BubbleML
收藏资源简介:
BubbleML数据集是由加州大学欧文分校的研究团队创建,专注于多物理场相变现象的研究。该数据集通过物理驱动的模拟生成,提供了多种沸腾场景的准确地面实况信息,包括池沸腾、流动沸腾和过冷沸腾。数据集包含79个模拟,覆盖了广泛参数,如重力条件、流速、过冷水平和壁面过热。这些数据对于开发和比较最先进的机器学习技术和模型至关重要,特别是在热管理和能源效率等关键领域。此外,数据集还提供了两个基准测试场景,用于光学流分析和操作网络学习温度动态,进一步推动了机器学习在多物理场相变现象研究中的应用。
The BubbleML dataset was created by a research team at the University of California, Irvine, focusing on the study of multiphysics phase transition phenomena. Generated via physics-driven simulations, this dataset provides accurate ground-truth information for various boiling scenarios including pool boiling, flow boiling, and subcooled boiling. It contains 79 simulations covering a wide range of parameters such as gravitational conditions, flow velocity, subcooling level, and wall superheat. This dataset is crucial for developing and comparing state-of-the-art machine learning technologies and models, especially in critical fields like thermal management and energy efficiency. Furthermore, the dataset provides two benchmark test scenarios for optical flow analysis and operational network learning of temperature dynamics, further advancing the application of machine learning in the research of multiphysics phase transition phenomena.

- 1BubbleML: A Multi-Physics Dataset and Benchmarks for Machine Learning加州大学欧文分校 · 2023年



