Dijo-404/mhd-nanofluid-ev-thermal-dataset

基于物理的合成数据集，用于训练电动汽车电池热管理的机器学习替代模型，采用磁流体动力学（MHD）混合纳米流体冷却技术。数据集包含5,000个样本，通过拉丁超立方采样从MHD混合纳米流体流动的控制方程（连续性方程、纳维-斯托克斯方程+MHD、能量方程与粘性耗散/焦耳热）生成。输入特征包括哈特曼数（Ha）、纳米颗粒体积分数（φ）和入口流速（u₀）。输出特征包括最大电池表面温度、努塞尔数、归一化熵生成等12个指标。物理系统参数包括冷却通道尺寸、混合纳米流体成分、电池热通量和入口温度。数据校准误差在0.4-5%之间。

Synthetic physics-based dataset for training ML surrogate models for electric vehicle battery thermal management using MHD hybrid nanofluid cooling. Contains 5,000 samples generated via Latin Hypercube Sampling from governing equations of MHD hybrid nanofluid flow (continuity, Navier-Stokes + MHD, energy with viscous dissipation/Joule heating). Input features include Hartmann number (Ha), nanoparticle volume fraction (φ), and inlet flow velocity (u₀). Output features comprise 12 metrics such as maximum battery surface temperature, Nusselt number, normalized entropy generation. Physical system parameters include cooling channel dimensions, hybrid nanofluid composition, battery heat flux, and inlet temperature. Data is calibrated within 0.4–5% error against published results.