Beyond the hype how can AI work as a power design tool?
We use it to replace a complex and disconnected datasheet with a testable map of true performance. We can search a component library and estimate the benefit of every component with exceptional speed and accuracy. Furthermore, we can generate predictive models which identify trade-offs, accelerating the product's time to market and ensuring optimum component selection.
With our AI-powered tool, we can provide collaborators with the data needed to elevate their power electronics designs to the next level.
What are the implications for products?
Design tasks that are laborious and often take weeks to accomplish can now be undertaken swiftly. Tasks that would usually take weeks to complete can now be done in just minutes. Using Artificial Neural Network (ANN) modelling of circuit simulation training data we can reduce simulation time by up-to 99%.
The aim in developing of our AI-powered tool is to build a bespoke model, drawing on an extensive high fidelity device model library. We model a range of components, to select and simulate the best topology for the application. The outcome is that the efficiency, cost, power density and performance of a customer's built design can be accurately predicted with the model.
Why is an AI tool needed?
We know our customers need to make informed choices, at pace, to deliver their technologies. Our aim is to accelerate and optimise product development so that world class products can reach the market more quickly.
Our methodology
- Simulations: We simulate designs using our electrothermal and mechanical software tools
- Benchmarking: We use state-of-the-art products to guide design decisions
- Characterisation: We go beyond the datasheet to ensure the best components are used in the product
- Build support: We have an advanced packaging lab to help bring the design to life
- Validation: We can confirm the performance of the built design
How can I find out more?
Join our exclusive webinar, where Professor Ingo Ludtke, Head of Power Electronics at CSA Catapult and visiting professor at Cardiff University will present state-of-the-art machine learning techniques being used to improve device efficiency and transform real-world applications in industries ranging from data centres, rail, automotive and aerospace, to energy and defence.
Sign up for the event on Wednesday March 5th, (12:00-12:45) via the Eventbrite page.
