基于改进模糊SVM的转子振动故障诊断技术

首先对常用的隶属度确定方法进行改进,提出了基于改进模糊支持向量机(FSVM)的融合故障诊断方法,并建立了改进FSVM故障诊断数学模型;然后,利用转子振动模拟实验台对四种典型的转子诊断故障进行模拟,并提取其故障信号特征;最后,通过实例计算分析,验证了该方法在转子振动故障诊断方面是可行的和有效的.     

基于在线LS-SVM算法的变参数混沌时间序列预测

研究利用最小二乘支持向量机(LS-SVM)预测变参数混沌时间序列。变参数混沌系统适合于描述现实中的复杂混沌现象,但由于参数的慢变导致系统动力学特性不断发生变化,基于Tan-kens嵌入定理的建模预测方法难以适用,其时间序列预测可以看作是小样本学习问题。最小二乘支持向量机是在二次损失函数下采用等式约束求解问题的一种支持向量机,保留支持向量机优点同时计算量大大减少。提出用一种具有遗忘机制的最小二乘支持向量机在线递推算法,并引入历史数据的高次项预测变参数混沌时间序列。对典型变参数混沌时间序列的预测结果表明,该方法具有较高预测精度,能快速跟踪预测变参数混沌时间序列。     

基于神经网络的战时装备维修保障能力评估

介绍了神经网络的概念,阐述了BP神经网络的基本原理与方法,在系统分析战时装备维修保障系统的基础上构建了战时装备维修保障系统能力评估指标,设计了BP神经网络模型,通过Matlab仿真软件对结果进行计算分析表明,BP神经网络具有很强的解决非线性问题能力,适用于战时装备维修保障能力评估。     

基于二叉树支持向量机的装备故障诊断研究

本文提出了一种基于二叉树支持向量机的某型装备故障诊断方法,应用该方法构造了多故障分类器,并通过某型装备故障实例验证了该方法故障诊断结果的正确性,证明了该故障诊断方法具有良好的鲁棒性和推广性。     

最小二乘支持向量机的在线学习算法研究

针对最小二乘支持向量机在线逼近过程中求解矩阵维数逐渐扩大的问题,提出了一种最小二乘支持向量机的在线学习算法。借助滚动窗口的思想,建立一个随时间滑动的建模数据区间,在线逼近中通过接受新数据删除旧数据来保持数据区间长度不变,同时数据不断更新,从而实现模型的在线学习。仿真结果表明了这种学习算法的有效性。     

一种具有电压支撑功能的改进无功注入策略

电压跌落是输配电系统面临的最严重的问题之一。随着分布式光伏电源在电力系统中的渗透率越来越高,国内外最新标准要求光伏逆变器必须具备低电压穿越能力。为了消除有功功率振荡,在传统的正/负序控制基础上提出了一种改进的有功/无功注入策略;针对现有电压支撑方法缺乏电压抬升范围控制,将三相电压幅值作为控制目标,详细推导了适用于不同电压跌落情况下的数学模型,得出为保证各相电压均保持在安全运行范围之内所需注入无功功率参考值。仿真结果验证了所提策略的正确性和有效性。     

基于灰色近似支持向量机的加速度计参数预测

温度、振动等环境载荷使得石英挠性加速度计参数产生漂移,直接影响了惯导系统的测量精度和性能,石英挠性加速度计参数变化趋势为非线性的,很难用常规的建模方法进行趋势预测。基于灰色理论适合进行小样本、贫信息不确定型系统建模以及近似支持向量机不需要求解二次规划就能求得非线性模型参数的优点,提出了基于灰色近似支持向量机进行石英挠性加速度计参数预测的方法。为了验证该方法的有效性,针对自然贮存的加速度计进行了固定周期的参数标定,结果表明灰色近似支持向量机具有很好的预测效果。     

海上待战模式下的导弹保障辅助决策模型

在大规模、高强度、海上待战的作战模式下,海军导弹武器装备的高效及时补充,对于作战部队战斗力的恢复和持续具有越来越重要的意义。文章根据时效性和经济性原则,围绕保障关系的决策建立辅助决策模型,帮助指挥员快速地、科学地确定保障关系,为制定精确、快捷、高效的保障方案提供辅助决策,具有一定的参考和借鉴意义,并通过实例对模型进行了验证。     

老龄飞机的故障控制

介绍了老龄飞机的故障控制方式，阐述了预防性维修和及时处理故障在老龄飞机保障中的重要性。     

Integration of lessons from recent research for “Earth to Mars” life support systems

Development of reliable and robust strategies for long-term life support for planetary exploration must be built from real-time experimentation to verify and improve system components. Also critical is incorporating a range of viable options to handle potential short-term life system imbalances. This paper revisits some of the conceptual framework for a Mars base prototype which has been developed by the authors along with others previously advanced (“Mars on Earth^®”) in the light of three years of experimentation in the Laboratory Biosphere, further investigation of system alternatives and the advent of other innovative engineering and agri-ecosystem approaches. Several experiments with candidate space agriculture crops have demonstrated the higher productivity possible with elevated light levels and improved environmental controls. For example, crops of sweet potatoes exceeded original Mars base prototype projections by an average of 46% (53% for best crop) ultradwarf (Apogee) wheat by 9% (23% for best crop), pinto bean by 13% (31% for best crop). These production levels, although they may be increased with further optimization of lighting regimes, environmental parameters, crop density etc. offer evidence that a soil-based system can be as productive as the hydroponic systems which have dominated space life support scenarios and research. But soil also offers distinct advantages: the capability to be created on the Moon or Mars using in situ space resources, reduces long-term reliance on consumables and imported resources, and more readily recycling and incorporating crew and crop waste products. In addition, a living soil contains a complex microbial ecosystem which helps prevent the buildup of trace gases or compounds, and thus assist with air and water purification. The atmospheric dynamics of these crops were studied in the Laboratory Biosphere adding to the database necessary for managing the mixed stands of crops essential for supplying a nutritionally adequate diet in space. This paper explores some of the challenges of small bioregenerative life support: air-sealing and facility architecture/design, balance of short-term variations of carbon dioxide and oxygen through staggered plantings, options for additional atmospheric buffers and sinks, lighting/energy efficiency engineering, crop and waste product recycling approaches, and human factor considerations in the design and operation of a Mars base. An “Earth to Mars” project, forging the ability to live sustainably in space (as on Earth) requires continued research and testing of these components and integrated subsystems; and developing a step-by-step learning process.