Researches On Application Of Markov Chain Theory To Landslide Early Warning

In this paper,according to the random properties of the displacement parameters obtained in the process of landslide evolution,comprehensively using three stages theory of landslide evolution,system clustering method,mean and standard deviation classification method,Markov chain prediction theory,etc,the landslide early warning model based on the theory of Markov chain is constructed,the landslide early warning criterion based on the theory of Markov chain is put forward,and the evaluation criteria of early warning model is given.The landslide early warning model based on Markov chain prediction theory is applied in the 4.17 landslide engineering example of on top slope of Pingzhuang West open pit.Through successive inspection and comparative analysis on the early warning effect of the 28 monitoring points near landslide area,the optimal model parameters of dynamic weighted Markov prediction model based on system clustering and dynamic weighted Markov prediction model based on mean and standard deviation are obtained separately.Finally,by comparing the early warning effects of the two models,the applicability of the model is obtained,which provides a guide for the selection of early warning methods.Study shows that both dynamic weighted Markov prediction model based on system clustering and that based on mean and standard deviation can give landslide early warning to single monitoring point,and they can also reveal the process of the initiation and evolution of the landslide from the region.The two models have their respective applicability,the former is applied to the condition with inadequate displacement data and has relatively higher false alarm rate and the intensive early warning time is relatively;the latter is applied to the condition with adequate displacement data and has almost zero false alarm rate and the intensive early warning time is more accurate.