多源异构数据驱动的综采设备群健康评价与多目标预知维护决策

As a key part of coal mining, the fully-mechanized coal mining equipment group's healthy operation is crucial to safe, efficient, green, and intelligent coal mining. The project aims at the health assessment and predictive maintenance of the fully-mechanized coal mining equipment group. Based on the complexity, uncertainties, and low density of the equipment group data, the project intends to solve the problems of multi-source heterogeneous data fusion processing with structured and unstructured coexistence. Its data describes key scientific issues such as modeling and storage query mechanism construction, data-driven health assessment and residual life prediction, and system multi-objective predictive maintenance decision model solving. The main researches include the construction of equipment health assessment model based on deep belief learning network theory, the effective extraction of evolutionary mechanism information of equipment health state degradation process, the research on the method for predicting the residual life of equipment group based on the theory of complex system and multivariate statistical methods, etc. The predictive maintenance decision model that takes into account the impact of the residual life of the equipment and the opportunity maintenance strategy reveals the relationship between the changes in the residual life of the equipment group and the decision results. The research results of the project provide new ideas and new methods based on big data analysis for the assessment of health status, life prediction and maintenance decision of the coal mining equipment group, and it has important scientific significance and application value for the development of coal mine production safety and intelligent mining technology.

综采设备群作为煤矿开采的核心关键部分，其健康运行对煤炭安全、高效、绿色、智能开采至关重要。项目以综采设备群的健康评价与预知维护为目标，基于设备群数据复杂性、不确定性和密度低等特点，拟通过结构化与非结构化并存的多源异构数据融合处理，解决其数据描述建模及存储查询机制构建、数据驱动健康状态评估与寿命预测和系统多目标预知维护决策模型求解等关键科学问题。主要研究包括基于深度置信学习网络理论的设备健康状态评估模型构建，实现设备健康状态退化过程演化机理信息的有效萃取；基于复杂系统与多元统计方法等理论的装备群剩余寿命预测方法研究；研究建立计及设备剩余寿命和机会维修策略影响的预知维护决策模型，揭示设备群剩余寿命的变化与决策结果的关联关系。项目研究成果对煤矿综采设备群健康状态评估、寿命预测和维护决策提供了基于大数据分析的新思路和新方法，对煤矿安全生产和智能化开采技术发展具有重要的科学意义和应用价值。

煤矿智能化是煤炭工业高质量发展的核心技术支撑，是我国能源技术革命的新方向。综采设备稳定可靠运行是煤矿安全、高效开采的必要保证。如何准确、高效、智能地监测、预测及维护综采设备群成为煤矿智能化进程中亟待解决的问题。本项目以综采设备群运维数据为基础，构建了基于Hadoop的综采设备群运维数据分布式计算和存储框架，实现了多源异构数据的高效存储；研究了时变工况下深度残差收缩网络改进方法、不完备特征下时空融合自编码网络方法等综采设备健康状态评估方法，提升了复杂工况下综采设备群健康状态评估精度；提出了特征冗余下基于核主成分分析的长短期记忆神经网络等剩余寿命预测方法，提升了综采设备群剩余寿命预测的准确性；提出了面向生产计划与维护活动的综采设备群多目标联合决策优化方法，缓解了设备群生产计划与维护活动间的冲突问题。项目组严格按照研究计划开展研究，取得了预期的研究成果，从研究深度和研究广度上均达到了项目的结题要求。项目组共发表学术论文42篇，其中SCI/EI索引16篇；申请国家发明专利4项（已授权2项）、软件著作权2项。与西安煤矿机械有限责任公司、铜川玉华煤矿等煤炭企业深入合作研发了综采设备群健康监测与运维平台3个，现已成功应用并推广；2020年获批陕西省科学技术进步二等奖1项。综上所述，本项目研究成果为复杂工况下的煤炭企业设备科学管理提供了理论基础，对促进煤矿安全生产和智能化开采技术发展具有重要的科学意义和应用价值。