Smart Local Energy Systems (SLES) are becoming more complex with the rapid increase of energy sources and storage options; at the same time, connectivity between multiple generation and distribution sub-systems vary over time and space. The stakeholders, users and beneficiaries involved in the energy sector are therefore becoming more diverse with a need to be proficient in overcoming SLES constraints and developing systems that are demonstrably futureproof.
Research will focus on Infrastructure and Cyber-Physical techniques as applied to smart local energy systems. We will investigate ways to cope with, and exploit the complexity that smart, local energy systems bring – technologically, economically and socially. Flexibility will be designed into the present and future SLES, through the use of Artificial Intelligence (AI), machine learning, agent-based systems and distributed decision-making approaches. To function effectively and securely, AI needs to be supported with appropriate and well-defined measurement and sensing together with cybersecurity approaches that protect the information generated and stored within the system and propagated without it.
When two energy sources, such as photovoltaics and the grid for example, are used together in an energy system, sophisticated controls are required to optimally manage the acquired, variable energy and its distribution to multiple consumers. Supervisory control should play an essential intermediary role in smart energy systems by synchronising where energy can, or needs to, come from and where it needs to go; when this is achieved, the energy system is balanced, efficient, waste-less and delivers high quality service, economically.
The complexity of the control system grows exponentially with many and diverse sources of energy, coupled with storage options, various load types and complex, constrained distribution grids. Each component of the system will have different functional parameters and will be subjected to a multitude of constraints, on different timescales and abstraction levels. Therefore, control of distributed energy systems will rely on substantial data streams that are gathered and consumed in real time and near real time to effect control actions.
We will endeavor to design not only effective and optimal distributed control algorithms but also sensing systems that gather the right data, at the right time, with high yield and representing the system state and its essential measurands. Moreover, such systems should transport the data efficiently (time/cost) to the location where it is needed within the distributed energy system.
Given the level of complexity in the design of sensing systems, state measurement and control, data mining through AI methods and supporting mathematical modeling and simulation/emulation tools will be developed.
The tasks to be undertaken are far from trivial with regards to both mathematical modeling and the conceptualisation of control. For example, at a simple, microgrid level alone, events happen on several timescales on both energy supply and demand side. These can range from instantaneous events, relating to signals which fluctuate rapidly, to longer timescales relating to energy storage and storage availability, to mega timescales for environmental effects of our energy systems.
This timescale complexity prevent traditional mathematical simulation and straightforward control theories being sufficient or adequate to enable smart energy systems to be conceived in such a way that they do what they were intending to do in the first place: offer greatest value at lowest cost and lowest environmental footprint.
AI mining and machine learning techniques can skip through levels of complexity where numerous multi-input multi-output time varying systems are involved. Complex models can be replaced with black box learning machines that can be more agile, reactive and adaptive to changes that take place in the system. AI can manage diverse timescales beyond mathematical models of simple orders. If enough data can be collected about a system’s functionality then AI techniques can be successfully used to improve the control of the energy system, allowing simulated learning over the long timescales to bring in improved efficiency and performance enhancement on the short timescales.
Data from sensors is a critical part of this process. Generating and transporting data within systems comes with cost; thus attention needs to be paid to what data needs to be generated and how this data will be used both inside and outside of the smart local whole energy system to provide value.
Smart local energy systems call for more and better use of data through advanced modeling (using AI techniques) and improved control. Increased use of data does introduce concerns around a system’s cyber security infrastructure. Very careful consideration has to be given to how secure the data is and protocols to ensure this have to be developed. The system and the data within it, has to be “wrapped” in a cyber-security shell that is appropriate to the context, including the type of users and the processes involved in data transfer.
A large amount theoretical and applied research in the scientific areas of AI and all aspects of Cyber-Physical Systems (CPS) has been carried out. However, practical validation and verification of innovations in real-life energy systems is less common, as is the evaluation of added value through integrative and whole system philosophy to the application of CPS advances for energy generation, distribution and storage. The transfer of demonstrated, commercially viable AI and cyber-based solutions to complex problems from domains such as transport and manufacturing has not been yet fully explored by the energy and CPS communities of researchers. Previous experiences have not yet been linked together in a cohesive way to really understand how AI, for example, can be exploited for energy systems design effectively and at what cost.
Further investigative research in partnership with systems designers and practitioners will reveal practical know-how for energy systems designers and together with fundamental advances for future uptake.
We propose that data gathered from every sub-component of an energy system can be used to answer questions about the energy eco-system using novel perspectives on data mining. Our work may help to answer key domain questions such as 'what is the energy transformation process across vectors and what are the transformative effects?' 'What is the energy being used for, by whom, and at what cost?' and 'how can additional, external data be brought into the wider energy system to enable the local energy system to perform more efficiently?'
Research in EnergyREV will piece together innovation and insights across the areas of AI, measurement and sensing, control and cyber-security, through a series of case studies that link up these key facets of science, in the context of the whole energy system, and bring advances into practice.
A state-of-the-art review will be carried out to investigate research that has already been undertaken associated with smart local energy systems in the key areas of AI, measurement and sensing, control and cyber-security to explore their practical potential. We will work with a selection of case studies that will enable a generic consideration of what does and doesn’t work. Case studies will include all of the PFER demonstrators as well as other local energy systems of which are technologically and economically advanced.
Streams of work with a range of contributions and innovations will be brought together to explore how the performance of specific smart local energy systems can be improved by layering a cyber-physical infrastructure onto them.
An evaluation of sensing, control and AI added value in practice, including biggest wins, will be carried out, as well as an evaluation of combinations of services that could increase the technology readiness level for selected techniques. We will provide guidelines on the utilisation of key combinations of cyber-physical infrastructure layers, and on how design practice needs to change to accommodate scientific developments, when cyber-physical infrastructure can be integrated into the design of smart local energy sytems to realise added value.