Material Design
The properties of the individual components play a major role in material development, as does the behavior of the ingredients among themselves. Materials research is thus a predestined application area for AI, since physical experiments and simulations are extremely time-consuming and costly.
-
Continuous description of the gas mixture
-
Fast results
-
Prediction instead of complex calculation
The Problem
Energy efficiency is more important than ever in today's world: modern glazing should therefore serve to reduce heating costs. Among other things, this is stipulated by the current Building Energy Act (GEG).
Insulating glass with appropriate gas filling saves energy. This often involves completely normal flat glass surfaces, between which there is a certain amount of low thermal conductive gas. Gases and gas mixtures have different thermal and acoustic conductivities. Determining the properties of different gas mixtures is possible quickly and efficiently with KI.
What we did
In order to train the AI model, data on the properties of the gases used were required. Data for the pure gases at different pressures and temperatures, as well as data from known mixing ratios, are used for this purpose. The AI can then be used to predict values for data that is not available. For this purpose, a neural network is trained in the AI tool dAIve for the specific use case.
The solution
The trained, optimized AI can then be used to fill databases with it, or to represent certain combinations. However, the AI model can also be implemented directly into the control system or used in engineering designs.
