A new method for constructing spectral emissivity models for measuring the real temperature of targets

The main problem in radiation pyrometry is the large error arising from the unknown or varying emissivity of target surface. In this paper, we design a method of combined neural network to set up emissivity model, we call Combined Neural Network Emissivity mode (CNNE model), which allows calculate true temperature and emissivity of any examined body from the measured data of continuous spectral emissivity. We optimize the training algorithm of the model by proposing single parameter dynamic search algorithm using the hybrid steepest desecent and Newton method. This optimization algorithm can quicken the constringency speed of CNNE model. Through theoretical derivation and Simulation experiment, we can see that this algorithm is useful for any target, and the measurement precisionis is very high.