Geological Multi Parameter Monitoring And Early Warning Platform And Geological Landslide Prediction Based On GA-BP Neural Network

Monitoring and early warning of geological disasters has always been a topic of general concern in the society,which directly affects people's life and property safety.In China,the geomorphic characteristics are complex,geological disasters are widely distributed,hidden,sudden and destructive.Therefore,it is necessary to focus on the development of scientific research on the monitoring technology of geological disasters in order to reduce the harm and economic losses caused by geological disasters.This thesis focuses on the acquisition and display of the data from RTU,designs a geological disaster monitoring and early warning platform,and studies the problem of efficient landslide prediction.In order to solve the problem of high concurrency,according to the type of RTU,this platform designs two kinds of data receiving background.The first kind of background mainly uses Java NIO technology to realize non blocking single thread socket communication.After receiving the sensor data,it directly processes the data.It is mainly for RTU which is online for a long time but only sends data when the sensor collects data.The second kind of background mainly uses Active MQ technology,uses the traditional blocking multi-threaded socket communication,uploads the data to the message queue middleware Active MQ after receiving the sensor data,and processes the data in other programs,mainly for a large number of RTUs that are online and centralized to send data at the same time for a short time.The main functions of the background are to receive data,analyze data,store data,manage equipment,judge whether to send warning SMS and upload data to Sichuan Geological Disaster platform.In order to facilitate users to view data,analyze data and manage equipment,the local disaster monitoring and early warning platform is divided into five modules: user management module,map display module,monitoring data module,early warning data module and equipment information management module.The monitoring data module can view real-time data and historical data.The historical data includes two forms: table and chart.The chart style is determined by the type of sensor.For the crack meter,displacement meter and other sensors,the chart adds filtering function based on Kalman filter algorithm.While displaying historical warning information,the early warning data module can also configure whether to send early warning messages to the same level or low level of the same sensor next time.The device information management module is mainly used to add,delete,modify and query the information of the site,RTU and sensor.It can also configure early warning information in this module,such as early warning threshold,early warning SMS,etc.The early warning short message telephone is divided into the primary early warning telephone and the secondary early warning telephone.If the primary early warning telephone is empty,the secondary early warning telephone can not receive the early warning short message.When the early warning information is configured,the primary early warning telephone can receive the test short message,but the secondary early warning telephone does not have this function.Because the time series of slope displacement has the characteristics of nonlinear variation,the GA-BP neural network model(BP neural network model optimized by genetic algorithm)is established to predict the stability of landslide.The model uses genetic algorithm to optimize the initial weight,threshold and learning rate of BP neural network,and solves the problem that BP neural network is easy to fall into local minimum through the characteristics of "survival of the fittest" of genetic algorithm.Finally,taking a slag storage expansion project in Yunnan Province as an example,the GA-BP neural network is used for prediction,and the prediction effect of GA-BP neural network and BP neural network is compared and analyzed.The experimental results show that the BP neural network optimized by genetic algorithm has higher prediction accuracy and greater application potential.