千米深井大型提升机主轴装置早期故障诊断及恶化状态评估

Main shaft device is one of the most key parts of mine hoist. Its serious failure could lead to catastrophic accident and affect the safety of mine production. With the increase of well depth, lifting speed and load, the large mine hoist will appear more complex nonlinear coupling and stronger noise interference, which are new challenges to the fault diagnosis of main shaft device. Research on early fault diagnosis and performance degradation assessment of the large hoist main shaft device of kilometer deep mine is carried out by using heterogeneous sensor data sets to extract the features of weak fault, and its main research contents are: with the proposed multi-scale noise auxiliary and evaluation based hybrid EMD method, the weak fault information hidden in the nonlinear and complex sensor signals is isolated by selecting these signals components sensitive to the faults of main shaft device; considering time-varying working condition and sensors with various sampling frequencies, a kind of stochastic window based statistical feature extraction method is proposed with the constructed synchronous sampling strategy to solve the problem of how to make full use of the various sensor data sets; with researching the correlation between features and working conditions of main shaft device, these features extracted from heterogeneous sensor data sets are used to designed orthogonal projection model and parameter optimality RVM for incipient fault diagnosis and degradation condition assessment of main shaft device. Research results of this project could provide technical support for the health monitoring and safe running of large mine hoist working in kilometer deep mine.

主轴装置是提升机的关键传动部件，其性能好坏直接影响矿井的安全生产。随着井深、提升速度和载荷的增加，千米深井大型提升机非线性耦合和强噪声干扰特点凸现，给主轴装置故障诊断带来了新的挑战。本项目从综合利用异构传感数据出发，以快速提取微弱的故障特征为手段，开展千米深井大型提升机主轴装置早期故障诊断及恶化状态评估方法研究，主要包括：针对主轴装置故障信息的非线性和微弱性，提出基于多尺度噪声辅助和评估的混合经验模态分解方法，在敏感分量定位基础上自适应地分离出主轴装置微弱的故障信息；提出同步采样策略，构建基于随机窗子空间序列的统计平均特征提取方法，解决变工况下主轴装置多源异构传感信号的统一表征问题；提出主轴装置性能退化关联特征优选方法，设计正交投影模型和参数优化相关向量机，建立基于多源异构数据驱动的主轴装置早期故障诊断及恶化状态评估策略。研究成果可为千米深井大型提升机的健康监测和安全运行提供技术支撑。

煤炭是我国的主体能源，每年约90%的开采量来源于地下。随着开采强度的增加，浅层煤炭日益枯竭，煤炭开采向深部化发展。据统计，我国深埋千米地层下的煤炭资源约占已探明总储量的50%，千米深井将成为井工开采的必然趋势。提升机是矿山井工开采的关键设备，承担地面和井下之间人员、物料、设备和矿石等的运输重任，素有矿井“咽喉”之称。作为矿井提升机的关键设备，主轴装置在提升系统中起着传递动力的重要作用，其性能好坏直接影响矿井的生产效率。.本项目以提升机主轴装置为研究对象，从综合利用多源传感器数据出发，以提取微弱故障特征为手段，进行千米深井大型提升机主轴装置早期故障诊断及恶化状态评估方法研究。项目主要开展了三个方面的研究内容：强噪声干扰下主轴装置微弱故障信息自适应分离、早期故障诊断和恶化状态评估。提出了一种混合经验模态分解方法和迭代奇异值分解重构方法，对主轴装置的非线性微弱故障信号进行自适应分离，实现了分离出被强噪声淹没的故障信号，为主轴装置微弱故障特征提取建立基础。在微弱信号分离基础上，通过包络分析和正交投影分析，实现了千米深井大型提升机主轴装置的早期故障诊断。提出了基于随机窗子空间序列的统计排列熵指标、多传感器敏感特征优选和融合策略、主轴装置恶化状态识别方法，实现了主轴装置故障状态识别。在国家自然科学基金的支持下，本项目发表SCI收录论文8篇；授权美国发明专利1件、俄罗斯发明专利1件、中国发明专利3件。本项目取得的成果对提高矿山生产的安全性、减少恶性事故的发生、避免人员伤亡或设备损坏、保障我国能源供应等具有重要意义。