With the advent of new technological capabilities thanks to sensor-based condition monitoring in combination with predictive and generative artificial intelligence (AI), the fundamental role of TPM is evolving. Real-time data from systems and grids allow continuous condition monitoring, in which AI-assisted forecast models provide early identification of patterns that indicate impending disturbances or efficiency losses. Generative AI provides detailed troubleshooting guides, precise inspection instructions, and clear root cause analyses for potential faults. Combined with event-driven automation, control systems can respond immediately to deviations and take preventive action. In the energy sector, therefore, TPM becomes an integrative system that seamlessly brings together predictive maintenance, field service support, intelligent decision-making, and automated implementation.

Digital twins for total productive maintenance

A digital twin is a digital representation of a physical system that is continuously supplied with real-time data for the purpose of analysis, simulation, and optimization. The term is frequently misunderstood, however, because a digital twin is not necessarily a single software solution or primarily a visual representation of reality. Rather, it is a distributed, data-driven system that continuously monitors the physical system, models its behavior, and predicts future conditions. This serves as a basis to present proposals for action or to initiate automated processes. An AI-assisted TPM solution based on real-time data meets precisely these criteria because it integrates operating data, carries out analyses, and paves the way for predictive maintenance. It therefore fulfils the definition of a digital twin.

The researchers at the division of Fraunhofer IIS in Dresden are working on the technological basis for the development and implementation of digital twins. This involves developing methods for the creation of digital twins from structured and unstructured data. Based on these digital twins, the algorithms developed by Fraunhofer IIS offer the ability to monitor the condition of assets, optimize their management, and predict suitable maintenance times.

A digital twin for TPM should not be seen as monolithic; rather, it is a highly distributed system. Solutions of this kind consist of components located both in the data center or cloud and directly in operational technology (OT) systems in the edge environment – and also encompass integration with devices and sensors. This diversity of infrastructure calls for a platform that extends uniformly across all environments and serves as a consistency layer, ensuring stability, compatibility, and governance. At the same time, the platform must also be optimized for the specific requirements of the environments in question. Red Hat offers the market-leading hybrid/multi-cloud platform. The platform provides a universal consistency layer for innovation in situations ranging from on-premises to public and sovereign clouds and even remote edge locations. This allows the continual development, distribution, updating, and operation of AI models and automation workflows.

No single piece of commercial software is sufficient for a digital twin. Indeed, it is essential to consistently combine and harmonize the entire IT setup with the OT world to create an interoperable overall system. It is not enough to program isolated automation processes; rather, there is a need for a company-wide culture of cross-team automation. This culture should be underpinned by a central platform on which all teams can work together transparently in order to develop, optimize, secure, and control processes. An infrastructure-as-code approach therefore also simplifies and strengthens security and compliance governance. It is only through this integrative approach that lasting synergies emerge between IT and OT with a view to creating long-term added value. For this, Red Hat provides a platform that allows collaborative automation across various IT levels, thereby considerably improving the scalability, efficiency, and forward compatibility of digital twins.

AI and Automation Lab

Developing and implementing AI and automation solutions for total productive maintenance in the energy sector is a complex undertaking. The aspects discussed in this article reflect some of the challenges and opportunities involved, but the diversity of use cases and sector-specific requirements goes far beyond that. This is where the collaboration between Fraunhofer IIS and Red Hat comes into play: By working together, the partners offer comprehensive expertise on everything from infrastructure to the development of tailor-made solutions.

A workshop with researchers from the Engineering of Adaptive Systems division of Fraunhofer IIS and Red Hat serves as a starting point. This workshop provides energy companies with an opportunity to outline possible use cases, examine challenges in greater depth, and discuss suitable approaches to implementation.

When it comes to implementing specific projects, the cooperation partners provide a Joint AI & Automation Lab. While Red Hat provides the scalable platform infrastructure, Fraunhofer IIS handles project management and the research and development of AI and automation components. This close collaboration paves the way for the development of innovative and practical solutions with productive applications.