Highly efficient energy conversion through the use of AI
Project title
Highly efficient energy conversion through the use of AI - New ways of increasing the efficiency of converter output stages using artificial intelligence
Brief description
Power electronics is a key to achieving a CO2-neutral society by 2050. In order to achieve CO2 neutrality, energy generation in the wind and solar energy sector must be expanded and the challenges of converting transportation to CO2-neutral drives must be solved. In addition, heat generation in households, e.g. via heat pumps, must also be CO2-neutral. Examples of this include hydrogen production and electricity generation using fuel cells. Power semiconductors are used in all the applications mentioned above. In future, all energy generated from renewable sources will pass through several such power-electronic converter stages before it is used in the target application. Poor efficiency requires a very large amount of additional energy generation equipment. An extremely high efficiency of these converters is therefore essential. At the core of these converters are power electronic switches, e.g. made of silicon, silicon carbide or gallium nitride. The latter two are new materials with outstanding physical properties in terms of on-state resistance and switching speed. However, these properties can only be utilized by means of new mechatronic construction technologies for power modules and connections to intermediate circuit capacitors. The power modules must be optimized in terms of their electrical and thermal properties. The same applies to the connections and busbar systems through to the capacitor. Until now, the geometric design/optimization of these complex systems has mostly been carried out in the classic way using simulation with only a systematic linear parameter variation. This means that the positions of chips or lengths and thicknesses of copper connections were systematically varied. In this case, the complexity of the changes is very limited and thus also the finding of an optimal structure. In the course of this project, the development process for highly efficient power modules is to be optimized with the help of artificial intelligence (AI). In other words, an AI should find the optimum positions, lengths, etc. In addition, the development process is to be accelerated through the use of AI. The challenge is to combine such complex structures in the simulation program with an AI algorithm. On the other hand, suitable AI algorithms and optimization criteria must be determined. This new approach is a small revolution in the development of power electronic converters and will lead to a further increase in efficiency, which will contribute to the success of the energy transition.
