Study On The Prediction Model Of Unstable Rock Block According To Kalman Filter Method

Unstable rock block is a severe and global natural disaster in mountainous region.Its site is covert and the consequence for its breakage is very grave. Thus, to predict thetransmutation of unstable rock block is quite important. This thesis takes a research onthe displacement of unstable rock transmutation, adopts Kalman Filter method andstudies on the prediction model of unstable rock block, which is showed as follows:According to the basic model of Kalman Filter method, visual method andprojection method are used to confirm the optimal linear prediction estimate, predictionmeasured value and the orthogonal component, so the Kalman Filter in discrete systemis deduced in detail. Based on the basic of discrete system, while the sample intervaltends to zero, its limit is selected to deduce the generic formulas of Kalman Filter modelin linear continuous system.Because of model parameters and the prior knowledge of noises demanded by theapplication of Kalman Filter, somehow, under a lot of conditions, they are unbeknownor nearly known. If the imprecise mathematical model and noise design is used, KalmanFilter might result in the distortion of estate estimation, even the radiation of filter. Toovercome that shortage, it put an important study on the method and models establishedby self-adaptive Kalman Filter with no-divergent filter.It analyzed the basic geological features for unstable rock block of Moziling,Yun-yang County, in Three Gorges Reservoir Area: spatial form, topography andgeomorphology, geological structure, hydrogeology, boundary conditions, etc. Takingfive unstable rock blocks in Moziling to monitor, monitoring data were obtained anddatasheets and graphs for its horizontal displacement were ploted.It analyzed the external horizontal displacement of unstable rock blocks. Itcalculated and fitted the monitoring data by using Kalman Filter. Moreover, it analyzedand predicted the displacement and developing trend for those rock blocks.It adopted Matlab software to manage the data, denoised and predicted theaccumulative amount of displacement for Moziling unstable rock block. Furthermore, itvalidated the data result, which was processed by the self-adaptive Kalman Filter.According to the analysis, it tests the application of foregoing method, which providesscientific basis for the mid-long term prediction and the real-time disaster alert. It showspreferable practicability.