Application of the improved BP neural network model to deformation analysis of an earth-stone dam

In recent years, the artificial neural networks theory and the application obtained the swift development. Especially the artificial neural networks BP model reflected the functional relations which need not to use the explicit function expression to indicate that but adapts through regulating network's weight and the error value, so we can avoid the error which caused by choosing an improper factor. Therefore, we use the BP model to analyze observed data of the dam which has become one newly research subject. But the BP model has some shortcomings, which causes its application to receive the very big limit. This article works in the foundation of the predecessor, proposed several corrective measures, and apply it in the analysis of some reservoir earth-stone dam distortion observed data. Finally, the analysis achievement indicated that these corrective measures have large useful value.