Abstract: Currently, the monitoring of landslides is often achieved by arranging multiple monitoring points on the landslide mass. To improve the reliability of landslide prediction making use of all the monitoring data, the technology of no-reset federated Kalman filter is proposed to synthetically extract the monitoring information, and meanwhile the consistent description of landslide is given. This method has such advantages as high fault tolerance and smaller computation. A simulation example shows that it is feasible and effective to apply this method to the fusion of landslide monitoring data of multiple sensors.

Key words: federated Kalman filter, information extraction, landslide, monitoring data