Smart Grid: Artificial Intelligence in the Field of Renewable Energies

It's been 33 years since I first encountered the then-nascent field of Artificial Intelligence (AI). I worked with the AI ​​programming languages ​​LISP and Prolog. Through the university network, I also came into contact with the internet. At the same time, the satellite television market was booming. From there, I developed my skills in the field of intralogistics, eventually arriving at my current position in photovoltaics.

The FAW Ulm (Research Institute for Application-Oriented Knowledge Processing) was founded in 1987 as the first independent institute for artificial intelligence. Companies such as DaimlerChrysler AG, Jenoptik AG, Hewlett-Packard GmbH, Robert Bosch GmbH, and several others were involved. I myself worked there as a research assistant from 1988 to 1990.

Meanwhile, AI has found its way into many fields, be it medicine, law, marketing, or computer games. Machine translations, for example with Google Translate or DeepL, are among the best known applications. AI is also used in the analysis and forecasting of stock price movements and in managing the flood of information in search engines.

Artificial intelligence is a subfield of computer science that deals with the automation of behavioral patterns, from which decision support can be derived and, ideally, leading to independent, autonomous processes. It is most often used when an excessively large or unstructured, yet unmanageable, amount of data needs to be managed and coordinated.

It is not always successful. For example, Amazon had to deactivate its AI for evaluating applicants because the automated rating system disadvantaged women.

And even in machine translations, there are still quite often some crude errors that, upon closer inspection, cause frowns or smiles.

So, artificial intelligence isn't actually that simple. The problem isn't really the amount of data, but rather its correct interpretation. Because Amazon had predominantly hired men, the AI ​​concluded that women had a performance deficit. However, it failed to consider that the low percentage of women in male-dominated professions has sociological causes.

The fundamental problem with artificial intelligence is that the programming of the algorithms and the source data are only as good as the subjective work of the developers who create and provide them. Deficiencies in objectivity due to individual emotions and intentions, as well as errors in interpretation and perception on the part of the developers, are adopted by the AI; it learns from them and further develops its capabilities. If a lack of knowledge about the interrelationships between things and processes (key skills) is added to this, the cycle is complete.

More on this: Artificial intelligence made simple

AI therefore needs a lot of development time and the courage to face setbacks before an efficient system can emerge.

Headlines such as “Artificial Intelligence (AI) as a driver of the energy transition” or “How logistics benefits from artificial intelligence” are media sensations that do not even begin to reflect the development and effort required, and that costs are the primary concern before financial profitability becomes apparent.

Artificial intelligence has so far been used in the energy industry primarily for monitoring or forecasting tasks.

 

Smart Grid – Intelligent Power

However, with the increasing share of electricity from renewable energies, it is becoming clear that AI will also control the processes of the energy system on a large scale in the future.

While centralized power generation has dominated electricity grids to date, the trend is shifting towards decentralized generation facilities. This is particularly true for renewable energy sources such as photovoltaic systems, solar thermal power plants, wind turbines, and biogas plants. This leads to a significantly more complex structure, primarily in the areas of load control, voltage regulation in the distribution network, and maintaining grid stability. Smaller, decentralized generation facilities, unlike medium to large power plants, also feed directly into lower voltage levels such as the low-voltage or medium-voltage networks.

 

Development of a smart grid

An intelligent power grid integrates all stakeholders into a comprehensive system through the interplay of generation, storage, grid management, and consumption. Power plants (including storage facilities) are already controlled to ensure that the amount of electrical energy produced always matches the amount consumed. Intelligent power grids incorporate consumers, as well as decentralized small-scale energy suppliers and storage facilities, into this control process. This results in a balanced consumption pattern across time and location (smart power/intelligent electricity consumption) and allows for better integration of non-dispatchable generation facilities (e.g., wind turbines and photovoltaic systems) and consumers (e.g., lighting).

