Simultaneous measurement of flame temperature and absorption coefficient through LMBC-NNLS and plenoptic imaging techniques

Jian Li, Md. Moinul Hossain, Jun Sun, Yudong Liu, Biao Zhang, Christos Tachtatzis, Chuanlong Xu

Research output: Contribution to journalArticlepeer-review

8 Citations (Scopus)
6 Downloads (Pure)


It is important to identify boundary constraints in the inverse algorithm for the reconstruction of flame temperature because a negative temperature can be reconstructed with improper boundary constraints. In this study, a hybrid algorithm, a combination of Levenberg-Marquardt with boundary constraint (LMBC) and non-negative least squares (NNLS), was proposed to reconstruct the flame temperature and absorption coefficient simultaneously by sampling the multi-wavelength flame radiation with a colored plenoptic camera. To validate the proposed algorithm, numerical simulations were carried out for both the symmetric and asymmetric distributions of the flame temperature and absorption coefficient. The plenoptic flame images were modeled to investigate the characteristics of flame radiation sampling. Different Gaussian noises were added into the radiation samplings to investigate the noise effects on the reconstruction accuracy. Simulation results showed that the relative errors of the reconstructed temperature and absorption coefficient are less than 10%, indicating that accurate and reliable reconstruction can be obtained by the proposed algorithm. The algorithm was further verified by experimental studies, where the reconstructed results were compared with the thermocouple measurements. The simulation and experimental results demonstrated that the proposed algorithm is effective for the simultaneous reconstruction of the flame temperature and absorption coefficient.
Original languageEnglish
Pages (from-to)711-725
Number of pages15
JournalApplied Thermal Engineering
Early online date27 Mar 2019
Publication statusPublished - 25 May 2019


  • 3D reconstruction
  • flame temperature
  • absorption coefficient
  • plenoptic imaging
  • non-negative least squares

Cite this