Building a Hierarchy of Digital Twins

Especially when modeling or tracking complex systems, a set of of real-time digital twin models can organize their behaviors in a hierarchy at multiple levels of abstraction, from device handling to strategic analysis and control. This enables their digital twin objects to implement analytics at multiple levels by exchanging messages within the hierarchy.

Consider a streaming analytics application that analyzes telemetry from the components of a wind turbine. This application initially uses three real-time digital twin models for blades, generators, and control panels. Digital twin objects created from these models can receive telemetry from their respective physical components and combine this with relevant contextual data, such as the component’s specifications, service history, and dynamic state, to enhance its ability to predict impending failures. In a real-world deployment, many instances of these objects would be created to track a large wind farm.

The following diagram illustrates how the digital twin objects created from the three models receive telemetry from three components of a hypothetical wind turbine (blades, generator, and control panel). Message-processing code analyzes the telemetry and generates feedback and alerts:

In this hypothetical example, the blades and generator work together to generate power managed by the control panel. Taking advantage of a hierarchical organization illustrated below, the real-time digital twin objects for the blades and generator could feed telemetry to a higher-level real-time digital twin object implemented by a Blade System digital twin model. This object manages the rotating components within the tower and their common concerns, such as avoiding over-speeds, while not dealing with the detailed issues of directly managing the two components. Likewise, the real-time digital twin object for the blade system and the control panel feed telemetry messages to a yet higher-level object implemented by a Master Control digital twin model. This object coordinates the overall wind turbine’s operation and generates alerts as necessary. Note that Blade System and Master Control objects do not communicate directly with physical data sources.

By partitioning the application into a hierarchy of digital twin models, the code can be modularized and thereby simplified with a clean separation of concerns and well-defined interfaces for testing.