A multi-stage regulation strategy of space manipulators with a free-swinging joint failure

To ensure tasks can be completed after a free-swinging joint failure occurs, a multi-stage regulation strategy of space manipulators is proposed. Considering all terms of the dynamics equation, an evaluation model of the regulation ability (EMRA) of active joints over the fault joint is established based on the fuzzy entropy. And then a multi-stage regulation strategy based on the EMRA is designed to regulate the fault joint. The strategy divides the regulation process into several stages, and select a certain active joint to regulative the fault joint in every stage. With this multi-stage regulation strategy, the fault joint can be regulated to the desired angle without huge torque on regulative joints. The simulation is carried out with a 7-DOF space manipulator, verifying the correctness and effectiveness of the multi-stage regulation strategy. The strategy has three advantages: Coriolis and centrifugal terms are both considered for the first time in selecting the regulative joint, making the selection result more in line with the actual regulation process; The influence of the model uncertainty is eliminated in establishing the EMRA, making the evaluation of regulative ability more precise; The fault joint is successfully regulated to the desired angle without huge torque on regulative joints.     

一种基于Deep-GBM的航空发动机中介轴承故障诊断方法

针对航空发动机中介轴承故障信号难于识别的特点，提出了一种深度梯度提升模型（Deep-GBM）对振动信号特征进行逐层学习以提高分类模型的准确率。开展某型航空发动机中介轴承故障模拟实验，并采用经验模式分解(EMD)方法对采集的振动信号进行分解，提取内蕴模式函数(IMF)分量非线性动力学参数样本熵作为原始故障特征。采用Deep-GBM对中介轴承内环故障、内环和滚动体综合故障、正常、滚棒剥落、滚棒划伤五种不同状态进行识别。实验结果表明，所提出的Deep-GBM故障诊断准确率达到87%，相对于传统的机器学习模型准确率最高提升了28%，并具有良好的泛化能力。      

AP聚类改进免疫算法用于航空发动机故障诊断

在免疫算法训练过程中引入近邻传播(AP)聚类与熵权法,对训练样本进行聚类与权值计算,将权值引入免疫算法中样本选择阈值的计算,以解决训练过程采用固定选择阈值所造成的检测器在部分区域过拟合,部分区域欠拟合的问题。结果表明:改进的免疫算法用于典型非线性函数的寻优时,迭代性能均优于传统免疫算法,并在大部分情况下优于粒子群算法与量子遗传算法,在进行某型发动机故障诊断的实例实验时,改进后的算法的诊断准确率达到98.06%,高于传统免疫算法的92.60%。      

预旋喷嘴径向角度对预旋特性影响的数值研究

为了降低低位进气预旋流路的气动损失，针对带有不同径向角度（0°~30°）预旋喷嘴的预旋系统进行了数值仿真，并对流动特性、温降特性和比熵增特性进行了分析。结果表明:随着预旋喷嘴径向角度的增大，预旋系统无量纲温降先增大后减小，流动阻力减小，预旋系统的流量随之增大。旋转雷诺数为2.3×107时，预旋喷嘴带径向角度的预旋系统无量纲温降比传统喷嘴最大可提高18.3%，存在某一角度使预旋温降特性达到最好。预旋系统内的耗散主要发生在预旋腔和共转盘腔内，径向角度为10°时其比熵增变化量分别占整个预旋系统总体比熵增的42.4%和30.2%；合理设计预旋喷嘴的径向角度，能改善预旋腔内气流的流动效果，并且可以减少整个预旋系统的不可逆损失。      

基于功能脆弱性的空中交通相依网络流量分配

依据空中交通管理与航班运行规则,采用复杂网络理论构建由机场、航路与管制扇区组成的相依网络模型,建立不同扰动策略的影响规则,提出以网络流量熵和交通流损失比变化率为指标识别网络功能脆弱性。并以网络总流量熵最小为目标,建立基于改进遗传算法的网络流量协调分配策略,以降低空中交通相依网络的脆弱性。以民航华北地区空域为原型,发现了其相依网络脆弱性表现规律和脆弱源,采用遗传算法求解网络流量分配方案,优化结果降低了网络总熵值和功能脆弱性,其中机场网络流量分配后效果最为显著,验证了方法的有效性,研究结果可为空中流量管理决策提供一定的理论支撑。     

