Abstract: A three-dimensional temperature distribution reconstruction algorithm based on Tucker decomposition and acoustic temperature measurement was proposed. In the algorithm, numerical computation was used to establish a priori dataset, the main characteristics of the boiler temperature field were extracted using Tucker decomposition, and acoustic temperature measurement was combined with two-dimensional temperature field interpolation to reconstruct the three-dimensional temperature field. The algorithm could improve the reconstruction accuracy of complex temperature fields and have a fast reconstruction speed. Results show that the method can reconstruct complex three-dimensional temperature fields in about 10 seconds, and reduce reconstruction errors by more than 10% compared to traditional acoustic temperature measurement. It also has strong applicability for temperature fields outside priori operating conditions, and has some guiding significance for combustion optimization and load adjustment in coal-fired power plants.