Study On The Technique Of Information Fusion Applied To The Prediction Of Landslide

China is a country where landslide disaster occurred frequently and the disaster loss is very serious. According to the preliminary statistics, the economic loss caused by collapse and landslide is high to between 30 and 50 billion dollars every year. At present, it is impossible in our country to have enough economic and technical force to comprehensively control about latent risky landslide disaster spot. Under this condition, if the occurring time of landslide can be predicted timely and accurately, we may take measures and countermeasures against natural disasters as soon as possible so as to reduce the loss caused by these kinds of disaster to the lowest degree.Based on the above purpose, in this paper, aiming at two knotty questions in the present landslide prediction that the false noise affecting landslide prediction can't be wiped out and the displacement material of several monitoring stations don't be used sufficiently, geological property and mobile characteristics of landslide, the landslide information preprocessing and the application of multi-sensor information fusion theory are emphatically researched on the basis of widely consulting the domestic and foreign documentations and materials related to landslides forecast and multi-sensor information fusion. According to the geological property and mobile characteristics of landslide and principle of dynamics, models about landslide are built in order to describe the true deformation of landslide in the every stage of landslide's motion; on the basis of analyzing the defaults of traditional information preprocessing, state estimation and tracking technique of maneuvering target in modern estimation theory are proposed to use to de-noise the monitoring information of landslide for the first time. On the basis, according to state estimation theory in multi-sensor information fusion, displacement information of several monitoring stations of landslide is fused and comprehensive displacement information is extracted reflecting the whole deform of landslide. And then landslide prediction is done according to traditional prediction model base on displacement. At last, typical landslide examples are chosen to verify, analyze and summarize the methods proposed in this paper.