Fast prediction of complicated temperature field using Conditional Multi-Attention Generative Adversarial Networks (CMAGAN)
作者：
Highlights：
• Widely applicable strategies of temperature prediction via computer vision technology.
• Presenting a conditional multi-attention module inspired by self-attention module.
• Obtaining temperature field rapidly from the structure by building a new GAN model.
• Improving stability, accuracy and overall perception compared with former studies.
• Exploring our GAN's performance in expressing thermodynamics phenomenon.
摘要
•Widely applicable strategies of temperature prediction via computer vision technology.•Presenting a conditional multi-attention module inspired by self-attention module.•Obtaining temperature field rapidly from the structure by building a new GAN model.•Improving stability, accuracy and overall perception compared with former studies.•Exploring our GAN's performance in expressing thermodynamics phenomenon.
论文关键词：Heat transfer,Generative adversarial network,Temperature field,Multi-attention,Self-attention
论文评审过程：Received 3 November 2020, Revised 3 August 2021, Accepted 4 August 2021, Available online 12 August 2021, Version of Record 16 August 2021.
论文官网地址：https://doi.org/10.1016/j.eswa.2021.115727
