|
| Solving the inverse heat conduction problem based on data driven model |
| Received:January 15, 2021 Revised:March 15, 2021 |
| View Full Text View/Add Comment Download reader |
| DOI：10.7511/jslx20210115004 |
| KeyWord:machine learning data-driven inverse heat conduction problem pipe inner wall identification tumor growth parameter estimation |
| Author | Institution |
| 陈豪龙 |
清华大学 航天航空学院 应用力学教育部实验室, 北京 |
| 柳占立 |
清华大学 航天航空学院 应用力学教育部实验室, 北京 |
|
| Hits: 1022 |
| Download times: 636 |
| Abstract: |
| Inverse heat conduction problems (IHCP) have good application values in engineering.In this paper,a data-driven model is developed to solve the IHCP,such as the inner wall identification of the pipe and growth parameter estimation of the skin tumor.For the pipe inner wall identification problem,we used a stochastic model and the finite element method to solve the direct heat conduction problems.By using the effective thermal conductivity transformation,the relationship between the measurement temperature and effective thermal conductivity is established by a machine learning model and the shape of the inner wall of the pipe is identified.Then the data-driven model is used to identify the growth parameters of the skin tumor and the influences of different measurement errors on the results are discussed.The numerical results show that the proposed method can accurately estimate the heat generation rate and blood perfusion rate of the tumors.They also show that the data-driven model has broad application prospects in solving the IHCP. |