Surrogate role of machine learning in motor-drive optimization for more-electric aircraft applications

Motor drives form an essential part of the electric compressors, pumps, braking and actuation systems in the More-Electric Aircraft (MEA). In this paper, the application of Machine Learning (ML) in motor-drive design and optimization process is investigated. The general idea of using ML is to train surrogate models for the optimization. This training process is based on sample data collected from detailed simulation or experiment of motor drives. However, the Surrogate Role (SR) of ML may vary for different applications. This paper first introduces the principles of ML and then proposes two SRs (direct mapping approach and correction approach) of the ML in a motor-drive optimization process. Two different cases are given for the method comparison and validation of ML SRs. The first case is using the sample data from experiments to train the ML surrogate models. For the second case, the joint-simulation data is utilized for a multi-objective motor-drive optimization problem. It is found that both surrogate roles of ML can provide a good mapping model for the cases and in the second case, three feasible design schemes of ML are proposed and validated for the two SRs. Regarding the time consumption in optimizaiton, the proposed ML models can give one motor-drive design point up to 0.044 s while it takes more than 1.5 mins for the used simulation-based models.     

Pose matters: Pose guided graph attention network for person re-identification

Person re-Identification (reID), aiming at retrieving a person across different cameras, has been playing a more and more important role in the construction of smart city and social security. For deep-learning-based reID methods, it has been proved that using local feature together with global feature could help to give robust representation for person retrieval. Human pose information can provide the locations of human skeleton to effectively guide the network to pay more attention to these key areas, and can also help to reduce the noise distractions from background or occlusions. Based on human pose, a Pose Guided Graph Attention (PGGA) network is proposed in this paper, which is a multi-branch architecture consisting of one branch for global feature and two branches for local key-point features. A graph attention convolution layer is carefully designed to re-assign the contribution weight of each extracted local feature by modeling the similarity relations. The experimental results demonstrate the effectiveness of our approach on discriminative feature learning. Our model achieves the state-of-the-art performance on several mainstream evaluation datasets. A plenty of ablation studies and different kinds of comparison experiments are conducted to prove the effectiveness of this work, including the tests on occluded datasets and cross-domain datasets. Moreover, we further design supplementary tests in practical scenario to indicate the advantage of our work in real-word applications.     

Vulnerability analysis of the Chinese coupled aviation and high-speed railway network

The integrated aviation and High-Speed Railway (HSR) transportation system plays a vital role for today’s inter-city transportation services. However, an increasing number of unexpected disruptions (such as operation failures, natural disasters, or intentional attacks) pose a considerable threat to the normal operation of the system, especially on ground transfer, leading to the extensive research on its vulnerability. Previous approaches mainly focus on interruptions within a single transportation mode, neglecting the role of ground transfer which serves as a coupled connection between aviation and High-Speed Railway. This paper proposes a network-based framework for evaluating the vulnerability of the Chinese Coupled Aviation and High-Speed Railway (CAHSR) network from the viewpoint of ground transfer interruption. Taking the end-to-end travel time and passenger flow information into consideration as an evaluation measure and analyzing from the perspective of urban agglomerations, an adaptive method is developed to identify the critical cities and further investigate their failure impacts on the geographic distribution of vulnerability. In addition, the proposed model explores variations of vulnerability under different failure time intervals. Based on the empirical study, some major conclusions are highlighted as follows: (A) Only a few cities show significant impacts on the network’s vulnerability when ground transfer interruptions occurred. (B) The distribution of vulnerability is not proportional to the distance between failure city and influenced city. (C) The vulnerability is more serious in the morning and evening when the ground transfer is disconnected. Our findings may provide new insights for maintenance and optimization of the CAHSR network and other real-world transportation networks.     

A hybrid deep neural network based on multi-time window convolutional bidirectional LSTM for civil aircraft APU hazard identification

Safety is one of the important topics in the field of civil aviation. Auxiliary Power Unit(APU) is one of important components in aircraft, which provides electrical power and compressed air for aircraft. The hazards in APU are prone to cause economic losses and even casualties. So,actively identifying the hazards in APU before an accident occurs is necessary. In this paper, a Hybrid Deep Neural Network(HDNN) based on multi-time window convolutional neural network-Bidirectional Long Short-Term M...     

Formation control for networked multiagent systems with a minimum energy constraint

Minimum-energy formation achievement problems for networked multiagent systems are investigated, where information networks with leaderless and leader-follower structures are respectively addressed and information networks are randomly switching. The critical feature of this work is that the energy constraint is minimum in the sense of the linear matrix inequality, but limited-budget control and guaranteed-cost control cannot realize a minimum-energy formation. Firstly, the leaderless minimum-energy formation control problem is converted into an asymptotic stability one via a nonsingular transformation and state space decomposition, and based on linear matrix inequality techniques, sufficient conditions for analysis and design of leaderless minimum-energy formation achievement are proposed, respectively, which can be solved by the generalized eigenvalue method. Then, main results of minimum-energy formation achievement of leaderless networked multiagent systems are extended leader-follower networked multiagent systems, where the asymmetric property of the leader-follower information network is well dealt with by two nonsingular transformations. Finally, two simulation examples are shown to verify the main results for minimum-energy formation achievements of leaderless and leader-follower networked multiagent systems, respectively.     

