Application of improved GM(1,1) models to prediction of landslide deformation

For the purpose of heightening prediction precision and general applicability of GM(1,1) models to prediction of landslide deformation,mathematical characteristics of GM(1,1) model were analyzed and several practical improved methods that aimed at inherent limitation of conventional GM(1,1) model were discussed in this paper.With referenced typical deformation monitoring data of the Lianziya hazard rock mass(logarithmic) and the Huanglongxi landslide(exponential),prediction models of conventional GM(1,1),unbiased GM(1,1),center approach GM(1,1),reconstruct ground value GM(1,1) and gray-neutral network were established to study deformation prediction.The prediction results show that,aimed at different monitoring data sequence in mathematical natures of slope deformation,compared with conventional GM(1,1) model,some improved GM(1,1) models have higher prediction accuracy and general applicability.