Smart Grids and Distributed Intelligence

Digitalisation as the focal point of modern electricity grids

The energy sector is undergoing an unprecedented transformation driven by digitalisation, the integration of distributed renewable sources and the growing demand for reliability and efficiency. Smart grids represent the technological response to these challenges, incorporating advanced monitoring, self-control and autonomous decision-making capabilities.

In this context, it is important to mention Industry 4.0, also known as the Fourth Industrial Revolution, which integrates new trends in digital technologies, automation, high-speed information exchange and intelligent system creation in various sectors. These systems can exchange information, perform actions to optimise the system as a whole, and manage equipment increasingly autonomously.

Specifically in the energy sector, data-driven technologies and approaches are integrated to optimise energy production and productivity and facilitate the development of a more sustainable energy ecosystem.

Among the different technologies and approaches that Industry 4.0 encompasses, the following stand out: the Internet of Things (IoT), cloud computing, optimisation, edge computing, cybersecurity, and digital twins.

New generations of electrical systems are tending towards the use of IoT devices that have their own IP address, allowing them to connect to the grid and provide services via the web. To prevent unauthorised access or possible information leaks during operation, it is essential to integrate cybersecurity measures.

Additionally, due to the increased calculation capacity of computer systems, the data collected by sensors can be sent to the cloud for processing and analysis. Alternatively, depending on the speed requirements based on these measurements, part of this processing can be carried out directly at the edge of the network (Edge Computing), allowing action to be taken in compliance with strict real-time and low latency requirements.

On the other hand, digital twin technology is one of the most promising technologies for improving efficiency, maintenance, and continuous monitoring of systems.

Digital Twin as the standard for monitoring and self-control

The definition of digital twin (DT) has evolved over time, with the first terminology established in 2003 by Michael Grieves [1]. Between 2003 and 2011, advances in communications and sensor technology, simulation techniques and big data analysis allowed this idea to be developed further. In 2012, NASA formalised the definition of this concept as ‘an integrated, multi-scale, multi-physics, probabilistic simulation of a vehicle or system as built, using the best available physical models, sensor updates, fleet history, etc., to reflect the lifetime of its corresponding physical twin’ [2].

Similarly, over the last decade, its description has continued to change, always focusing on the idea of representing a physical system dynamically, adapting to variations and operational conditions through the collection of information/data from acquisition systems online.

Given the generality of the definitions, there is considerable confusion in the literature about what a DT really is, and it is worth taking the time to explain the difference between the various approaches to digitising a system. There are three fundamental types: the digital model, the digital shadow and, finally, the digital twin.

• Digital Model: characterised by being a digital representation of a physical system or object in which there is no interaction with the physical entity in question. An example of this could be a static model of a synchronous machine in simulation environments such as Matlab, PLECS, etc.
• Digital Shadow: refers to a type of model that replicates the behaviour of the physical system by feeding on physical measurements of the real entity. There is only one-way communication between the two objects. This definition fits quite easily with the origins of the digital twin.
• Digital Twin: is generated when information flows bidirectionally between both entities, allowing for complete integration. In this scenario, any modification in the physical system triggers an automatic adjustment in the digital counterpart and likewise, the virtual object produces changes or actions on the physical system to improve its overall performance.

Currently, ‘intelligent DT’ is starting to be developed, where this new intelligence integrates characteristics such as proactivity, real time and predictability into the digital model. In other words, it could be summarised as the ability of the virtual replica to work actively and online, providing relevant information when necessary, with a specific objective and the ability to anticipate situations in order to modify the behaviour of the real entity. Advances in AI simplify the implementation of this evolution of the concept.

Levels of abstraction and distributed intelligence: the beginning of autonomous nodes

The digital replica modelling approach is fundamental, especially in the context of the electricity sector, where there are different levels of resolution, available computing power and abstraction to represent the most significant parts of the physical entity. If the DT provides its output within a guaranteed time frame, below its execution step, it can be said that the replica has real-time capabilities. Furthermore, if this period is well below the execution time interval, it means that the twin can provide relevant information sufficiently in advance to be used in control or scheduling decisions. Conversely, if the calculation of results substantially exceeds operating times, it remains only suitable for planning and design applications.

On the other hand, depending on the fidelity of the model, these digital twins can be classified as topological, static, or dynamic.

• Topological: tthey only describe the structure of the system and the interconnection between the different components. An example could be a map of a particular factory.
• Static: model based on algebraic equations, inputs, states, and configurable outputs that emulate a specific operating point. Another example in this case could be an economic model of a specific market.
• Dynamic: equations or code that provides a description of a dynamic system. This could be the thermo-electrical modelling of an engine.

Based on the relations in the execution and form of representation of the model, different variants of twins can be glimpsed.

Another crucial issue to consider is that the deployment of distributed intelligence represents a paradigm shift in electricity grid management. Instead of relying on a single control centre, decision-making and data processing are decentralised, allowing elements located at the periphery (such as converters, inverters and other electronic devices) to act autonomously as smart nodes.

The main advantages of this approach are:

• Scalability: The grid can grow and adapt without overloading the central control.
• Robustness: A failure in one node does not compromise overall operation, improving fault tolerance.
• Response times: Local decisions are made in milliseconds or microseconds, which is key to managing critical events.
• Local optimisation: Each node can efficiently manage its resources and contribute to the overall balance.