Nightside 135.6 nm emission enhancements of mid/low latitudes during geomagnetic storms as observed by the ionosphere PhotoMeter (IPM) on the Chinese meteorological satellite FY-3D

IPM has detected nightside 135.6 nm emission enhancements over a wide latitude range, from the sub-auroral latitudes to the equatorial regions during geomagnetic storms. Our work, presented in this paper, uses the data of IPM to understand these 135.6 nm emission enhancements during of geomagnetic storms and studies the variations of total electron content (TEC) and the F₂ layer peak electron density (NmF2) in the region of enhanced emissions. Middle and low latitude emission enhancements are presented during several medium storms in 2018. The variations of both the integrated electron content (IEC) derived from the nighttime OI 135.6 nm emission by IPM and TEC from the International GNSS Service (IGS) relative to the daily mean of magnetically quiet days of per each latitude bin (30°≦geographic latitude < 40°, 15°≦geographic latitude < 30°, 0°≦geographic latitude < 15°, ?15°≦geographic latitude < 0°, ?30°≦geographic latitude < -15°, ?40°≦geographic latitude < -30°) are investigated and show that on magnetically storm day, IEC by IPM always increases, while TEC from IGC may increase or decrease. Even if both increase, the increase of IEC is greater than that of TEC. From the comparison of IEC and TEC during magnetic storms, it can be seen that the enhancement of the nighttime 135.6 nm emissions is not entirely due to the ionospheric change. The time of IEC enhancements at each latitude bin is in good agreement, which mainly corresponds to the main phase time of the geomagnetic storm event and lasts until the recovery phase. The available ground-based ionosonde stations provide the values of NmF2 which match the 135.6 nm emissions measured by IPM in space and time. The variations of NmF2 squared can characterize the variations of the OI 135.6 nm emissions caused by O⁺ ions and electrons radiative recombination. The study results show that the OI 135.6 nm emission enhancements caused by O⁺ ions and electrons radiative recombination (where NmF2 squared increases) are obviously a contribution to the measured 135.6 nm emission enhancements by IPM. The contribution accounts for at least one of all contributions to the measured 135.6 nm emission enhancements by IPM. However, where the NmF2 squared provided by ionosonde decrease or change little (where the OI 135.6 nm emissions cause by O⁺ ions and electrons radiative recombination also decrease or change little), the emission enhancements measured by IPM at storm-time appear to come from the contributions of other mechanisms, such as energetic neutral atoms precipitation, or the mutual neutralization emission (O+ + O-→2O + h? (135.6 nm)) which also occupies a certain proportion in 135.6 nm airglow emission at night.     

Orbital error propagation considering atmospheric density uncertainty

Due to the influence of various errors, the orbital uncertainty propagation of artificial celestial objects while orbit prediction is required, especially in some applications such as conjunction analysis. In the orbital error propagation of artificial celestial objects in low Earth orbits (LEOs), atmospheric density uncertainty is one of the important factors that require special attention. In this paper, on the basis of considering the uncertainties of position and velocity, the atmospheric density uncertainty is also taken into account to further investigate the orbital error propagation of artificial celestial objects in LEOs. Artificial intelligence algorithms are introduced, the MC Dropout neural network and the heteroscedastic loss function are used to realize the correction of the empirical atmospheric density model, as well as to provide the quantification of model uncertainty and input uncertainty for the corrected atmospheric densities. It is shown that the neural network we built achieves good results in atmospheric density correction, and the uncertainty quantization obtained from the neural network is also reasonable. Moreover, using the Gaussian mixture model - unscented transform (GMM-UT) method, the atmospheric density uncertainty is taken into account in the orbital uncertainty propagation, by adding a sampled random term to the corrected atmospheric density when calculating atmospheric density. The feasibility of the GMM-UT method considering atmospheric density uncertainty is proved by the further comparison of abundant sampling points and GMM-UT results (with and without considering atmospheric density uncertainty).     

Quick identification of guidance law for an incoming missile using multiple-model mechanism

A guidance law parameter identification model based on Gated Recurrent Unit (GRU) neural network is established. The scenario of the model is that an incoming missile (called missile) attacks a target aircraft (called aircraft) using Proportional Navigation (PN) guidance law. The parameter identification is viewed as a regression problem in this paper rather than a classification problem, which means the assumption that the parameter is in a finite set of possible results is discarded. To increase the training speed of the neural network and obtain the nonlinear mapping relationship between kinematic information and the guidance law parameter of the incoming missile, an output processing method called Multiple-Model Mechanism (MMM) is proposed. Compared with a conventional GRU neural network, the model established in this paper can deal with data of any length through an encoding layer in front of the input layer. The effectiveness of the proposed Multiple-Model Mechanism and the performance of the guidance law parameter identification model are demonstrated using numerical simulation.     

遥感场景分类的批处理协方差神经网络方法

针对卫星遥感图像场景分类数据集中存在的局部区域特征异常问题，提出一种采用批处理协方差层的神经网络(CovNN)模型进行遥感场景分类的方法。该方法通过计算全输入通道的局部区域均值实现一种3D批处理协方差算法，能够有效消除局部区域均值的影响，从而更好地处理局部光照过强和局部区域存在无关特征的问题。将其应用于存在局部光照异常和局部无关特征问题的卫星采集AID数据集和NWPU RESISC45数据集中，实验表明CovNN在两个数据集上均取得了超过现有卷积神经网络(CNN)的召回率，可有效降低图像局部区域特征异常的不利影响。