Research On Key Technology Of Landslide Monitoring And Auxiliary Decision

In China,landslide disaster is one of the major geological disasters with high frequency,destructive and heavy damage.To protect the people 's life and property safety,so the current research we focus on adopting timely and effective means for monitoring landslide.The existing landslide monitoring system uses the wireless sensor network technology to realize real-time monitoring of landslide,real-time data transmission and data visualization,and it plays an important role in the monitoring area where the human cannot complete the work.But the system lacks the regional detection of landslide,the function of holistic analysis and the efficient localized regional landslide prediction model.Therefore,the prediction is not timely,the prediction accuracy is not high.Therefore,the prediction is not timely,the prediction accuracy is not high.In this paper,the application of machine learning technology supports the system to analyze the landslide law,identification,statistics and landslide grade division,which can assist the experts in landslide warning analysis and improve the accuracy of landslide prediction.The main work done is as follows:1.we adopt the methods including demising,smoothing and normalization of data to handle the data of landslide monitoring.Then,we clear the data in the landslide monitoring data set with noise,inconsistent,incomplete,redundant and missing data,and further demised,smoothed and normalized to improve the quality of the data set.Finally,we get ideal training data set.2.A landslide region fragmentation detection method was designed by using machine learning,R-tree,data correlation and so on.The purpose is to realize the functions of landslide regional testing and holistic analysis,to count the occurrence probability of landslide.And according to the landslide level,each slice area is divided,and the landslide probability values of each slice area are automatically marked and displayed.So as to achieve landslide regional testing,holistic analysis.3.Bringing forward a model for local landslide prediction.using RBM to adjust slightly the model parameters.This model draws on the idea of multi-hidden layer of deep-learning convolution neural network.In the process of training the model to join the multi-hidden layers of the number of selection settings,continue to low-level data features passed to the last layer,until the last layer,and get a data feature set.Then the data featureset is trained again and again,and build the local area landslide prediction model.The results show that the system has certain auxiliary decision function in three actual landslide scenes in Nanshan Scenic Area,Jiangxi Zhou and Parson's homes in Zhenjiang City,Jiangsu Province.