Fast prediction of surface-temperature fields based on the thermal response function method

Fast prediction of surface-temperature fields is required in thermal protection, spacecraft thermal analysis, infrared signature prediction and iterative thermal design, where repeated transient evaluations make full-domain transient solvers expensive. This work proposes a physics-interpretable acceleration algorithm for surface-temperature prediction based on thermal response functions (TRFs). The algorithm predicts surface temperatures through linear superposition of TRFs precomputed from the target’s thermal model, giving it a direct physical interpretation. The thermal response function method (TRFM) implements this algorithm by reusing the TRFs for repeated predictions of a fixed target. In the three-dimensional tests, TRFM predictions agree well with the adopted FVM baseline. With direct FVM and TRF construction using the same discretization, a representative configuration achieves approximately 14– 42 × online speed-ups. Its combined offline–online cost becomes lower than that of direct FVM after 7–19 repeated predictions over the tested horizons. The results demonstrate that TRF reuse and boundary-only online calculations provide fast, physically interpretable surface-temperature prediction while maintaining close agreement with the corresponding FVM baselines.