University of Twente Student Theses


Particle detection and tracking in avalanches

Broeke, D.A. van den (2023) Particle detection and tracking in avalanches.

[img] PDF
Abstract:To prevent avalanches one requires their reliable prediction. For this purpose one has to study the avalanche dynamics and conditions under which such dynamics occurs. In this thesis the avalanche of granular materials is studied from an experimental and data-driven perspective. The data are collected by taking images of the rotating drum filled by a granular material with the help of the high speed camera. Initially, the individual particles in images are detected by use of circular Hough transform and the existing expert knowledge. Furthermore, the particles are tracked by use of the nearest neighbor method in combination with the unscented Kalman filter. This approach has shown to have good performance as one may easily account for missing particle detections and uncertainties about particles exact position. Thus, the initial particle position is modelled as uncertain, and its evolution is further predicted using the coordinated turn model. With the help of the proposed tracking method the particle position is further updated by assimilating the measurement data with the prior knowledge. The method is evaluated on unseen data set, and is able to predict the global avalanche dynamics such as angle of repose, and movement of particles that is not too stochastic. On the other hand, prediction of the individual movement of particles in the middle of the avalanche is shown to be more challenging.
Item Type:Essay (Master)
Faculty:ET: Engineering Technology
Subject:52 mechanical engineering
Programme:Mechanical Engineering MSc (60439)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page