【作者】 程哲；

【导师】 胡茑庆；

【作者基本信息】 国防科学技术大学 ， 机械工程， 2011， 博士

【摘要】 直升机被广泛应用于抗灾救援、科学研究和反恐维稳等诸多领域,在国民经济发展和国家安全中发挥着重要作用。作为直升机的重要组成部分,传动系统的运行环境恶劣多变,导致其关键部件容易产生故障,又因为传动系统的机械部件一般为无冗余设计,一旦出现故障往往引发严重事故。故障预测对于监测、预报直升机传动系统关键部件的运行状态、保障其安全运行具有重要意义。在直升机传动系统关键部件的故障预测中,常常存在故障机理不明确、故障演化数据难以获取、早期损伤检测困难、预测特征难以量化选择等问题。为此,本文以直升机传动系统中的核心部件——行星轮系为研究对象,开展了损伤建模与故障预测理论及方法的研究工作。开展的主要研究包括:1.系统地研究了行星轮系典型损伤的建模方法,以深入分析典型损伤对行星轮系动态特性的影响。(1)研究了基于集中参数动力学理论的典型损伤建模方法。通过分析行星轮系典型损伤的机理及其对时变啮合刚度的影响,建立了行星轮系太阳轮齿根疲劳裂纹、点蚀、胶合和缺齿等常见损伤模式的集中参数动力学模型。(2)提出了基于多体动力学模型的行星轮系典型损伤建模方法。通过分析行星轮系典型损伤的接触函数,建立了行星轮系太阳轮点蚀和缺齿等典型损伤模式的多体动力学模型。对比研究表明,上述动力学模型有效刻画了太阳轮常见典型损伤对行星轮系动态特性的影响。2.基于损伤模型和统计分析,深入研究了行星轮系典型损伤的特征提取方法;在此基础上,研究了基于灰色关联分析的早期损伤检测与模式识别方法。(1)基于行星轮系典型损伤模型的仿真数据分析,提出了基于多种变换域信息的特征生成方法,提出了基于统计算法的特征敏感度和稳定度的评估方法和特征权重方法。(2)基于所提取的特征向量,将灰色关联分析应用于行星轮系的早期损伤检测与识别。实验数据验证表明,基于损伤模型的特征生成、选择与权重方法可有效提取行星轮系太阳轮典型损伤的特征,特征的敏感度和稳定度优于常用特征指标;基于灰色关联分析的检测和模式识别方法可有效检测和识别行星轮系早期损伤。3.针对行星轮系的退化状态识别和故障预测,提出了基于典型损伤演化信息的预测特征提取和选择方法。在此基础上,将灰色关联分析和特征权值向量相结合,提出了基于灰色概率关联分析的行星轮系退化状态识别方法。研究表明,上述方法可建立预测特征提取和选择的量化标准,所提取的预测特征适于行星轮系典型损伤的退化状态识别。实验验证结果证实了基于灰色概率关联分析的方法对行星轮系太阳轮齿根裂纹退化状态识别的有效性。4.针对行星轮系太阳轮齿根疲劳裂纹的故障预测问题,研究了基于损伤演化机理的剩余使用寿命预测方法,提出了基于改进灰色模型的剩余使用寿命预测方法。(1)研究了基于Paris公式的行星轮系太阳轮齿根疲劳裂纹的演化机理模型,研究了基于机理模型的剩余使用寿命预测方法和预测结果准确度的评价方法。(2)将损伤演化信息与灰色模型相结合,提出了灰色模型的改进方法,并将改进灰色模型应用于行星轮系太阳轮齿根疲劳裂纹的剩余使用寿命预测。研究表明,通过对损伤演化数据和运行剖面数据的修正,有效提高了基于机理模型预测方法的准确度。同时,与传统灰色模型相比,基于改进灰色模型的预测方法具有更高的准确度。5.分析了面向直升机传动系统的故障预测与健康管理系统体系结构,开发了相关的硬件系统和软件系统,并在实验室环境下对该系统进行了验证。验证结果表明,直升机传动系统故障预测与健康管理原型系统具有数据采集与状态实时监控、信号回放与特征选择、早期损伤检测与模式识别、故障趋势预测与剩余使用寿命预测和数据、信息、知识管理等基本功能。更多还原

