School of Information and Control Engineering, China University of Mining and Technology

输送带和驱动装置是带式输送机的主要组成部分且为故障高发部位，以输送带故障和驱动装置故障为切入点，分析了输送带跑偏、打滑、损伤、堆料撒料等故障及驱动装置滚筒、托辊、减速器等故障的机理，重点阐述了知识驱动和数据驱动的带式输送机故障诊断方法研究进展。知识驱动法以知识处理技术为基础，实现符号处理和数值处理的统一、推理过程和算法过程的统一，主要包括专家系统、故障树分析法。数据驱动法采用机器学习和数据挖掘等技术对历史数据进行分析处理，建立诊断模型，达到故障诊断目的，主要包括支持向量机（SVM）、比差法、基于声音和视觉的诊断方法。分析了带式输送机故障诊断方法目前存在的挑战和未来发展趋势：① 结合历史故障数据和实时数据推断设备健康状况，预测早期微小故障，提醒工作人员进行预测性维护。② 揭示带式输送机耦合故障的关联关系，利用人工智能等新兴技术研究耦合故障联合诊断方法。③ 利用多模态机器学习技术研究带式输送机多模态信息融合利用机制，开发带式输送机多模态信息融合故障诊断方法。④ 将故障知识图谱和带式输送机领域知识相结合，实现带式输送机设备故障追踪、故障超前预警，通过知识查询、知识推理和辅助决策功能，提高故障处理、精准挖掘设备潜在故障风险的能力。

The conveyor belt and driving device are the main components of the belt conveyor and are the high-risk areas for faults. Taking conveyor belt faults and driving device faults as the starting point, this paper analyzes the mechanisms of conveyor belt deviation, slipping, damage, stacking and scattering, as well as the faults of the driving device roller, idler and reducer. It focuses on the research progress of knowledge-driven and data-driven fault diagnosis methods for belt conveyors. Based on the knowledge processing technology, the knowledge-driven method realizes the unification of symbol processing and numerical processing, the unification of reasoning process and algorithm process. It mainly includes expert system and fault tree analysis. The data-driven method uses machine learning and data mining techniques to analyze and process historical data. It establishes diagnostic models, and achieves fault diagnosis purposes. It mainly includes support vector machines (SVM), comparison method, and diagnosis methods based on sound and vision. This paper analyzes the current challenges and future development trend of belt conveyor fault diagnosis methods. ① The historical fault data and real-time data should be combined to infer equipment health. The early minor faults should be predicted so as to remind the staff to carry out predictive maintenance. ② The correlation between coupling faults of belt conveyors should be revealed. The emerging technologies such as artificial intelligence should be used to study joint diagnosis methods for coupling faults. ③ The multimodal machine learning technology should be utilized to study the mechanism of multimodal information fusion and utilization of belt conveyors. Fault diagnosis methods for multimodal information fusion of belt conveyors needs to develop. ④ The fault knowledge graph and the belt conveyor domain knowledge should be combined to realize the belt conveyor equipment fault tracking and fault early warning. Through the knowledge query, knowledge reasoning and auxiliary decision-making functions, the capability of fault handling and precise mining of potential fault risks of equipment can be improved.

国家自然科学基金面上项目 (61873272)。

0 引言
1 带式输送机故障分类及故障机理
1.1 输送带故障
1.2 驱动装置故障
2 带式输送机故障诊断方法
2.1 输送带故障诊断方法
2.1.1 知识驱动的输送带故障诊断方法
2.1.2 数据驱动的输送带故障诊断方法
2.2 驱动装置故障诊断方法
2.2.1 知识驱动的驱动装置故障诊断方法
2.2.2 数据驱动的驱动装置故障诊断方法
3 现存问题及发展趋势
3.1 带式输送机故障诊断方法现存问题
3.2 带式输送机故障诊断方法发展趋势
4 结语

杨春雨，曹博仕，张鑫，等. 带式输送机系统故障诊断方法综述[J]. 工矿自动化，2023，49(6)：149-158.

YANG Chunyu, CAO Boshi, ZHANG Xin, et al. Summary of fault diagnosis methods for belt conveyor systems[J]. Journal of Mine Automation，2023，49(6)：149-158.