基于神经网络的航空发动机全包线PID控制

提出一种基于神经网络的航空发动机全包线PID控制器参数整定方法,在全包线内选定若干离线整定点,在这些点离线整定PID控制器参数kp,ki,kd.以离线整定点参数为训练样本,离线训练BP神经网络,该网络可映射高度H,马赫数Ma与kp,ki,kd的非线性关系,便可用该网络在线整定包线内任意点的kp,ki,kd.用发动机非线性部件级模型为被控对象的数字仿真表明,用上述方法设计的发动机PID控制器在全包线内,都能获得理想的动静态品质.该方法简单易行,效果好,具有实用价值.     

采用主流FPGA的数字电路在线生长进化方法

采用主流FPGA器件构建了在线进化平台,提出了一种适合较大规模数字电路在线进化的生长进化方法。该方法模拟植物生长机理进化以解决进化速度缓慢问题,采用增长验证评估方法取代传统的穷举式验证评估方法来解决在线验证评估难题;应用免疫遗传算法克服遗传算法的早熟收敛问题;采用多参数级联十进制整数编码方法缩短染色体长度;采用生长进化方法成功地进化出了16位加法器和8位乘法器。对比实验结果表明,采用生长进化方法无论是进化出的电路规模,还是进化速度均优于传统的直接进化方法。     

The solar cycle variation of plasma parameters in equatorial and mid latitudinal areas during 2005–2010

Based on the ISL data detected by DEMETER satellite, the solar cycle variation in electron density (Ne) and electron temperature (Te) were studied separately in local daytime 10:30 and nighttime 22:30 during 2005–2010 in the 23rd/24th solar cycles. The semi-annual, annual periods and decreasing trend with the descending solar activity were clearly revealed in Ne. At middle and high latitudes, there exhibited phase shift and even reversed annual variation over Southern and Northern hemisphere, and the annual variation amplitudes were asymmetrical at both hemispheres in local daytime. In local nighttime, the annual variations of Ne at south and north hemispheres were symmetrical at same latitudes, but the annual variation amplitudes at different latitudes differed largely, showing obviously zonal features. As for Te, the phase shift in annual variations was not as apparent as Ne with the increase of latitudes at Southern and Northern hemisphere in local daytime. While in local nighttime the reversed annual variations of Te were shown at low latitudinal areas, not at high latitudes as those in Ne. The correlation study on Ne and Te illustrated that, in local daytime, Ne and Te showed strong negative correlation at equator and low latitudes, but during the solar minimum years the correlation between Ne and Te changed to be positive at 25–30° latitudes in March 2009. The correlation coefficient R between Ne and Te also showed semi-annual periodical variations during 2005–2010. While in local nighttime, Ne and Te exhibited relatively weak positive correlation with R being about 0.6 at low latitudes, however no correlation beyond latitudes of 25° was obtained.     

锥形预混火焰羽流脱落过程的实验诊断

为了研究锥形预混火焰的羽流脱落特性,以丙烷-空气预混火焰为实验对象,通过粒子图像测速技术对锥形预混火焰的羽流速度场进行了实验研究,动态模态分解法对得到的速度场进行了模态分解和主要特征重构还原。同时采用光电倍增器对火焰热释放强度信号进行实时采集,采用双热电偶法对然后热气体的温度进行了重建。结果表明在浮力诱导下燃尽气体与周围环境空气相互剪切形成剪切层（BSL）, BSL 的 KH 不稳定性诱导产生了涡,涡作用于燃尽区附近的流场,扰动传至火焰面附近区域,与火焰剪切层（FSL）的 DL 不稳定性耦合,表现为火焰面的脉动和火焰热释放率的脉动。通过近红外自发光成像得到的热气体自发光强度与温度信号有较强的相关性,热气体呈现出与速度场相似的周期性传播模式,为研究预混火焰的羽流脱落特性提供了新的思路。     

基于人为因素的航空安全模糊综合评价研究

人为因素的控制是航空安全事故预防与控制的关键,通过分析人为因素对航空安全的影响,结合各方面专家的研究成果,建立了基于人为因素的航空安全综合评价体系,利用层次分析法确定了各评价指标权重,构建了模糊综合评价模型,并对人为因素对航空安全的影响进行了综合评价,得出了影响航空安全主要人为因素的综合评价等级。研究结果表明,航空安全状态良好,可以达到飞行安全的目的。     

