An Audio-based Intelligent Fault Classification System for Belt Conveyor Rollers

This paper researches how to realize the fault classification of the roller components in the belt conveyor of the coal preparation plant, which is selected as the actual industrial scene for the application of fault. An actual fault state classification system based on audio data and designed for rollers of belt conveyor is built. The system is consists of hardware parts and software parts. The hardware part applies with wired communication to achieve on-line dynamically measuring in an awful spot environment. The software applied with machine learning algorithms is designed for fault classification to achieve the goal of more than 90% classification accuracy. The application demonstrates that the system has the characteristic of simple structure, higher interference-free capability, moving steadily, high precision, good expansibility, productivity-improving, and advanced performance-to-price ratio.