Application Of Complex Event Processing To Landslide Monitoring Sensor Data Integration Research

Of the complex geographical terrain and climate conditions, resulting in heavy losses caused by geological disasters, landslides of geological disasters are in the proportion of more than half, in which the rainfall-induced landslides account for 90% of the total landslides. The average number of deaths due to landslides disasters is up to thousands of people, and property loss is up to tens of billions. Therefore, in a region affected by natural disasters usually, environmental monitoring as a critical work that can ensure public safety, establish continuous information service mechanism, and provide input to the spatial decision support systems.For the above purpose, to change the type of rainfall landslide research at home and abroad focusing on the two traditional angles of statistical forecasting and analysis of the mechanism, this paper puts forward the innovative points of view of event-driven. Through a lot of literature review, teases out several rain-induced landslides earth science events, and selects three typical model as the basis for the event feature extraction, use of statistical methods, GIS technology, theoretical physics and image processing means of clearly different models and determined conditions applicable threshold range, to establish an association rule learning events and the observed data. CEP using complex event processing as a key technology, from the mass dynamic sensor data flow, abstracts demand information for detecting an abnormality of landslide behavior or behavior patterns. Using JAVA language development, the open source event engine project -Esper technical expansion, and EQL (Event Query Language) language as an input of event metadata, through the rules engine and data model, find the compound event, and trigger event handouts and action modules. Campus by building the campus rain-induced landslides simulation field to simulate experiments, verifies the stability and applicability of event detection algorithms.Final results shows that the proposed rule of event-driven angle to sensor data and landslide events associated with effective verification, and the predicted rainfall type landslides in several different stages are basically in line with the actual situation, so research methods can be event-driven angle-type landslide monitoring rainfall prototyping.