GSA-SVM: Gravitational Search Algorithm with Support Vector Machine for Fault Diagnosis of Scraper Conveyors
Year: 2026
Type: conference-paper
Abstract: Scraper conveyors serve as important transportation equipment in underground coal mining operations, yet their complex mechanical structures and harsh working conditions frequently led to critical component failures. However, the existing method has limited capability to handle complex and highly nonlinear relationships among multiple conveyor sensor signals, due to its performance depends strongly on the proper selection of kernel and penalty parameters. This limitation affects the accurate identification of different conveyor fault conditions, when the operating environment and sensor patterns become complex. To solve these challenges, this research proposed Gravitational Search Algorithm with Support Vector Machine (GSA-SVM) for fault diagnosis scraper conveyor. The proposed method collects the data including vibration, current, temperature, speed and load signals from key machinery components. Then, the raw data passed through pre-processing stage to filter background noise and segment continuous data windows. Similarly, time domain statistical metrics are derived through feature extraction, while the GSA dynamically initializes candidate solutions to automatically optimize the penalty parameter and kernel parameter of the SVM classifier. The optimized model maps non-linear feature vectors into high dimensional space to categorize conveyor health into normal operation. The experimental validation demonstrates that the proposed GSA-SVM method achieves an accuracy of 99.40%, precision of 99.20%, recall of 98.80% and F1-score of 99.00%.
Author: Shanshan Li
Institution: Lanzhou Vocational Technical College
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7225129649