Research On Landslide Monitoring And Early Warning Based On Multi-source Information Fusio

In recent years,influenced by frequent human activities in many regions,coupled with global climate change and frequent extreme disasters,landslides have become the most serious natural disasters affecting and endangering our people,and currently,seriously restricting our social and economic development.Therefore,how to provide timely and accurate early warning of landslides has become an urgent problem.Among the many methods proposed for this problem,multi-source information fusion processing and decision early warning analysis is one of the main methods for landslide early warning at present.The multi-source information fusion technology can effectively combine the time series data of landslide monitoring area with spatial information to obtain reliable and accurate prediction of landslide movement status.Therefore,this paper takes multiple regions in Guizhou Province as the research area object.By monitoring landslide multi-sensor data and combining with spatial information analysis,a landslide monitoring and early warning system is built to realize data-level fusion in the underlying hardware and decision-level fusion in the remote platform,while the monitoring data rely on the Io T cloud server to realize remote storage,remote monitoring and remote warning of data.The main work of this paper is as follows.(1)Establishing multi-source information fusion algorithm model.First,12 landslide sensitive factors combining spatial information and time series information are used to analyze the danger state of the mountain.Then,a TS-based fuzzy neural network model is constructed to achieve decision-level multi-source information fusion results.Finally,in order to avoid the risk of the algorithm falling into local optimum,the global optimization property of ant colony algorithm is used to optimize the weights of TS network,which makes the network performance better.(2)Applying the theoretical study of this model,a landslide monitoring and warning system with multi-source information fusion is designed.The monitoring system is mainly divided into three parts: acquisition,transmission and application.The acquisition is divided into geological information and environmental information of the monitoring area;among them,geological information is analyzed by Arc GIS platform to find out 5 spatial landslide condition factors of the monitoring area;environmental information is collected by multi-sensor hardware monitoring system to collect 7 time series of landslide condition factors.The transmission adopts realtime monitoring integrated data encapsulation format and transmission scheme based on MQTT protocol,and builds EMQ server cluster to ensure the real-time,reliability and stability of monitoring data with high concurrent data processing.The application adopts a hybrid architecture landslide monitoring and warning system with unified interface,multi-platform architecture system(Node-RED,Lab VIEW,MATLAB),based on MQTT interface,to build a unified integrated data display platform for different functional requirements and take advantage of the architecture.(3)In order to verify the effectiveness of this model in landslide monitoring,early warning results,multiple areas in Dafang,Bijie City,Guizhou Province,are selected as landslide areas for study.On the basis of extracting spatial information,favorable monitoring points were selected and hardware monitoring platform was deployed.The operation results show that the multi-source information can obtain richer and more effective monitoring area information,eliminate the limitation of single landslide condition factor on landslide grade evaluation,and improve the reliability and accuracy of overall landslide grade evaluation.