To measure temperature field of power plant boilers with acoustic method, reconstruction algorithm with regularization method is presented in this paper and compares with the reconstruction algorithm of temperature field commonly used in the least square method. The main difference between this algorithm and the reconstruction algorithm used in the least square method is that this algorithm increases the reconstruction accuracy by applying a priori information of flame during the reconstruction of temperature field when there are few data. Computing simulation examines the effect of reconstruction of temperature field with different temperature distribution functions of this algorithm and the least square method, as a result, this algorithm is more accurate than the least square method. This algorithm can implement 2-D temperature field reconstruction in power plant boilers with the algorithm fast and accurately without measuring a lot of data.