转子碰摩故障振动特征的数值分析与辨识方法

为一类集中参数非线性转子碰摩行为进行了数值仿真，获得了各种响应解形态，同时获得了两类混沌路径，即拟周期进入混沌路长和混沌运动突发到周期运动路径，以数值分析结果为基础开展试验研究，观测到的响应形态与仿真结果定性一致，最后，对碰摩故障的振动特征进行了总结并提出了相应的辨识方法，这些结果对工程实际具有一定的参考价值。     

不平衡激励对碰摩转子振动特性的影响

研究了 Jeffcott转子发生动静件碰摩时的非线性振动特性。根据数值计算的结果 ,分析了系统结构参数确定时 ,不平衡激励水平对转子振动分叉特性的影响 ,并给出了系统响应随转速变化的分叉图 ,以及拟周期、混沌时的轴心轨迹、Poincare截面图和频谱 ,揭示了碰摩转子的内在规律     

转子叶轮—定子触碰故障的分析与辨识

基于航空涡轮发动机和工业透平机中的转子叶轮－定子间触故障背景,以力学建模力指导搭建了两类简支型的转子－定子触碰故障实验台架,考察了转子－定子间正常无触碰、早期尖锐型轻微触碰、中期兰尖锐型触碰,晚期平钝型触碰4种状态态时转子振动位移的时域轨迹,观测到了早期尖锐型轻微触磁时伴生的混沌现象;采用FFT方法对上述4种状态试验数据的频率结构作了分析,结果表明当出现三分频以下（包括三分频）的分频特征时可以初步判断叶片发生了早期触碰,而当转子振动位移的频谱图中仅存有两个分频分量甚至抽频分量时,叶片－转子间的触碰故障已非一日之寒。     

加速环境试验技术及其相关研究

加速环境试验是可靠性试验工程的一个新领域。这里简要介绍了加速环境试验技术的基本概念、基本原理和技术现状,并阐述了该技术的相关研究内容     

基于模糊H网模型与面向对象方法的加工设备融合型故障诊断专家系统

以模糊Ｈ 网模型结合面向对象方法,提出一种新型的符号与数值融合型模糊推理专家系统,给出该专家系统的知识表示、知识存储、推理机制与策略。该系统使数值分类算法（如神经网络、模式识别）与符号推理无本质区别,被统一处理。以ＪＣＳ－ ０２０（ＦＡＮＵＣ）加工中心为对象建立了柔性加工设备故障诊断专家系统,以其进给轴驱动电机为例,说明该专家系统的可行性与有效性     

Layout problem of multi-component systems arising for improving maintainability

To improve the mainlainability design efficiency and quality, a layout optimization method for maintainability of multi-component systems was proposed. The impact of the component layout design on system maintainability was analyzed, and the layout problem for maintainability was presented. It was formulated as an optimization problem, where maintainability, layout space and distance requirement were formulated as objective functions. A multi-objective particle swarm optimization algorithm, in which the constrained-domination relationship and the update strategy of the global best were simply modified, was then used to obtain Pareto optimal solutions for the maintainability layout design problem. Finally, application in oxygen generation system of a spacecraft was studied in detail to illustrate the effectiveness and usefulness of the proposed method. The results show that the concurrent maintainability design can be carried out during the layout design process by solving the layout optimization problem for maintainability.     

Sequential fault diagnosis strategy with imperfect tests considering life cycle cost

The problem of sequential fault diagnosis is to construct a diagnosis tree that can isolate the failure sources with minimal test cost. Pervious sequential fault diagnosis strategy generating algorithms only consider the execution cost at application stage, which may result in a solution with poor quality from the view of life cycle cost. Furthermore, due to the fact that uncertain information exists extensively in the real-world systems, the tests are always imperfect. In order to reduce the cost of fault diagnosis in the realistic systems, the sequential fault diagnosis problem with imperfect tests considering life cycle cost is presented and formulated in this work, which is an intractable NP-hard AND/OR decision tree construction problem. An algorithm based on AND/OR graph search is proposed to solve this problem. Heuristic search based on information theory is applied to generate the sub-tree in the algorithm. Some practical issues such as the method to improve the computational efficiency and the diagnosis strategy with multi-outcome tests are discussed. The algorithm is tested and compared with previous algorithms on the simulated systems with different scales and uncertainty. Application on a wheel momentum system of a spacecraft is studied in detail. Both the simulation and application results suggest that the cost of the diagnosis strategy can be reduced significantly by using the proposed algorithm, especially when the placement cost of the tests constitutes a large part of the total cost.     

机械设备智能监测与诊断系统的研究与实现

在深入研究一般机械设备结构特点和功能关系的基础上 ,采用层次诊断模型设计思想 ,建立了一个多知识表示形式、多推理机制相互融合的机械设备状态监测与故障诊断专家系统 ,并成功地应用于摩托车发动机和传动系统的故障诊断中     

基于Wigner分布的齿轮箱振动信号相位估计

时域同步平均是齿轮箱诊断技术的基础,目前这种方法依赖于转速传感器提供相位同步信号。探讨了在没有转速传感器的前提下,由振动信号本身得到相位过零信号的方法。建立了齿轮箱振动信号的一种简化理论模型,通过理论分析得到了一种基于Wigner分布的相位估计方法。将这种方法应用于实际的齿轮箱振动数据,证明这种方法是有效的。     

步降应力加速寿命试验（下篇)——统计分析篇

围绕步降应力加速寿命试验的统计分析问题进行讨论，首先建立了步降应力试验的3步分析方法，然后在Weibull分布场合对3步分析方法中的关键步骤——数据折算过程进行了深入探讨，给出了数据折算过程中估计问题的求解及算法，最后利用实验数据进行验证。该统计分析方法不依赖于加速模型，降低了统计分析的复杂度，求解精度高，迭代收敛性好，具有明显的程序化特征，便于加速寿命试验CAE的编程实现。