A Comparative Study with Traditional and Transfer Learning-enhanced Machine Learning Algorithms for Geotechnical Characterisation of Coal Spoil

Abstract

The characterisation of materials is a prerequisite for evaluating and predicting the stability of mining waste dumps. Over the past three decades, the BHP Mitsubishi Alliance Coal framework has been a cornerstone in Australian coal mines for characterising waste dump materials. However, its reliance on subjective human observations has introduced potential inaccuracies and subjectivity into the process. In response to these limitations, this study proposes an innovative approach to classify coal spoil attributes by remotely acquiring images through phones/tablets. Automated image-based classification relies on feature extraction and a substantial amount of data. Nevertheless, the inherent complexity of geological factors contributing to the formation of both rare and dominant materials leads to imbalanced data. Recognising the need for classification mechanisms to overcome these challenges in spoil classification, the study explores and compares the use of convolutional neural networks, hybrid deep learning, and traditional techniques. Among the twelve models evaluated in this study, the ResNet18-k nearest neighbour model emerges as a powerful tool in geotechnical characterisation. However, it is essential to address issues of interpretability and adaptability to diverse datasets. As this study evolves, the field of geotechnical characterisation of spoil can anticipate the development of more robust methods in the future.