Recognition and classification of microseismic event waveforms based on histogram of oriented gradients and shallow machine learning approach

• Published in: Journal of applied geophysics 2024-11, Vol.230, p.105551, Article 105551
• Main Authors: Shu, Hongmei, Dawod, Ahmad Yahya, Dong, Longjun
• Format: Article
• Language: English
• Subjects: Histogram of oriented gradients; Image classification; Machine learning; Microseismic signals; Support vector machine
• ISSN: 0926-9851
• DOI: 10.1016/j.jappgeo.2024.105551

Accurate identification of microseismic events is vital for understanding underground rock deformation, rupture behavior, and mechanical properties. This study proposes a method that combines the Histogram of Orientation Gradient (HOG) and shallow machine learning techniques for microseismic waveform recognition. HOG features are extracted from event waveform images, and five classifiers including Linear classifier (LC), Fisher Discriminant (FD), Decision Tree (DT), K-Nearest Neighbors (KNN), and Support Vector Machine (SVM) are compared. Experimental results show good accuracy and efficiency, with the SVM classifier and FD classifier achieving the best performance at 97.1 % and 96.9 % accuracy, respectively. Compared to previous studies, this method offers simplicity, ease of use, and low computational resource requirements, making it valuable for real-time monitoring and disaster prediction applications. It provides a foundation for evaluating mine geological structure stability. •Proposed a multi-channel microseismic event recognition and classification method using HOG and shallow machine learning.•Constructed and compared five different classifier models: SVM, Linear, Decision Tree, KNN, and Fisher Discriminant.•Compared to traditional methods, this study's automatic feature extraction requires no manual intervention, offering simplicity, ease of use, and low computational costs.