壁温比对高速钝锥边界层转捩天地差异的影响研究

首先对比了天上飞行状态与地面风洞状态下钝锥边界层的转捩特性,然后利用基于线性稳定性理论的e^N方法对飞行状态与风洞状态下的钝锥边界层进行了转捩预示,最后研究了壁温比对高速钝锥边界层的稳定性及转捩的影响。研究结果表明,在低壁温比条件下,圆锥迎风中心与侧面的边界层先于两者之间区域转捩,转捩形貌与飞行实验结果相似;在高壁温比条件下,圆锥迎风面区域迟于侧面及背风面区域转捩,转捩形貌与风洞试验结果相似。壁温比是造成高速钝锥边界层转捩天地差异的重要影响因素。     

地基设备布站对空间碎片观测能力的影响分析

建立空间碎片地面观测站时,需要用数值仿真工具对其观测能力进行评估和分析.评估观测站能力的重要指标是测站设备对空间碎片的观测量和可观测频度.先进的设备指标和合适的测站位置能够大大提高观测能力.本文使用STK,PROOF等软件,对不同测站雷达和光学设备的观测能力(过境率和探测率)进行了数值仿真和比较分析,得到了一些初步结论,对于空间碎片监测网的设计有一定的参考作用.      

基于深度卷积网络的海洋涡旋检测模型

传统的利用遥感数据检测涡旋的方法通常是基于物理参数、几何特征、手工特征或专家知识。本文重点研究了基于深度学习技术从海表面高度图中识别海洋涡旋的方法。针对海洋卫星拍摄的海洋表面高度图中的涡旋检测问题,提出了一种基于卷积神经网络的多涡旋检测模型,该模型能够准确提取涡旋的特征信息,拟合语义信息与海面高度之间的关系。同时,在用于涡旋检测的最新公开数据集SCSE-Eddy上进行模型训练,以评估基于人工智能的涡旋检测方法性能,该数据集涵盖了15年来位于中国南海及其东部部分海域的每日卫星遥感海表面高度数据。实验结果表明,与现有的方法相比,本文模型取得了更好的检测结果,能够更好地区分相距较近的涡旋。     

基于SAR图像匹配结果可信度评价的INS/SAR自适应Kalman滤波算法

在惯性导航系统（INS）/合成孔径雷达（SAR）组合导航系统中,SAR图像易受斑点噪声的影响,图像匹配的精度对整个导航系统精度的影响十分明显,能够准确地分析SAR图像匹配过程中的误差特性,利用有效的图像匹配信息辅助INS进行组合定位尤为重要。针对上述问题,在加权Hausdorff距离匹配算法的基础上,对影响SAR图像匹配精度的因素进行了分析,提出了一种基于模糊推理的匹配结果可信度评价准则,经过可信度筛选,将有效的匹配信息与INS进行组合;对合理范围内的匹配误差变化引起量测噪声统计特性发生变化,进而导致Kalman滤波精度下降的问题,研究采用改进的Sage-Husa自适应滤波算法对量测噪声方差阵进行动态调整,使其更加接近系统的当前状态。搭建仿真验证平台对所提算法进行了验证,结果表明,该算法能够在合理的匹配误差范围内,有效地筛选出可信的图像匹配结果,相比常规Kalman滤波算法,显著地提升了INS/SAR组合导航系统水平方向的定位精度。     

Approach and landing guidance design for reusable launch vehicle using multiple sliding surfaces technique

An autonomous approach and landing (A&L) guidance law is presented in this paper for landing an unpowered reusable launch vehicle (RLV) at the designated runway touchdown.Considering the full nonlinear point-mass dynamics,a guidance scheme is developed in threedimensional space.In order to guarantee a successful A&L movement,the multiple sliding surfaces guidance (MSSG) technique is applied to derive the closed-loop guidance law,which stems from higher order sliding mode control theory and has advantage in the finite time reaching property.The global stability of the proposed guidance approach is proved by the Lyapunov-based method.The designed guidance law can generate new trajectories on-line without any specific requirement on off-line analysis except for the information on the boundary conditions of the A&L phase and instantaneous states of the RLV.Therefore,the designed guidance law is flexible enough to target different touchdown points on the runway and is capable of dealing with large initial condition errors resulted from the previous flight phase.Finally,simulation results show the effectiveness of the proposed guidance law in different scenarios.