The temperature distribution in the furnace of power plant boilers is an important parameter to indicate the pulverized coal combustion state. The real-time and precise monitoring of the temperature field in the furnace is essential to ensuring the safe operation of the power plant and improving production efficiency. Acoustic thermometry is a typical noncontact temperature measurement and one of its cores is to derive the temperature distribution of the original temperature field by reconstruction algorithms. The existing temperature field reconstruction algorithms do not perform satisfactorily, and there are some problems such as incomplete reconstruction results, low reconstruction precision, and poor anti-interference ability. In order to further improve the reconstruction performance, an acoustic thermometry reconstruction algorithm based on logarithmic-quadratic (LQ) radial basis function (RBF) and singular value decomposition (LQ-SVD) is proposed in this article. This algorithm first uses the linear combination of the LQ RBFs to fit the reciprocal distribution of the acoustic velocity, and then uses the SVD method to solve the inversion model. The simulation results show that, compared with the commonly used algorithms, the proposed algorithm can obtain complete reconstruction results with significantly improved reconstruction precision, stronger robustness, and better anti-interference ability. In addition, the proposed algorithm also has good performance in the actual experiment, which verifies the feasibility and effectiveness of the algorithm in the engineering application.