盘腔进气位置对径向预旋系统的影响

为减少径向预旋系统的流动损失,运用数值模拟方法对不同盘腔进气位置的径向预旋系统进行分析,结果表明:随着盘腔进气径向位置的增加,预旋喷嘴出口气流旋流比随之逐渐减小,径向预旋系统的温降系数及总压损失系数均随之逐渐增大。当旋转雷诺数等于7.9×106,盘腔进气位置由低位向高位变化时温降系数最大可增加525%,同时总压损失系数增加3.93%。径向预旋系统内比熵增主要发生在预旋喷嘴和共转腔,约占系统总体比熵增的80%。随着盘腔进气径向位置的增加,径向预旋系统总体比熵增降低,预旋喷嘴比熵增占比逐渐增大,共转腔比熵增占比逐渐减小。     

高超声速钝头体边界层转捩试验

为研究高超声钝头体边界层转捩以及头部钝度对转捩的影响,在FD-14和FD-14A两座激波风洞中开展了热流、压力扰动和高速纹影显示等综合测量。试验结果表明,转捩雷诺数关于钝度雷诺数的变化显示出转捩反转的趋势。压力扰动的功率谱密度（PSD）分析结果以流向离散分布云图形式显示,边界层高速纹影图像显示了第二模态波的发展、湍流的生成和熵层对边界层结构的显著影响。大头部钝度带来的强熵梯度熵层流动对边界层压力扰动频谱特性和流动结构影响显著,在转捩反转机理中起到重要作用。此外,马赫数对转捩的影响不容忽视。     

液体火箭发动机推进剂泵诱导轮与离心轮的匹配

为获得诱导轮离心轮周向匹配的时序效应对离心泵外特性以及压力脉动的影响规律，阐释相关作用机制，采用基于分离涡仿真（DES）的离心泵三维全流道数值仿真方法，引入熵产理论以及压力脉动强度系数等先进分析方法对不同匹配角度下离心泵内能量损失机制及压力脉动特性进行了研究。结果表明：离心轮诱导轮的时序效应对泵外特性有一定的影响，随着匹配角度的增加，扬程和效率均呈现先减小后缓慢增大的趋势，扬程变化为0.8%，效率变化为1.2%，其影响机制由不同匹配角度下叶轮通道分离涡、叶轮叶片尾迹以及靠近隔舌处扩压器通道回流涡变化决定；时序效应对离心轮扩压器动静干涉效应影响显著，当诱导轮叶片尾缘位于离心轮相邻主叶片中间位置时，能够有效消除3倍频成分，显著降低泵内压力脉动水平，其中动静干涉区域以及隔舌处扩压器叶片表面压力脉动平均降幅分别达到14.5%和16.7%。     

Characterization of ionospheric amplitude scintillations using wavelet entropy detrended GNSS data

The extensive monitoring networks of Global Navigation Satellite System (GNSS) ionospheric scintillation have been established to continuously log observation data. Further, the amplitude scintillation index and the phase scintillation index, which are derived from scintillation observations, are anticipated to accommodate the accuracy requirement of both the user level and the monitoring station level. However, raw scintillation observations essentially measure superposed waveform impairments of GNSS signals propagating through ionosphere and troposphere. It implies that fluctuations of raw scintillation observations are caused by multiple factors from the entire radio propagation environment. Hence, it is crucial to characterize ionospheric scintillations from GNSS observation data. And the characterization is implemented through extracting fluctuations of raw observations merely induced by ionospheric scintillations. Designed to address this problem by means of Fourier filtering detrending, the present work investigates the influence of varying detrending cutoff frequencies on wavelet statistical energy and wavelet entropy distributions of scintillation data. It consequently derives criteria on the optimum detrending cutoff frequency for three types of raw amplitude scintillation data, which are classified by their wavelet energy distributions. Results of the present work verify that detrending with specific optimum cutoff frequencies rather than the fixed and universally applicable one renders the validity and credibility of characterizing ionospheric scintillations as the part of GNSS observation fluctuations purely induced by ionosphere electron density irregularities whose scale sizes are comparable with or smaller than the Fresnel scale.     

转子振动故障的过程功率谱熵特征分析与定量诊断

针对传统旋转机械振动故障定性诊断的不足,提出了1种以信息熵理论为基础的转子故障特征分析与定量诊断方法。在转子试验台上模拟转子振动的4种典型故障,分别得到4个测点多转速下的振动过程故障数据;对这些故障数据进行分析和处理,提取反映其振动过程的故障特征——功率谱信息熵,建立能描述转子振动过程变化规律的多转速多测点下的故障信息熵矩阵,并对振动故障进行分析;通过对转子振动故障信号的实例计算和定量诊断分析,验证了该方法在转子振动故障分类和故障严重程度判断方面是可行的。