Skip to content

idrl-lab/recon-data-generator

 
 

Repository files navigation

Data Generator for Temperature Field Reconstruction of Heat Source Systems

This project is based on FEniCS and used for data generation of temperature field reconstruction samples. This data generator mainly supports the paper "A Machine Learning Modelling Benchmark for Temperature Field Reconstruction of Heat-Source Systems"

Samples

1

Support functions

  • Configurations
    • Size of board
      • 2-D
        • length
        • width (default: equal to length)
    • Scale Number
      • mesh grid(nx
    • Boundary Conditions(bcs
      • Heat Sink (Dirichlet BC)
      • Sine function boundary (Dirichlet BC )
      • default (Neumann BC)
    • Components
      • type
        • rectangle
        • circle
        • capsule
        • triangle
      • size of units
        • length
        • width
      • powers
        • constant
        • one from a given set
        • uniform sampling
      • 功率类型
        • 固定功率
        • 高斯分布功率
      • 组件位置和角度
        • 组件中心位置坐标(positions)
        • 组件放置角度(angles)
      • 组件数量
    • 存储格式
      • mat格式
    • 测点选取策略
      • random
      • uniform
      • center
      • from_mat(从mat文件中读取,默认读取变量为u_pos)
    • 特殊样本
      • 打开关闭(special)
      • 特殊组件数量(功率为0的组件数量,全部组件情况为所有组件功率相同)

Installation

本生成器依赖 fenics 作为有限元求解器,可参照 fenics 安装文档,推荐以下两种方式安装,如果没有 docker 使用经验推荐 Anaconda 方式

  • Anaconda (LinuxMac)

    1. 使用 conda 创建并激活环境
    conda create -n fenicsproject -c conda-forge fenics mshr
    source activate fenicsproject
    
    1. use pip to install the released version

      • pip install -U recon-data-generator
      • or use unreleased version from master branch
      pip install -U git+https://github.com/shendu-sw/recon-data-generator.git
      
  • Docker (Linux, Win, Mac)

FAQ

  • Windows 下可以使用 Docker 方式安装,或在应用商店安装 Ubuntu WSL

  • pip 安装前可使用国内源如清华

    pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
    
  • 仅支持 Python3.6 以上版本

  • 如果按照以上 anaconda 安装方式,别忘了切换到 fenicsproject 环境

Visualization

  • 可视化生成配置文件入口config_generate.html
  • For example:
image-20210824224114153 Example0Example1
heat source system visualization samples by data generator

Easy Starting

  • data generate执行如下命令
recon_data_generator generate
  • plot执行如下命令
recon_data_generator plot

Temperature Field Reconstruction Dataset (TFRD)

  • Baidu netdisk: link, Password: tfrd

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 95.7%
  • HTML 3.7%
  • Dockerfile 0.6%