Artificial intelligence (AI) and machine learning are becoming an increasingly integral element in digital twins to maximise their benefits and enable their use at scale.

This brings value not only in industries such as manufacturing but also in key areas such as health, urban planning, advanced energy management, and climate action and preparedness.

AI also supports connected digital twins[1] which unlock more advanced insights and capabilities than single-asset digital twins – this is key for complex issues that span multiple sectors.

Some examples of AI working with digital twins today include: improving infrastructure and traffic flow at a busy transport hub[2]; boosting manufacturing efficiency and warehouse automation[3];  wind power forecasting for energy trading[4]; accelerating asset commissioning[5]; and managing the impact of flooding on energy, water and telecoms networks[6].

With greater use of AI and integration of connected digital twins, further potential[7] has been identified in areas such as monitoring ice sheets to inform climate science and predict impacts; better personalising medicine; and reaching net zero goals[8].

The role of AI in digital twins

AI advances digital twins in several important ways[9]:

• Data analysis: At the heart of digital twins is data. AI algorithms can analyse vast amounts of data at high speed to spot patterns, trends and anomalies.

• Predictive analytics: By analysing historical and real-time data, AI models can forecast future outcomes and potential problems, such as when a machine will need maintenance or how busy a road is likely to be at a certain point in time.

• Optimisation: Digital twins can leverage AI to optimise processes and resources to improve performance and efficiency.

• Decision support and decision intelligence: AI uses data in digital twins to better predict the future and help humans make decisions based on all available data.

• Autonomous control: AI can enable digital twins to autonomously control physical systems or processes, using data to understand changing conditions and constantly optimise performance.

• Generative AI: Generative AI is also expected to play a growing role in digital twins. Key use cases include improving the ability to create simulations and interactive environments, enabling real-time queries and interactions through natural language processing[10], and advanced scenario generation capabilities. Another example is the use of generative modelling[11] such as Generative Adversarial Networks (GAN)[12]. These sophisticated AI models deploy neural networks to simultaneously generate and evaluate data, refining the generation process over time to potentially enhance the richness and diversity of data in digital twins.

Challenges

Challenges remain[13] with the successful implementation of AI in digital twins across various sectors, such as limited data, model evaluation, and infrastructure design. These tie in closely with some of the challenges facing digital twins[14] more broadly, such as data issues, skills, and organisational readiness.

Examples

• Climate Resilience Demonstrator (CReDo)

• Building a digital twin specification: Using wind energy to extract green hydrogen from seawater

• Remapping China’s commute with digital twins

• Scaling wind power forecasting for energy trading

• Power forecasting for wind farm maintenance scheduling optimisation

• Assuring a digital twin

• Transforming commissioning with digital twins

References

[1] Bennett, H. et al. (2023) Towards ecosystems of Connected Digital Twins to address global challenges. Available at: https://zenodo.org/records/7840266 (Accessed: 22 February 2024).

[2] Unreal Engine/Epic Games (2021) Remapping China’s commute with digital twins. Available at: https://digitaltwinhub.co.uk/case-studies/remapping-china’s-morning-commute-with-digital-twins-r48/ (Accessed: 22 February 2024).

[3] Connected Places Catapult Falcon Digital Twin Integration Platform. Available at: https://digitaltwinhub.co.uk/case-studies/falcon-digital-twin-integration-platform-r53/ (Accessed: 22 February 2024).

[4] Jungle Scaling Wind Power Forecasting for Energy Trading. Available at:  https://digitaltwinhub.co.uk/case-studies/scaling-wind-power-forecasting-for-energy-trading-r42/ (Accessed: 22 February 2024).

[5] Mott MacDonald Five Fords Treatment Plant: Transforming Commissioning with Digital Twins. Available at: https://digitaltwinhub.co.uk/case-studies/five-fords-treatment-plant-transforming-commissioning-with-digital-twins-r21/ Accessed: 22 February 2024).

[6] Connected Places Catapult (2022) Written evidence submitted by CReDo. Available at: https://committees.parliament.uk/writtenevidence/106677/pdf/ (Accessed: 22 February 2024).

[7] Willcox, Karen (2022) How "digital twins" could help us predict the future. Available at: https://www.ted.com/talks/karen_willcox_how_digital_twins_could_help_us_predict_the_future/    (Accessed: 22 February 2024).

[8] Digital Twin Hub. What is CReDo? Available at: https://digitaltwinhub.co.uk/credo/credo/ (Accessed: 22 February 2024).

[9]  Emmert-Streib, Frank. (2023) What Is the Role of AI for Digital Twins? Available at: AI | Free Full-Text | What Is the Role of AI for Digital Twins? (mdpi.com) (Accessed: 22 February 2024).

[10] Marr, Bernard (2023) Digital Twins, Generative AI and the Metaverse Available at https://www.forbes.com/sites/bernardmarr/2023/05/23/digital-twins-generative-ai-and-the-metaverse/?sh=a14623573627 (Accessed: 22 February 2024).

[11] Emmert-Streib, Frank. (2023) What Is the Role of AI for Digital Twins? Available at: AI | Free Full-Text | What Is the Role of AI for Digital Twins? (mdpi.com) (Accessed: 22 February 2024).

[12] Tsialiamanis G, Wagg DJ, Dervilis N, Worden K. (2021*) On generative models as the basis for digital twins.* Available at: https://www.cambridge.org/core/journals/data-centric-engineering/article/on-generative-models-as-the-basis-for-digital-twins/331809017A261E6A63402D10C418F9B8 (Accessed: 22 February 2024).

[13] Emmert-Streib, Frank. (2023) What Is the Role of AI for Digital Twins? Available at: AI | Free Full-Text | What Is the Role of AI for Digital Twins? (mdpi.com) (Accessed: 22 February 2024).

[14] Digital Twin Hub (2022) Digital Twin Roadblocks Available at: https://digitaltwinhub.co.uk/files/file/117-dt-roadblocks-full-report/ (Accessed: 22 February