Recent progress of machine learning in flow modeling and active flow control

In terms of multiple temporal and spatial scales, massive data from experiments, flow field measurements, and high-fidelity numerical simulations have greatly promoted the rapid development of fluid mechanics. Machine Learning(ML) provides a wealth of analysis methods to extract potential information from a large amount of data for in-depth understanding of the underlying flow mechanism or for further applications. Furthermore, machine learning algorithms can enhance flow information and automat...     

我国深空探测对航天材料及工艺的需求

随着我国首次月球采样返回和火星探测器"天问一号"任务的圆满完成,我国深空探测进入了新的发展阶段。本文首先对我国深空探测的现状和发展趋势进行了分析,进而对深空探测面临的极端温度、强太阳电磁辐射、强粒子辐射、尘与尘暴、酸性大气等环境及对深空探测任务的影响进行了梳理,进而从材料及结构的轻量化、高效热控制、可靠的辐射防护与抗辐射能力、提供可持续的能源、具有较强的耐腐蚀性能、具有较好抗尘与尘暴损伤性能、在轨组装与制造等角度梳理了深空探测对航天材料与工艺的需求,最后从轻质结构机构材料、高效热控制材料、组合辐射防护及耐辐射材料、耐腐蚀材料、耐尘与尘暴材料、高可靠能源材料、3D/4D打印技术等方面给出了深空探测材料与工艺的发展方向。     

An effective crack position diagnosis method for the hollow shaft rotor system based on the convolutional neural network and deep metric learning

In recent years, the crack fault is one of the most common faults in the rotor system and it is still a challenge for crack position diagnosis in the hollow shaft rotor system. In this paper, a method based on the Convolutional Neural Network and deep metric learning (CNN-C) is proposed to effectively identify the crack position for a hollow shaft rotor system. Center-loss function is used to enhance the performance of neural network. Main contributions include: Firstly, the dynamic response of the dual-disks hollow shaft rotor system is obtained. The analysis results show that the crack will cause super-harmonic resonance, and the peak value of it is closely related to the position and depth of the crack. In addition, the amplitude near the non-resonant region also has relationship with the crack parameters. Secondly, we proposed an effective crack position diagnosis method which has the highest 99.04% recognition accuracy compared with other algorithms. Then, the influence of penalty factor on CNN-C performance is analyzed, which shows that too high penalty factor will lead to the decline of the neural network performance. Finally, the feature vectors are visualized via t-distributed Stochastic Neighbor Embedding (t-SNE). Naive Bayes classifier (NB) and K-Nearest Neighbor algorithm (KNN) are used to verify the validity of the feature vectors extracted by CNN-C. The results show that NB and KNN have more regular decision boundaries and higher recognition accuracy on the feature vectors data set extracted by CNN-C, indicating that the feature vectors extracted by CNN-C have great intra-class compactness and inter-class separability.     

Sequential dynamic resource allocation in multi-beam satellite systems:A learning-based optimization method

Multi-beam antenna and beam hopping technologies are an effective solution for scarce satellite frequency resources. One of the primary challenges accompanying with Multi-Beam Satellites(MBS) is an efficient Dynamic Resource Allocation(DRA) strategy. This paper presents a learning-based Hybrid-Action Deep Q-Network(HADQN) algorithm to address the sequential decision-making optimization problem in DRA. By using a parameterized hybrid action space,HADQN makes it possible to schedule the beam patte...     

IABC: An iteratively automatic machine learning boosted hand-eye calibration method for laser displacement sensor

An on-machine measuring (OMM) system with a laser displacement sensor (LDS) is designed for measuring free-form surfaces of hypersonic aircraft’s radomes. To improve the measurement accuracy of the OMM system, a novel Iteratively Automatic machine learning Boosted hand-eye Calibration (IABC) method is proposed. Both the hand-eye relationship and LDS measurement errors can be calibrated in one calibration process without any hardware changes via IABC. Firstly, a new objective function is derived, containing analytical parameters of the hand-eye relationship and LDS errors. Then, a hybrid calibration model composed of two kernels is proposed to solve the objective function. One kernel is the analytical kernel designed for solving analytical parameters. Another kernel is the automatic machine learning (AutoML) kernel designed to model LDS errors. The two kernels are connected with stepwise iterations to find the best calibration results. Compared with traditional methods, hand-eye experiments show that IABC reduces the calibration RMSE by about 50%. Verification experiments show that IABC reduces the measurement deviations by about 25%-50% and RMSEs within 40%. Even when the training data are obviously less than the test data, IABC performs well. Experiments demonstrate that IABC is more accurate than traditional hand-eye methods.     

