Abstract
A hybrid method combining Tikhonov regularization and generalized singular value decomposition (TR–GSVD) was proposed to reconstruct temperature distributions in radiant enclosures. The regularization parameter, which is crucial to accurately inverse temperature distributions, could be fast updated in TR–GSVD. Numerical reconstruction of three-dimensional temperature fields in a 10 m × 10 m × 20 m furnace proved TR–GSVD possesses the same computational accuracy but higher efficiency compared with TR. Experimental reconstruction results from practical flames in the furnace also showed the regularization parameter changed significantly in some combustion conditions. It is necessary to update the regularization parameter during the reconstruction.