Efficient and robust prediction of internal temperature distribution and boundary heat flux in participating media by using the Kalman smoothing technique

Abstract

The Kalman smoothing (KS) technique, which is consisted of the Kalman filtering and Rauch-Tung-Striebel smoothing technique, is introduced to resolve the inverse radiation-conduction heat transfer problem by using the future temperature measurement information. For the forward problem, the discrete ordinate method is employed to solve the radiative transfer equation and the energy equation is resolved by using the finite volume method. The boundary time-dependent heat flux and internal temperature filed in participating media are retrieved in near real time from measurement temperature on the right surface. Different forms of time-dependent heat flux are imposed on the left surface to examine the performance of the proposed algorithm. All the reconstruction results indicate that the KS technique is effective and robust for resolving the retrieval of the boundary time-dependent heat flux and internal temperature fields in near real time. The effect of different parameters on the accuracy and stability of the reconstruction results, including the future temperature information, sampling interval, measurement noise covariance, initial state error, and initial state error covariance, are analyzed in detail. Compared with the KF technique, the reconstruction accuracy of KS technique can be improved obviously. Meanwhile, the time delay and oscillation of reconstructed heat flux are reduced significantly and the deviation of the retrieval temperature is also decreased greatly.