Abstract: Briefly describes the modeling procedure of the Dynamic Data System Method. By means of the Faster Fourier Transform, a set of engineering data is processed and the frequency spectrum obtained. The set of data is thereafter resampled and treated with the DDS method. Since the results of modeling different samples of the same stochastic process are distinct, a modeling program for multiple-sample data is put forward. Using this new method, a model satisfying different samples is obtained and proves itself efficient .Typical application cited shows that the error variances of azimuth and pitch angle are reduced by 70% and 84% respectively after filtering using the model.The improved modeling procedure for Dynamic Data System method can be used to process other cases involving multiple-sample data.