The operation is also technically demanding: In the plant park, very different generation and storage technologies are used in order to optimise the use of sun, wind and storage in such a way that a reliable and economic power supply results.
In the future, the security of electricity supply will depend to a large extent on the resilience of virtual power plants. Resilient systems are characterized by their ability to respond to cyberattacks, bugs, or technical failures and automatically return to the desired operating state. The basic principle is a so-called Resilience Cycle with the five phases “Prepare — Prevent — Protect — Respond — Recover”. In sum, all phases contribute to minimizing the effects of a malfunction and to restoring the function of the system.
The research project, which is funded by the German Federal Ministry for Economic Affairs and Energy (BMWi) and supported by Project Management Jülich, began on 1 April 2021 and is scheduled to run for 36 months. The total funding volume amounts to 2.7 million euros. Scientists from the Fraunhofer Institutes for Energy Economics and Energy System Technology (IEE) and for Secure Information Technology (SIT) as well as the Hannover University of Applied Sciences and Arts are jointly developing an information system for faults in the decentralized power supply in cooperation with companies from the industry.
Fraunhofer IEE and SIT are working on the sub-project “Failure detection and resilience strategies for distributed energy systems”. Hannover University of Applied Sciences and Arts is researching AI-based methods for detecting attacks and DECOIT GmbH is developing an incident information system (SIS).
The project partners ENERTRAG AG and ANE GmbH & Co. KG contribute their know-how from the operation of virtual power plants to the work and support field tests.
The special thing about SecDER is that it takes an equal look at physical operation and data processing and bundles them into one application. “Our goal is to have a security system that can detect and resolve decentralized distributed attacks and technical failures. In order to provide a holistic picture of the IT security situation, it is important to evaluate individual reports according to relevance and urgency, to aggregate them clearly and to suggest solutions,” reports Stefan Siegl, Group Leader Applied Energy Informatics, Fraunhofer IEE.
In concrete terms, the scientists are interested in collecting comprehensive data and making it available in a Big Data Store. “The large amount of data is necessary to comprehensively detect anomalies,” Siegl explains.
On this basis, a detection system for attacks and technical faults can then be set up in a virtual lab. With this simulation environment, the real behaviour of the plants can be virtually simulated and strategies for resilient defence against cyber attacks and technical faults in virtual power plants can be tested. These results are then further developed in practice.
As part of the work package on the detection and treatment of technical faults, the IEE researchers plan to combine physical measured variables with statistical models in order to provide a real picture of the condition of the plants. In order to identify any unusual behaviour of the equipment, the data should be compared with operating figures for normal operation.
A particular challenge of a decentralized energy system lies in the constant operability of the technical systems. To this end, the SIS is intended to provide an overview of the condition of all components in use and to make recommendations for action. “A simple warning message is only the first step. We want the virtual power plant to be able to remedy as many of its faults as possible independently in interaction with the SIS,” Siegl emphasizes.
Another focus of the research project is on cyber attacks. Up to now, IT protection systems have been developed individually for each virtual power plant. Within the framework of SecDER, common IT systems are to be adapted to the needs of the energy industry. This includes, in particular, security requirements and standards for the platforms and networks used.
Anomaly detection systems are known from other applications that classify unusual behavior as an indication of an attack. Specifically for the operation of virtual power plants, it is now to be investigated whether such systems also increase safety here.
The concept of machine learning has so far been used to control vehicles and drones and also for speech recognition. In practice, however, it is highly susceptible to interference. The aim of the research project is therefore to make these procedures more robust.