An improved algorithm based on Radial Basis Function for temperature fields reconstruction in furnaces

According to a small number of average temperatures or time on the stagecoach acoustic path, it is a very difficult inverse problem to get the temperature distribution. The furnace temperature image is transformed by Discrete Cosine Transformation and the coefficient matrix Q is obtained, which is adopted as parameters of the furnace temperature image. Based on the linear relationship between the average temperature and coefficient matrix which is obtained by RBF (Radial Basis Function), the average temperature on each sound of stagecoach path has a corresponding coefficient matrix. The furnace temperature is restored by Inverse discrete cosine transform of this coefficient matrix. 128 test samples are generated to test the effect of RBF neural network. The analysis results show that the target output is similar to the actual one. The root mean square of the 128 sets of test samples locates in 1%-4%.