As in most cases, the art lies in striking a balance. Depending on the computing capabilities and the objective of the twin, it can be integrated autonomously, running in parallel with the electrical node, as in the control system of a power converter, or deployed in the cloud to take advantage of the potential offered by current artificial intelligence.

Smart converters that integrate some form of DT as a whole allow them to act as autonomous nodes, evolving into smarter devices capable of operating self-sufficiently, enabling services such as the following, among many other functionalities:

• Real-time analysis of the status of the local and global electricity grid.
• Make autonomous decisions about energy management (e.g. storage, injection or disconnection).
• Communicate with other nodes to coordinate actions and optimise energy flow.
• Perform predictive maintenance on the most critical components, which are the most expensive to replace.
• Participate in digital energy markets.

This distributed architecture reduces dependence on central control and enables more efficient management, especially in grids with high penetration of renewables and distributed resources.

Applications in the electricity grid, case studies, future prospects

Digitisation and in particular DT in conjunction with the latest developments in AI, allows a multitude of services/applications to be integrated into the centre of power converters. The following are some relevant applications:

1. Advanced real-time monitoring: DT enables each converter to report its status with unprecedented granularity, including monitoring the correct operation of the equipment and comparing the outputs of the physical and virtual models to find discrepancies and act accordingly.

2. Predictive maintenance of components and service life management: the digital replica is capable of merging real data with degradation models that enable the detection of faults in IGBTs, ageing of DC capacitors, loss of control capacity due to thermal degradation in inductors, etc.

3. Real-time analysis of system stability: the versatility of the services/applications that a DT can integrate through the implementation of multiple layers allows algorithms such as the analysis of converter stability margins when connecting to a specific network to be introduced. This possibility is fundamental in the mass integration of power converters, as it allows the optimisation of control laws and the modification of control bandwidth to adapt to complex scenarios and weak networks.

Digitisation based on the integration of autonomous DTs is currently a key issue with high penetration in the energy sector industry, with several companies developing multiple lines of work.

General Electric (GE) holds four patents directly related to DTs. Two of these are specifically associated with wind farms [4]. The company conceptualised a twin for a wind farm that incorporates two communication grids. The first grid establishes a connection between the control systems of the individual wind turbines in the farm. The other grid similarly interconnects the digital models. These virtual replicas are constantly updated based on the data recorded by the first grid. This system provides real-time monitoring of the operational status of the wind turbines using sensors, while also allowing their functions to be managed through the digital models.

Siemens leveraged this technology in the field of electrical systems and wastewater treatment plants. The twin was designed with the aim of improving the planning, operation and maintenance of an electrical system in Finland. This initiative introduced substantial improvements in automation, data usage and decision-making processes [5].

Another example can be seen at IBM, where digital replicas were used in autonomous vehicles to analyse vital parameters such as engine speed and oil pressure. This approach not only helps prevent breakdowns, but also encourages the development of a more efficient engine [6].

Despite all these advances, the transition to fully smart and distributed electricity grids faces significant challenges:

• Interoperability: It is essential to establish common standards to ensure communication and coordination between devices from different manufacturers.
• Cybersecurity: The increase in connected nodes increases the possibility of a cyberattack, requiring robust protection systems.
• Data management: The volume of information generated requires advanced storage, processing and analysis solutions.

On the other hand, the opportunities are equally significant: greater energy efficiency, massive integration of renewables, reduced emissions and a more resilient grid that is adaptable to future needs.

Conclusion: Summary and Perspective

Smart grids and distributed intelligence represent the cutting edge of energy transformation. The combination of prediction algorithms, advanced sensors, digital twins and decentralised control enables the creation of more efficient, resilient and sustainable systems. Autonomous converters and programmable electronics open the door to a new era in which each node actively contributes to the stability and optimisation of the grid. Although challenges remain, the future of energy lies in digitalisation, decentralisation and intelligent collaboration between devices, operators and users.

For professionals in the sector, understanding and mastering these technologies will be key to leading the transition towards a smarter, safer electricity system that is prepared for the challenges of the 21st century.

References

1. Grieves, Michael. (2016). Origins of the Digital Twin Concept. 10.13140/RG.2.2.26367.61609.
2.  E. Glaessgen and D. Stargel, “The digital twin paradigm for future nasa and u.s. air force vehicles,” 04 2012.
3. Sergio de López Diz, Digital Twin Technology for Enhanced Monitoring and Stability Assessment of DC/AC Three-Phase Power Converters, Thesis 2024.
4. J. A. C. Arnold M. LundKarl MochelJeng-Weei LinRaimundo OnettoJayanthi SrinivasanPeter GreggJeffrey Eric BergmanKenneth D. Hartling, “Digital wind farm system,” US20160333855A1
5. Siemens, “Siemens expands digitalization solutions for the process industries,” 2018. [Online]. Available: https://www.siemens.com/press/en/pressrelease/?press=/en/pressrelease/2018/processindustries-drives/pr2018030215pden.htm
6. IBM, “An engine can become a platform with a digital twin,” 2018. [Online]. https://www.ibm.com/internet-of-things/trending/digital-twin

Image of Jain Temple of Ranakpur | Rajasthan, India.