20 POWER ELECTRONICS DESIGN https://www.celus.io/ Issue 3 2026 Power Electronics Europe www.power-mag.com Using AI to Successfully Navigate the Complexity of Power Electronics Design By Tobias Pohl, CEO, CELUS Power electronics design is a demanding engineering discipline. As modern systems look to blend efficiency and power with more compact form factors, engineers must balance performance with thermal behavior, electromagnetic compatibility, component availability, cost, and physical constraints. Artificial intelligence (AI) cannot remove these challenges, but it can help designers navigate them more effectively. Power electronics has always involved trade-offs: a component that performs well electrically may introduce thermal challenges; increasing the switching frequency can reduce the size of transformers and inductors but promote greater electromagnetic interference (EMI); a device selected for efficiency may have different implications for cost or system lifetime. These trade-offs are becoming more complex as engineers work with higher power densities and increasingly capable semiconductor technologies. Silicon carbide (SiC) and gallium nitride (GaN), for example, enable faster switching and new design possibilities, but can also make factors such as switching behavior and layout, along with parasitic effects, far more significant. At the same time, engineers must work through an expanding volume of datasheets and application information, as well as component data. This is where AI can have a particularly useful role: not by taking engineering decisions away from the designer, but by making the information and design process easier to navigate. From Component Data to Engineering Insight Selecting components for a power electronics application is rarely just about finding a device with the right voltage and current ratings. The engineer will also need to consider factors like switching characteristics, losses, thermal performance, package requirements, availability, lifecycle, cost, and compatibility with the wider design. A suitable power electronics device therefore emerges from the interaction between multiple requirements rather than from a single specification. AI can help by making this information easier to interrogate. Instead of manually searching through seemingly endless datasheets and comparing individual parameters, an engineer can take advantage of natural-language interaction to investigate a requirement or identify potential candidates. AI not only helps with surfacing relevant devices, but also with comparing their characteristics and identifying information that may require closer attention. This capability is particularly useful during the early stages of a power electronics project, when the design space is still open. Rather than spending significant time narrowing the options manually, engineers can investigate several potential approaches before committing to a particular implementation. Crucially, AI is merely helping to explore the design space; the engineer remains responsible for deciding which characteristics matter and whether a proposed component is genuinely appropriate for the application. Interconnected Trade-Offs The potential value of AI becomes even clearer when considering the wider behavior of a power design. A change in one part of the system can propagate through several other design considerations. Increasing switching frequency, for example, can influence efficiency, switching losses, thermal behavior, EMI, and the requirements placed on passive components. Likewise, changing a
RkJQdWJsaXNoZXIy MjQ0NzM=