The experimental and modelling research on the parameter-dependence nonlinearity of two key parameters (integration time and responsed gray level) of the CCD detector was carried out by using the self-developed CCD spectroradiometer. By giving the CCD′s characterization function of differential nonlinearity, the nonlinear correction was achieved for automatic integration time adjustment measurement of CCD spectroradiometer in multi-scale on the model function. The response characteristics of two-parameter adjustment were tested by the laser beam incidence, and the nonlinear characteristics of CCD photoelectric conversion were analyzed. The data coordinates is reconstructed according to the structural features of the data sets curves, and the fitting model of nonlinear change relationship with the dependent parameters of CCD is built in the new representation coordinates. The results show that, in the multi-scale range, the model can describe the precise functional relationship of the nonlinear deviation caused by CCD spectroradiometer through two parameters, that is to say, the precise quantitative characterization of the two-parameter correlation nonlinear deviation of CCD photoelectric conversion is achieved. The comparison between the response value after nonlinearity corrected by the fitting model and the monitoring value of power meter shows that the model can effectively reduce the nonlinearity deviation to less than ±2%.