A Computational Method to Solve for the Heat Conduction Temperature Field Based on Data-Driven Approach
Kun Li,Shiquan Shan,Qi Zhang,Xichuan Cai,Zhijun Zhou
DOI: https://doi.org/10.2298/tsci200822165l
IF: 1.971
2021-01-01
Thermal Science
Abstract:In this paper, a computational method for solving for the 1-D heat conduction temperature field is proposed based on a data-driven approach. The traditional numerical solution requires algebraic processing of the heat conduction differential equations, and necessitates the use of a complex mathematical derivation process to solve for the temperature field. In this paper, a temperature field solution model called hidden temperature method is proposed. This model uses an artificial neural network to establish the correspondence relationship of the node temperature values during the iterative process, so as to obtain the "Data to Data" solution. In this work, one example of 1-D steady-state and three examples of 1-D transient state are selected, and the calculated values are compared to those obtained by traditional numerical methods. The mean-absolute error of the steady-state is only 0.2508, and among the three transient cases, the maximum mean-square error is only 2.6875, indicating that the model is highly accurate in both steady-state and transient conditions. This shows that the hidden temperature method simulation can be applied to the solution of the heat conduction temperature field. This study provides a basis for the further optimization of the hidden temperature method algorithm.