A point cloud deep neural network metamodel method for aerodynamic prediction

Aiming to reduce the high expense of 3-Dimensional (3D) aerodynamics numerical simulations and overcome the limitations of the traditional parametric learning methods, a point cloud deep learning non-parametric metamodel method is proposed in this paper. The 3D geometric data, corresponding to the object boundaries, are chosen as point clouds and a deep learning neural network metamodel fed by the point clouds is further established based on the PointNet architecture. This network can learn an end-to-end mapping between spatial positions of the object surface and CFD numerical quantities. With the proposed aerodynamic metamodel approach, the point clouds are constructed by collecting the coordinates of grid vertices on the object surface in a CFD domain, which can maintain the boundary smoothness and allow the network to detect small changes between geometries. Moreover, the point clouds are easily accessible from 3D sensors. The point cloud deep learning neural network, which employs re-sampling technique, the spatial transformer network and the fully connected layer, is developed to predict the aerodynamic characteristics of 3D geometry. The effectiveness of the proposed metamodel method is further verified by aerodynamic prediction and robust shape optimization of the ONERA M6 wing. The results show that the proposed method can achieve more satisfactory agreement with the experimental measurements compared to the parametric-learning-based deep neural network.     

缺陷检测中的标签噪声形态恢复方法及其应用

为了保证碳纤维材料产品的可靠性，消除各种可能存在的缺陷，有必要采取有效的手段对其质量进行检查。结合图像识别算法的基于X射线无损检测技术被认为是一种快速有效的解决方案。然而，加工材料的表面通常附有包含各种信息的标签，这些标签会在检测中对缺陷的识别造成干扰，甚至被误检为缺陷。主要研究基于图像特征的产品标签噪声恢复方法及其在缺陷检测中的应用，该方法可以有效地消除噪声，而不影响其余的图像信息，从而确保算法正确识别材料中的缺陷。     

月表陨坑检测轻量化深度学习方法

针对目前基于深度学习的陨坑检测方法存在的模型参数量大和检测速度慢的问题，提出了一种轻量化的深度学习陨坑检测方法。首先，采用通道剪枝方法删减卷积神经网络中冗余的卷积核，得到结构紧凑高效的陨坑检测模型。然后，使用轻量化的深度可分离卷积操作替换基础陨坑检测模型中的标准卷积操作，进一步降低了模型的复杂度。仿真实验结果表明，所提出的轻量化陨坑检测模型能够保证较高的像素预测精度，并且能够适应亮度、图像噪声等干扰因素的影响。同时，与轻量化处理前的模型相比，参数量减少了99.2%，检测速度提升了94%。     

基于DKF的IMU误差预测算法

作为导航领域常用的组合导航方式,全球导航卫星系统(GNSS)/惯性导航系统(INS)组合导航在GNSS信号失锁后,由于惯性测量单元(IMU)误差随时间迅速积累,其定位结果会偏离载体真实位置,导航精度下降.针对此问题,提出了一种长短期记忆网络(LSTM)辅助的算法,称之为深度卡尔曼滤波(DKF)算法.DKF算法的核心思想是使用LSTM训练IMU误差模型,然后通过训练出的模型预测IMU误差,最后将预测的IMU误差代入IMU数据以校正导航结果.仿真结果表明:在200s测试数据上,DKF算法将误差从1.1537m/s降低到0.3746m/s.与平均预测、卡尔曼预测和最小二乘估计等方法相比,DKF算法的误差最小,具有更优越的导航性能.     

An online ensemble semi-supervised classification framework for air combat target maneuver recognition

Online target maneuver recognition is an important prerequisite for air combat situation recognition and maneuver decision-making. Conventional target maneuver recognition methods adopt mainly supervised learning methods and assume that many sample labels are available. However, in real-world applications, manual sample labeling is often time-consuming and laborious. In addition, airborne sensors collecting target maneuver trajectory information in data streams often cannot process information in real time. To solve these problems, in this paper, an air combat target maneuver recognition model based on an online ensemble semi-supervised classification framework based on online learning, ensemble learning, semi-supervised learning, and Tri-training algorithm, abbreviated as Online Ensemble Semi-supervised Classification Framework (OESCF), is proposed. The framework is divided into four parts: basic classifier offline training stage, online recognition model initialization stage, target maneuver online recognition stage, and online model update stage. Firstly, based on the improved Tri-training algorithm and the fusion decision filtering strategy combined with disagreement, basic classifiers are trained offline by making full use of labeled and unlabeled sample data. Secondly, the dynamic density clustering algorithm of the target maneuver is performed, statistical information of each cluster is calculated, and a set of micro-clusters is obtained to initialize the online recognition model. Thirdly, the ensemble K-Nearest Neighbor (KNN)-based learning method is used to recognize the incoming target maneuver trajectory instances. Finally, to further improve the accuracy and adaptability of the model under the condition of high dynamic air combat, the parameters of the model are updated online using error-driven representation learning, exponential decay function and basic classifier obtained in the offline training stage. The experimental results on several University of California Irvine (UCI) datasets and real air combat target maneuver trajectory data validate the effectiveness of the proposed method in comparison with other semi-supervised models and supervised models, and the results show that the proposed model achieves higher classification accuracy.