Abstract: For reconstruction of three-dimensional temperature field in furnace of power plant boiler,a mathematical model based on acoustic theory was constructed.Two regularization algorithms based on Singular Value Decomposition (SVD) algorithm were proposed.Three typical temperature fields were simulated.Anti-noise ability of algorithms was tested by Gaussian noise with standard deviations.It indicates that regularized SVD algorithm is able to solve severely ill-posed reconstruction problems.Reconstruction temperature field reflects accurately temperature distributions,and algorithms have good anti-noise ability.TSVD regularization algorithm with faster reconstruction speed and better anti-noise ability is suitable for power plant boiler with complicated combustion.