With the increasing share of renewable energies, it is becoming more important to align fluctuations in energy production with fluctuations in energy consumption. Besides the possibility of storing electrical energy using energy storage systems or pumped-storage power plants, demand-driven electricity generation (e.g., through hydroelectric power plants or bioenergy), and the expansion of electricity grids for rapid distribution over large areas, there is also the option of adjusting electricity consumption to the electricity supply.

“Electricity generation from solar and wind power plants makes the supply system significantly more fragmented and weather-dependent than the operation of conventional power plants. Furthermore, consumption must be more closely aligned with electricity supply. The necessary flexibility cannot yet be managed with the existing infrastructure. A decentralized system can only function through real-time digital processes and automated decisions,” explains Prof. Dr. Clemens Hoffmann, Director of the Fraunhofer IEE. Hoffmann sees digitalization as the foundation for the next steps in the energy transition: “The coordination and decision-making processes of a decentralized renewable energy supply are extremely complex. Only artificial intelligence will make it possible to connect different systems, such as electricity and heat supply as well as mobility, on a large scale through automated decisions. By building an ecosystem for cognitive energy systems, we are advancing the applications of AI in the energy sector.”

 

A decentralized energy system needs AI

There is already a concrete need for AI in various areas of the energy sector. For example, in automated energy trading, the focus is on systems that independently identify trading strategies and trigger buy or sell orders. Photovoltaic and wind power plants, as well as charging stations and electrolyzers, can use AI to optimize their operation, thereby reducing maintenance and extending their lifespan. In the grid sector, the technology is used to analyze a wide range of information, identify critical situations, and support their resolution.

Fraunhofer IEE has been working for 15 years on artificial intelligence for predicting weather-dependent electricity generation from solar, wind, and bioenergy. An automated trading system for the EPEX Spot power exchange is also under development in Kassel.

 

Research for AI in the energy sector

“Artificial intelligence is a key technology for the further development of the energy transition: The shift away from a centrally organized power plant industry based on fossil fuels to an energy system based on renewable sources is a highly complex process that can only be managed through intelligent control,” said Hesse’s Minister of Science, Angela Dorn. “The Competence Center for Cognitive Energy Systems provides scientists with space for new ideas and research approaches to innovations in the energy sector. I am delighted that we are supporting its establishment. Now it is crucial to combine the expertise of researchers with strong partners from industry.”

Therefore, a new competence center for cognitive energy systems is being established in Kassel. The research project on artificial intelligence in the energy system is seeking partners from academia and industry and sees excellent opportunities for Germany as a business and research location to achieve global innovation leadership in this field. For this reason, the state of Hesse is supporting the establishment of the new competence center, which is being run by the Fraunhofer Institute for Energy Economics and Energy System Technology IEE.

These application areas of AI are being researched by the new Competence Center for Cognitive Energy Systems in Kassel, whose establishment is being funded by the Hessian state government with a total of 5.8 million euros between 2020 and 2022.

 

The K-ES

The Competence Center for Cognitive Energy Systems (K-ES) has been under development by the Fraunhofer IEE since mid-2020 to research cognitive energy economics, cognitive energy networks, and cognitive energy system technology. The development process is planned to last ten years. The K-ES aims to become a national and international center for artificial intelligence in research and teaching.

The Competence Center for Cognitive Energy Systems (K-ES) examines energy system tasks from an AI perspective and further develops them in three areas: Cognitive Energy Economics, Cognitive Energy Networks, and Cognitive Energy System Technology. “A cognitive energy system independently determines its state based on available information and learns to achieve predefined goals. Artificial intelligence is not opposed to human intelligence, but rather engages in constant exchange with it and supports it. With the further development of the technology, both sides will change,” explains IEE project manager André Baier.

The energy sector can also build on insights from other industries. AI is already fundamentally changing the automotive industry, retail, and the insurance and financial sectors. For the energy transition with renewable energies and sector coupling, the most important areas of digitalization are smart producers and consumers, virtual power plants, smart grid technologies, and real-time energy management.

