In combination with the advantages of classical variational assimilation and regularization constraint, a multi-parameter regularization constraint assimilation method is studied. Unlike the weight of background and observation to the objective functional of classical variational assimilation, the regularization constraint can adjust the weight of observation term and give the weight function on the basis of Huber-estimator weight function during regularization parameter optimization. Simulation brightness temperature experiment is made for the water vapor channel of the Hyper-spectral Atmospheric Infrared Sounder (AIRS). The result shows that the variation method studied is better than the classical variational assimilation method. The influence of observation data on analysis fields is diagnosed on the basis of the degree of freedom for signals. The result shows that the method studied can effectively extract the brightness temperature information from the water channel of AIRS.