A new method that incorporates AI is used to design power electronic converters

Artificial intelligence is being used to create an innovative and effective way to design power electronic converters.

Power converters and common uses

A power converter is an electrical device used for converting electrical energy. It can be used to convert alternating currents (AC) into direct currents (DC), convert DC into AC, or change the voltage or frequency of a current completely.

The voltage is the difference between two points of an electric field or current, while the current is the rate of flow of the electrons. In other words, the voltage causes the current. Currently, power converters are used in devices such as mobile phones, televisions and computers.

The study was published in the journal IEEE Transactions on Power Electronics.

AC and DC currents

Alternating currents (AC) are electric currents that can reverse magnitude and direction within intervals of time, while direct currents (DC) can only flow in one direction. Often, both are used in power supplies and can flow through various conductors and semiconductors. However, AC currents are better at building electric generators and motors due to its efficiency and easier ability to change voltage, when compared to DC currents.

Current methods used to create converters

There are present techniques used to create power converters. These methods rely largely on mathematical models, which can be both costly and timely. The current method requires complex numerical calculations that takes up computational time and requires a complicated design process.  A well-made converter needs to be lightweight, highly efficient, and have both a low expense and low failure rate.

The purpose of constructing a power converter is to make sure it meets the criteria of being designed appropriately, while also identifying what could be traded off among its performance, to improve the competence of the converter.

Using artificial intelligence to select a new design model

The researchers studied artificial intelligence and the ways it could be used to create a design model for power electronic converters. The technique uses a type of artificial intelligence (AI) called artificial neural networks (ANN), which uses algorithms and computing systems to imitate the interlinked neural networks of the human brain.  The AI-based ANN was trained using a dataset that consisted of more than 2,000 designs. The team then selected the most suitable layout to produce the desired productivity and power intensity.

Components of the ANN design

Researchers decided to use four components to design the power electronic converters with algorithms from the artificial neural networks. The four components were comprised of  elements that included field-effect transistors (FETs – a semiconductor that controls the movement of current using an electric field), inductors, capacitors, heat sinks, and the power gallium nitride (GaN).

The team performed tests on a GaN-based single-phase inverter that was produced using the AI inspired design to corroborate the accuracy and efficiency of the process. The team also wanted to check to see if the new technique was designed within the range of current devices, allowing it to be both competitive and commercially practical.

“Accurate and fast transient modeling/simulation approaches are essential to efficiently and to rapidly optimize the performance of wide bandgap power electronics systems,” said Dr. Wenlong Ming, senior lecturer at Cardiff University and senior research fellow at CSA Catapult. The research team wants the ‘automated power electronics design optimization’ to be a viable method for creating the power electronic converters, using inspiration from artificial intelligence.