 

Concepts and applications for the economy

The concept for establishing the K-ES (Competence Center for Energy Systems) was developed by the Fraunhofer IEE. The initiative stems from an agreement in the coalition agreement of the Hessian state government. The development phase has now begun. The primary goal is to create an ecosystem for innovation and build a community of experts. The new competence center will be part of the Fraunhofer IEE campus currently under construction in Kassel and will complement the research portfolio for the transformation of energy systems.

The first step involves setting up premises and IT infrastructure with a cloud system. Following this, a digital platform will be created to facilitate exchange between partners from industry and research. The initial phase will focus on recruiting scientists and developing expertise. "Our aim is to connect scientists who share a common goal, regardless of where in the world they are based," says Baier.

Until the planned official establishment of the competence center, the focus will be on acquiring partners and securing application projects from industry. A close connection with the energy sector is a key part of the concept: K-ES's services for energy companies range from consulting and concept studies to prototypes and turnkey systems. "We welcome applications from researchers and companies alike, because such an ecosystem thrives on the networking between theory and practice," emphasizes Hoffmann.

 

The goal: A community of international renown in Germany

Over the next ten years, the K-ES is expected to have around 100 experts working in the fields of data science, advances in machine learning, recommender systems, and digital innovation management. Currently, 15 employees at the Fraunhofer IEE are working in these areas. The new institution aims to become one of the leading AI communities in the energy sector in Germany.

To reflect the high degree of internationality in AI research, the competence center also offers visiting scientists from around the world the opportunity to participate. “Thanks to the specialized training infrastructure, appropriate hardware and software, and a comprehensive model and data repository, we can conduct efficient and cross-site AI research for the energy system,” explains Christoph Scholz, scientific director of K-ES, regarding the available possibilities.

Globally, intensive work is underway on the development of AI. Germany has so far spent significantly less on this research than its competitors, the USA and China. As part of the German government's Corona-related economic stimulus package, €5 billion is now to be invested in AI by 2025. "With regard to AI in the energy system, Germany, as a business and research location, is well-positioned to achieve global innovation leadership. To this end, it is crucial that all stakeholders work together to advance this topic," said Hoffmann.

 

Cognitive Systems

A cognitive system is a digital system with interfaces between the digital world and the environment, capable of perceiving and understanding things, drawing conclusions, and learning. Cognitive systems are able to independently develop solutions to human problems. They can interact and cooperate with other digital systems, interpret contexts, and are adaptable.

Cognitive systems are being used in an increasing number of areas and, for example, represent the fundamental technology for self-driving vehicles, intelligent personal assistants, Industry 4.0, and the Internet of Things. A typical characteristic of such systems is their ability to process large amounts of data in a short time and their integration into a higher-level system (system of systems). By 2020, tens of billions of euros had been invested worldwide in this technology.

A cognitive system can independently determine its own state and that of its assets based on available information and, through its ability to adapt, learn to achieve predefined goals autonomously. Cognitive energy systems are a key technology for the energy transition. Applications in the electricity sector can be found in grid management and the management of generation and consumption.

Within the ecosystem for cognitive energy systems, access to AI is being facilitated for the various market roles. The tasks of plant operators, metering point operators, balancing group managers, and direct marketers are being automated to such an extent that they can be carried out autonomously. The "Energy Avatar" model (see above) illustrates how easily a homeowner with a solar power system can participate in the energy market when all processes are automated. The Energy Avatar is currently being developed in collaboration between the Fraunhofer Institutes IEE and IOSB-AST.

A close connection with the energy sector is part of the concept: K-ES's services for energy companies range from consulting and concept studies to prototypes and turnkey systems. The ecosystem thrives on the networking between theory and practice.

Automation and autonomization. Read more about it here: “CO2 neutrality – Learning from Amazon”