AIPE — AI for Power Engineering
AIPE connects research on device modelling, converter optimisation and open engineering tools. The aim is to make engineering data and models useful within an inspectable design workflow.
AI-Assisted Semiconductor Device Modelling
A modelling and assessment workflow was developed to link fitted device characteristics to converter losses. The SiC static ANN study explains the network branches, training and a buck-converter comparison up to 3 kW. Semiconductor modelling overview.
Multi-Objective Converter Optimisation
An integrated component-modelling and converter-design framework was proposed, and its development continued through AIPE. Explore the design framework.
Engineering Design Agents
The public converter optimisation repository connects component calculations, optimisation and surrogate modelling as a focused example of AI-assisted engineering tools.
AIPE PCB Agent connects structured requirements to KiCad schematics, PCB layouts, checks and exports. The project page introduces its automation framework and a 5 kW SiC buck converter with separate power and DSP control boards.
AIPE CAD Agent builds parametric FreeCAD models of transformers, heat sinks, cold plates and converter assemblies, and checks them for interference, fit and clearance. The project page introduces its structure and the mechanical models of a 10 kW DAB converter.
AIPE Sketcher turns circuit descriptions into SVG schematics and checks each drawing by reading its connections back from the wires. The project page introduces its structure and a dual-active-bridge schematic.
AIPE Field Simulation drives Gmsh and Elmer to solve thermal, magnetostatic and eddy-current fields, checking analytical estimates against verified finite-element results. The project page introduces its structure and the heat sink of a 10 kW DAB converter, solved from the CAD Agent’s model.
AIPE IEEE Paper Agent turns a research idea and the evidence recorded by the design agents into an evidence-linked paper plan and a manuscript workspace on the official IEEE template of the chosen venue, without inventing results or references. The project page introduces its structure and a paper plan for the 10 kW DAB benchmark.