Abstract: The traditional GM (1, 1) model was not high accuracy and strong adaptability for deformation forecast, the paper improved traditional GM (1, 1) from multiple angles, and integrated the traditional GM (1, 1) and its improved models, established prediction model base. By using gray evaluation model to evaluate the prediction results of model base, so that the intelligent prediction was achieved; and the forecast model base and evaluation model were achieved-by using programming language. Finally, the model base was test by Pangushan Tungsten Mine deformation data; the result indicates that the model base can predict the different characteristics of data and intelligently get the best prediction results; and has important application value

Key words: gray theory, model base, deformation intelligent prediction, GM (1, 1)