【作者】 刘学工；

【导师】 李占国；

【作者基本信息】 长春理工大学 ， 机械设计及理论， 2013， 博士

【摘要】 “重型车辆油液监测关键技术研究”是近年国内研究与发展的热点技术之一,是重型车辆状态监测的有效手段。本论文针对某类型特种重型车辆状态监测的实际需求,对重型车辆润滑油中的磨损颗粒进行识别与特征提取,预测传动系统的可靠性寿命,进行传动系统磨损状态故障诊断,提出试验方案,开发专家系统等,目的是给出车辆维护决策,以减少非预期停机损失,降低维修费用。论文首先介绍了油液监测的研究背景及意义,综述了国内外油液监测与故障诊断的研究现状及发展趋势,总结了油液监测手段和内容、故障诊断理论及油液监测的作用。通过对某类型特种重型车辆传动系统的结构、组成元素、磨损模式进行研究,揭示某重型车辆的磨损机理、磨粒特点以及主要故障模式,为进一步研究该车辆的油液监测技术和故障诊断奠定基础。论文基于油液光谱监测数据,建立一套重型车辆参数体系和运行状态监测模型,并运用因子分析理论,对高维的参数体系进行降维处理；根据投影寻踪理论,对降维后的数据进行处理,应用遗传算法进行优化,得到最佳投影方向向量,进而获得综合评价的各因子权重；利用最大熵原理,估计出最无偏概率密度函数和概率分布函数,进而建立车辆监测的诊断标准。论文在分析油液光谱监测特征的基础上,开发铁谱图像处理和识别模块,对光谱分析的不足进行补充。为此对油液监测的铁谱磨粒图像进行了特征提取,并运用改进的灰色关联度模型,对磨粒特征进行降维处理且不改变磨粒特征参数的原始性质；利用K-means聚类算法对油液特征进行二次压缩,在剔除冗余特征、选取有效特征方面提高了计算速度和分类识别率；结合特征聚类,对磨粒进行智能化识别,通过实例验证正确,进而为建立故障知识库提供理论模型。论文提出了基于路面影响因子的油液金属元素浓度预测方法,得到车辆磨损的可靠性寿命；结合重型车辆传动系统故障统计和拆检鉴定分析,得出该型号重型车辆磨损损坏阶段的基本里程,从而确定了该重型车辆的可靠性实车试验方案。最后,论文通过多种软件技术相结合建立了基于不同车型编号的磨损故障诊断系统,该系统实现了油液磨损状态规则的制定以及油样数据磨损状态的诊断,为车辆维护提供了决策。通过本论文的研究内容,将重型车辆油液监测的故障模式、诊断标准、寿命预测及试验方案结合起来,形成一套故障诊断系统,可以对任意重型车辆的油液监测数据进行处理,并提供维护决策,具有较大的工程应用价值。相信在不久的将来,通过发展和完善,可使之逐步走向产业化的道路。更多还原

【Abstract】 Lubricant condition monitoring, indicating the mechanical status of heavy vehicle, has been widely studied in recent years. In this study, the characteristics of wear granules in the lubricating oil were used to predict the reliability and wear state of transmission system of heavy vehicles. Based on theses works, test procedures of heavy vehicles were presented and an expert system for fault diagnosis was developed to decide when and how heavy vehicles should be maintained. The software for wear fault diagnosis was helpful to reduce maintenance costs and unexpected loss.Firstly, the history and roles of oil monitoring technology were reviewed. The methods of oil monitoring technology and their application in device monitoring and failure diagnosis were described. The principle of oil monitoring was summarized.Secondly, mechanical structure, metal element and wear mode of transmission system of certain heavy vehicle were analyzed, and its wear mechanism, particle feature and mostly failure mode were revealed. These are beneficial for the further study of failure diagnosis and oil monitoring of heavy vehicle.The parameter system and monitoring mode of running state were established by using spectrum data of the lubricating oil. The parameters of high dimention were depressed by using factorial analysis theory, Based on the projection pursuittheory, depressed dimensions was optimized by gene arithmetic, and best projection vector was founded, and factorial right weight of synthetical estimate was obtained. Moreover, the paper used most entropy principle to estimate probability density function without warp and probability distributing function of spectrum data, so that diagnosis criterion of oil monitoring of heavy vehicle was established.Thirdly, in order to reinforce monitoring characteristics of oil spectrum analysis, image processing and recognizing mode of ferrography technology were employed to analyze figure, veins and color characteristic of lubrication particles with grey correlation theory, and then K-mean clustering arithmetic was used to actualize the second compressing of the oil characteristics which improved the processing rate and classily recognition rate. Based on above, the paper recognized the oil wear particles and founded the diagnosis repository.Based on the results of oil spectrum and ferrography analysis and impact factor of road surface, forecast method of metal element concentration was presented. The fault statistics combined with that the transmission system was disconnected, to receive basic course of development of abrasion phases. Consequently, reliability test project of heavy vehicle was stipulated.Finally, wear fault diagnosis system on different vehicles was established by multi-software technology, which realized establishment of regulation and fault diagnosis of wear state of transmission system and offered maintenance decision-making of heavy vehicle. Anyway, research contents of the paper combined wear characteristics, criterion of synthetical estimate and life forecast with test project to come into being a suit of software system for oil monitoring of certain heavy vehicle, which will exhibit important value of engineering application. It is believed that in future, the wear fault diagnosis system will be competitive in state monitoring of certain heavy vehicle.更多还原