| 研究実績の概要 |
In continuation of previous work and in line with plans, the focus of research this year continued to be on understanding thermophysical properties of liquids as well as understanding how statistical methods could be applied to understand their properties. In particular, it is of importance to understand how modern machine learning techniques can be made applicable to modern thermal design of such liquids, so much focus this year was on this area.
It was found in particular that existing machine learning methods, while interesting from a predictive point of view, were insufficient to explain the physical mechanisms, which is important for scientific understanding of the results obtained and therefore for subsequently predicting the properties in a previously unseen environmental situation. It is for this reason that the research went beyond simply trying to predict the outcome but also exploring the ability of causal inference techniques to interpret the results.
The results of this exploration yielded interesting insight into the applicability of such methodology to a range of situations, not least the thermal properties of liquids that are important for modern coolant technology.
|
