Optimum Identifications of Spectral Emissivity and Temperaturefor Multi-Wavelength Pyrometry

The main problem of the traditional radiation pyrometry is that fatal errors will be caused by the unknown or varying emissivity. Based on the combined neural networks (CNNE model), we propose an improved method for emissivity modeling. The model structure and the optimum algorithm are described. This method being used, the spectral emissivity and temperature can be fast computed accurately from the spectral radiation measured.