Application of BP Neural Network in Multispectral Thermometry

摘要

A new method was introduced for the data processing of a multiwavelength pyrometer by BP neural network, without the assumption between emissivity and wavelength. The experimental results show that the temperature differences between the calculated values and the theoretical ones are less than +-25K, and the values of the calculated emissivities are in good agreement with that of the theoretical emissivities. The method by BP neural network is an effective one for the actual temperature and emissivity measurements.