Knowledge-Based alarm System with Identified Deformation Predictor (KASIP)

Because of increasing settlement activities of people in mountanious regions and the simultanous appearance of extreme climatic conditions the investigation and alerting of landslides becomes more and more important. Within the last few years a significant rising of disastrous slides could be registered which generated a broad pulic interest and the request for security measures. The investigation and installation of alarm systems aims for a raise of security and a restriction of human, economical and environmental damage. Our new vision is the combination of monitoring data (e.g. GPS or tacheometer measurements) with a numerical model which represents the structure of the slope. The model is planned to overtake the calculation of precise simulations and to support the prediction of critical states of the slope as reaction to environmental influences (like mass excavation). It will be one central component of a new type of data- and knowledge-based alarm system. Central elements of the data-based part are the monitoring system and the numerical slope model which is adapted to reality. For the model calibration step we are planning to use adaptive KALMAN-filtering tech-niques and replace the statistically non assured try and error methods. The knowledge-based part acts as a superordinated alarm manager which combines and evaluates the calibration, simulation and/or prediction results of the numerical model with additional hybrid expert knowledge. This will be measuring results from the monitoring system, additional local deformation models (e.g. polynomials or spectral analysis) and heuristic knowledge from landslide experts. The goal is to establish a widely automated decision process whether to keep the current alarm level of the slope or to change to another one.

The project "KASIP" (Knowledge-based Alarm System with Identified Deformation Predictor) deals with the investigation of a new type of alarm system for landslides. The central research idea is the combination of real-time monitoring data (e.g. tacheometer and radar measurements) from the landslide with a numerical model of the slide (slope model). The major goal is to get an improved slope model which is adapted to the real slope deformations and is able to perform realistic forecasts for the future behaviour of the slope. In this context a new laminar monitoring method for landslides is developed further: the ground based radar monitoring. This method enables the observation of a landslide from distances up to 4 km with a measuring rate of ca. 6 min and mm-precision. Dynamic effects like accelerations and consolidations of the slope as a result of changing meteorological conditions and even sudden rockfall events can be observed. Another big topic in the project is the optimal estimation of a priori unknown material parameters in the slope model with adaptive Kalman-filtering techniques. It is shown that it is possible to estimate precisely parameters like friction and cohesion which are responsible for the slope stability.