【Abstract】 As a type of low-flying craft, helicopter is widely used and plays a significant role in many areas of national economy. Due to its non-redundancy structure and severse operation condition, the faults in the parts of transmission system of helicopter are common, which always cause catastrophic accident.The development of Prognostics technology promotes the security of helicopter transmission system. There exist some problems such as the fault mechanics is undefinited, run-to-failure data is hard to obtain, incipient damage is difficult to detect, prognosis feature selection is not quantitative etc. After that, the key component named as planetary gear set in helicopter transmission system is selected, and the corresponding theories and methods of damage modeling and failure prognosis are researched. The detailed contents and innovative work can be summarized as follows.1. The modeling methods of normal damages in planetary gear set are researched systematically, in order to analyze the impact of normal damages to the dynamical characteristics of planetary gear set.(1) The lumped parameter dynamical models of the normal damages in planetary gear set are investigated. By the analysis of damage mechanics and time-varying meshing stiffness, the common damages such as root fatigue crack, pitting, scuffing and tooth breakage in the sun gear of planetary gear set, are developed.(2) The multi-body dynamical modeling methods of common damages in planetary gear set are presented. Based on contact function analysis, the multi-body dynamical models of the damages such as pitting and tooth breakage are built.The research shows that the models above represent dynamically the response of planetary gear set with common damages, which are the basis of incipient damage detection, degradation state recognition and failure prognosis.2. Feature extraction method based on simulation data and statistical algorithm is presented. And then, a grey relational analysis (GRA) based method is proposed for incipient damage detection and pattern recognition.(1) Multiple domain information based feature generation method is researched based on simulation data of the models above. And the feature evaluation and weighting approach is processed utilizing the sensitivity analysis and stability analysis. .(2) Based on the features extracted above, GRA is utilized into incipient damage detection and pattern recognition for planetary gear set.The above research shows the features extracted above are effective for sun gear damages in planetary gear set, which have the merits of sensitivity and stability compared to the traditional indicators. The effectiveness of the detection and recognition approach based on GRA is validated by damage seeded test data. 3. A prognosis feature extraction and selection method based on the simulation information of damage propagation is derived, in order for degradation state recognition and failure prognosis. On the basis of the prognosis feature extracted above, GRA and feature weighting vector are united to recognize the degradation states of planetary gear set.The research shows that a quantitative criterion is set to extract and select prognosis feature, which is suitable for degradation state recognition. The test results validate that the approach based on grey probability relational analysis is effective in degradation state recognition for sun gear root crack of planetary gear set.4. To resolve the problems of failure prognosis for sun gear tooth crack in planetary gear set, the residual useful life (RUL) prediction approach based on damage evolution mechanics is researched, and then the RUL prediction approach based on a modified grey model is presented firstly.(1) A damage evolution mechanics model of sun gear root fatigue crack of planetary gear set based on Paris equation is researched. After that, the RUL prediction approach based on mechanics model is researched. And then the accuracy of prognosis result is evaluated.(2) The grey model is modified and utilized to predict the RUL of planetary gear set with sun gear root fatigue crack for the first time.The research shows that the damage evolution mechanics model need modification by run-to-failure data and operation profile data. And the RUL prediction method based on mechanics model holds high accuracy. Compared to the traditional grey model, RUL prediction method based on modified grey model has merit on the accuracy of prediction results.5. Aiming at the helicopter transmission, the architecture of PHM system is analyzed. After that the corresponding hardware system and software system are developed. After all, the system is tested in experiment condition.The research shows that the PHM prototype system of helicopter transmission system has five main functions, which are composed of data acquisition and real-time condition monitoring, signal playback and feature selection, incipient damage detection and pattern recognition, conditon trend prognosis and RUL prediction, and data & information & knowledge management.更多还原