If you have any problems related to the accessibility of any content (or if you want to request that a specific publication be accessible), please contact us at firstname.lastname@example.org.
Detection and Monitoring of Tailings Dam Surface Erosion Using UAV and Machine Learning
AltmetricsView Usage Statistics
The recent failure that occurred in Burmadinho had yet added the list of tailings dam failure within the mining industry. The industry is currently in search of a better way to avoid them from happening again in the future. Most of the tailings dams guides suggest the mine to develop a proper surveillance program to detect possible causes of failure. Machine learning applications currently have been adapted throughout multiple engineering fields. However, the mining industry just recently started to possibly implement it into the daily operation. Parameters suggested to be monitored by the guides mostly require visual inspection, and UAV technologies are convenient and most effective for this task. Though UAVs are not foreign to most mine operations, the combination of rill detection with machine learning and UAV data is promising to aid the extensive in-person visual inspection of TSF walls. In addition, the automation will help solve the constant changes of managerial within the operation. This dissertation serves as proof of concept in utilizing semantic segmentation with UNet architecture and weighted cross entropy loss function for the detection of rills on tailings dams. The final model showed a promising result with precision, recall, and F1-score of 83.3%, 72.0%, 77.2% respectively.