自适应循环发动机性能智能在线寻优算法研究

为解决自适应循环发动机性能在线自主优化问题,提出一种适用于三流道自适应循环发动机的智能自主优化控制方法,通过深度确定性策略梯度算法（DDPG）在线优化压比计划,实现控制规律自主寻优。本文在基准发动机特性图上给出了不同外涵面积下的等推力特性曲线,在此基础上给出了基准的最低油耗控制规律曲线。当发动机性能退化、或存在个体差异等偏离时,基准控制规律不再能使发动机性能最优,DDPG算法利用前期存储的数据对压比指令修正量进行训练学习,自主调整发动机控制规律。整机数值仿真结果表明,调整后亚声速巡航单位耗油率降低7.63%,与最低油耗点比较偏差0.03%。超声速巡航单位耗油率降低5.04%,与最低油耗点比较偏差0.01%。性能寻优算法可以在发动机性能偏离情况下实现发动机控制规律自主调节,达到最低油耗。     

无监督深度学习单目视觉里程计研究

针对传统单目视觉里程计存在的尺度漂移和尺度不一致问题,提出了一种基于无监督深度学习的单目视觉里程计。首先,联合使用空间几何约束和图像相似性约束,得到长序列尺度一致的深度估计网络和鲁棒的光流估计网络;然后,对密集光流进行采样,得到精确的稀疏对应关系,减少尺度漂移;最后,根据改进的ORB-SLAM初始化方法,选择最优跟踪方式,结合深度信息进行尺度对齐,从而得到全局尺度一致的视觉里程计。在KITTI数据集上进行大量实验,结果表明,相较于ORB-SLAM2和基于深度学习的端到端的视觉里程计系统,该算法在通用评估指标方面性能有明显提升,验证了该算法的有效性。     

人工智能技术在航空发动机孔探检测中的应用进展

孔探是当前航空发动机检修过程中应用最多的无损检测方法,也是孔探图像的唯一获取途径。近年来,深度学习等人工智能方法开始被应用到航空发动机损伤分类、检测中,为实现航空发动机检修智能化提出了一些现行有效的方法,具有重要的工业应用价值。本文概述了航空发动机孔探检测的发展和优缺点,综述了专家系统和机器学习人工智能方法在发动机孔探图像方面的应用进展,总结了基于孔探图像实现航空发动机孔探检测智能化面临的一些挑战。     

Some implications of high luminosity and fast variability in active galactic nuclei

An analysis of the variability timescale against bolometric luminosity for Active Galactic Nuclei shows that a number of sources violate the Eddington limit. The average ratio (L/L_E) is found to change according to group classification. Whilst Seyfert Galaxies have luminosites well within the Eddington limit, Quasars and BL Lac object tend to approach and exceed this limit. Furthermore, BL Lac objects may be further subdivided on the basis of their (L/L_E) ratio. The data on luminosity and variability timescale indicate the existence of two types of active galaxies, one having highly anisotropic emission, probably collimated into jets with pointing angles within few degrees to the line of sight, and the other relating to isotropic emission of photons from the nuclear region. The results are discussed in the light of the high γ-ray luminosity suggested by recent observations of active galaxies.     

利用BP神经网络对卫星无热敏设备温度的估测

卫星上测温资源有限,只有部分设备有测温点,难以准确获得其他无测温点设备的温度.基于反向传播(BP)神经网络对复杂非线性系统优秀的拟合能力,建立了估测卫星上无测温点设备温度的神经网络,以在轨有测温点设备温度为输入层,以在轨无测温点设备为输出层,并使用卫星热试验获得的星上温度遥测数据和在轨无测温点设备的热电偶温度数据进行训...     

基于失效概率的涡轮后机匣全局灵敏度分析

涡轮后机匣是航空发动机安全的关键部件,但是其具有工况复杂、不确定性因素多的缺点。为了探究输入随机变量的不确定性对涡轮后机匣结构失效概率的影响,建立参数化有限元模型进行确定性分析。考虑材料性能、几何参数及外部载荷的不确定性,对涡轮后机匣两种典型失效模式：强度失效以及刚度失效建立极限状态函数;通过构造自适应Kriging 代理模型并结合重要抽样方法评估涡轮后机匣结构失效概率,利用基于失效概率的全局灵敏度方法对涡轮后机匣结构可靠度的不确定性来源进行分析,对各输入随机变量重要性进行排序,构建一种涡轮后机匣全局灵敏度分析框架。结果表明：涡轮后机匣在两种失效模式以及系统失效模式下,发动机推力以及线性膨胀系数对结构失效概率影响最为显著,应对其重点考虑;内外机匣长度以及材料弹性模量对涡轮后机匣结构失效概率影响较小,可对其适当忽略。     

柔性航天器姿态机动轨迹设计及跟踪控制

针对柔性航天器姿态机动的“快速性”及“稳定性”矛盾,研究了一种优化与控制综合的姿态机动轨迹设计与跟踪控制方法。首先,考虑柔性航天器姿态机动过程中既快又稳的需求,建立姿态机动的多目标多约束条件,优化获得姿态机动轨迹,在满足快速性基础上,最大限度提高稳定性;其次,设计新型的快速鲁棒输入成形器(FRIS),与传统输入成形器相比,FRIS具有更短的作用时间及更强的鲁棒性,能够有效抑制柔性附件振动,为姿态机动的“快速性”及“高精度”奠定基础;最后,设计新型自适应连续终端滑模控制器(ACTSMC),避免增益过估计,提高控制精度,实现对期望姿态轨迹的有限时间快速高精度跟踪控制。数值仿真校验了所提方法的有效性。     

A step parameters prediction model based on transfer process neural network for exhaust gas temperature estimation after washing aero-engines

The prediction of Exhaust Gas Temperature Margin(EGTM) after washing aeroengines can provide a theoretical basis for airlines not only to evaluate the energy-saving effect and emission reduction, but also to formulate reasonable maintenance plans. However, the EGTM encounters step changes after washing aeroengines, while, in the traditional models, a persistence tendency exists between the prediction results and the previous data, resulting in low accuracy in prediction. In order to solve the pr...     

浅谈Jscript中使用构造函数创建树对象

本文介绍一种树的方式对网站内容实现动态有序管理，就象windows资源管理器一样，方便直观。最后举了一个部门结构树的例子来阐述。     

Gust load alleviation wind tunnel tests of a large-aspect-ratio flexible wing with piezoelectric control

An active control technique utilizing piezoelectric actuators to alleviate gust-response loads of a large-aspect-ratio flexible wing is investigated. Piezoelectric materials have been exten-sively used for active vibration control of engineering structures. In this paper, piezoelectric mate-rials further attempt to suppress the vibration of the aeroelastic wing caused by gust. The motion equation of the flexible wing with piezoelectric patches is obtained by Hamilton's principle with the modal approach, and then numerical gust responses are analyzed, based on which a gust load alle-viation (GLA) control system is proposed. The gust load alleviation system employs classic propor tional-integral-derivative (PID) controllers which treat piezoelectric patches as control actuators and acceleration as the feedback signal. By a numerical method, the control mechanism that piezo-electric actuators can be used to alleviate gust-response loads is also analyzed qualitatively. Further-more, through low-speed wind tunnel tests, the effectiveness of the gust load alleviation active control technology is validated. The test results agree well with the numerical results. Test results show that at a certain frequency range, the control scheme can effectively alleviate the z and x wing-tip accelerations and the root bending moment of the wing to a certain extent. The control system gives satisfying gust load alleviation efficacy with the reduction rate being generally over 20%.