模糊极小极大神经网络在火箭发动机故障分类中的应用

研究了一种用模糊集表示火箭发动机故障模式的神经网络分类器。模糊集是由模糊超立方体聚集形成的集合体，模糊超立方体是一个极小点和极大点构成的用隶属函数表示的ｎ维方盒。极小点和极大点的确定用包含有扩张与收缩阶段的模糊极小极大学习算法实现，这种算法能在一次循环学习中形成非线性模式边界，无需对已知故障模式重新训练就可融合新样本和精炼已存在的故障模式。模糊集用于故障模式分类自然地提供了故障更刘水平分类的有用信     

空间光学精密展开机构展开方法的初步探讨

文章对空间光学精密展开机构（以下简称展开机构）进行分类，分析了几种典型可展开空间光学系统的展开方法和特点。讨论了展开机构设计需要考虑的主要因素，并给出设计建议。     

基于层合板基本性能数据的复合材料层合结构的强度校核方法

通过对[0/45]_8,[0/9]_8两种铺层的玻璃/环氧管状试样做双向载荷强度实验,发现基于层合板基本强度数据的Tsai—Wu准则失效包络线比基于单向板基本性能数据的失效包络线与实验结果的吻合要好得多。特别是按两个单轴强度比值做双轴强度实验确定相互作用系数,其强度曲线比选取相互作用系数F_(xy)~(?)=-1/2的方法与实验结果更为吻合。     

战术导弹全弹FMECA方法研究

论述了故障模式影响及危害性分析(FMECA)的目的和意义，尤其强调了战术导弹全弹FMECA对提高导弹可靠性的作用。讨论了导弹详细设计阶段硬件的FMECA方法，包括自下而上综合方法、约定层次划分、故障严酷度分类等问题。最后，结合工程实践，给出了导弹全弹FMECA的部分应用实例。     

航天电连接器失效预防

介绍了航天电连接器常见的接触不良，绝缘不良，固定不良和密封不良等致命失效形成的原因，探讨了预防航天电连接器失效应采取的措施，并针对不同的失效机理提出了预防失效的可靠性检验项目。     

考虑模糊因素的缺陷安全可靠性评定

在模糊理论和随机可靠性理论的基础上，通过对缺陷安全可靠性评定中各参数的模糊随机性分析，建立了缺陷安全可靠性评定模糊失效概率计算模型，并且结合算例进行验证，取得了较符合实际的结果。     

不完整结构系统刚度可靠性分析

基于结构强度可靠性分析的基础之上，提出了不完整结构(结构系统中有部分元件已失效，但结构未变成机构仍具有一定的承载能力)刚度可靠性的分析方法。该方法考虑了元件因强度失效对结构系统刚度可靠性的影响，以条件概率的形式给出了不完整结构因刚度失效所对应的模式失效概率。并且导出了条件概率所对应的等效安全余量的形式，进而计算在各失效历程中的各失效模式之间的相关性和刚度失效概率。算例表明这样分析符合结构在变成机构之前在使用各阶段刚度可靠性的真实情况，为结构的合理利用起指导作用。     

直升机型号研制中FMEA工作几个问题的考虑与建议

本文根据型号研制中FMEA(故障模式及影响分析)工作所遇到的问题,首次提出了型号生产制造中开展FMEA工作的见解;并就系统与成附件二次配套协调中FMEA工作提出了建议;文中对FMEA工作的实施所要注意的问题进行了讨论。     

Nonthermal Radiation Mechanisms

In this paper we review the possible radiation mechanisms for the observed non-thermal emission in clusters of galaxies, with a primary focus on the radio and hard X-ray emission. We show that the difficulty with the non-thermal, non-relativistic Bremsstrahlung model for the hard X-ray emission, first pointed out by Petrosian (Astrophys. J. 557, 560, 2001) using a cold target approximation, is somewhat alleviated when one treats the problem more exactly by including the fact that the background plasma particle energies are on average a factor of 10 below the energy of the non-thermal particles. This increases the lifetime of the non-thermal particles, and as a result decreases the extreme energy requirement, but at most by a factor of three. We then review the synchrotron and so-called inverse Compton emission by relativistic electrons, which when compared with observations can constrain the value of the magnetic field and energy of relativistic electrons. This model requires a low value of the magnetic field which is far from the equipartition value. We briefly review the possibilities of gamma-ray emission and prospects for GLAST observations. We also present a toy model of the non-thermal electron spectra that are produced by the acceleration mechanisms discussed in an accompanying paper Petrosian and Bykov (Space Sci. Rev., 2008, this issue, Chap. 11).     

Reliability and safety analysis of redundant vehicle management computer system

Redundant techniques are widely adopted in vehicle management computer （VMC） to ensure that VMC has high reliability and safety. At the same time, it makes VMC have special characteristics, e.g., failure correlation, event simultaneity, and failure self-recovery. Accordingly, the reliability and safety analysis to redundant VMC system （RVMCS） becomes more difficult. Aimed at the difficulties in RVMCS reliability modeling, this paper adopts generalized stochastic Petri nets to establish the reliability and safety models of RVMCS. Then this paper analyzes RVMCS oper- ating states and potential threats to flight control system. It is verified by simulation that the reli- ability of VMC is not the product of hardware reliability and software reliability, and the interactions between hardware and software faults can reduce the real reliability of VMC obviously. Furthermore, the failure undetected states and false alarming states inevitably exist in RVMCS due to the influences of limited fault monitoring coverage and false alarming probability of fault mon- itoring devices （FMD）. RVMCS operating in some failure undetected states will produce fatal threats to the safety of flight control system. RVMCS operating in some false alarming states will reduce utility of RVMCS obviously. The results abstracted in this paper can guide reliable VMC and efficient FMD designs. The methods adopted in this paper can also be used to analyze other intelligent systems＇ reliability.