Bird strike event monitoring in a composite UAV wing using high speed optical fiber sensing system

In this study, high speed bird strikes on a composite structure were successfully monitored using optical fiber sensors. Four multiplexed optical fiber sensors in a single cable were surface-bonded on the leading edge of a composite UAV wing box. In order to acquire those high frequency signals, a newly developed interrogation system was used to process strain signals from four sensors simultaneously at a sampling frequency of 100 kHz. Before the bird strike tests, pre-impact tests using a rubber hammer were performed to verify the suitability of the FBG signal acquisitions. The pre-test data were used in the neural network training procedures to estimate the bird strike locations. Then, the bird strike tests were accomplished using dummy projectiles and a pneumatic gun. The one-pound dummy birds, made of gelatin, hit the leading edge with a maximum speed of 201 km/h. The impact signals were successfully recorded during the tests and their frequency characteristics were then analyzed. Finally, the strike locations were estimated with the neural network which was trained through the pre-tests. The average error was 33.6 mm.     

扑翼飞行器测力平台设计与振动特性仿真分析

扑翼飞行器是基于鸟类仿生学理论衍生出的新型无人飞行器,主要通过机翼周期性上下扑动来提供飞行器所需的升力和推力,在军用和民用飞行器领域均有广阔的应用前景。扑翼飞行器气动力测量作为样机气动性测试的重要手段,多维气动力的准确测量可为新型扑翼飞行器设计优化和飞控品质的提高提供试验数据支持。本文介绍了一种新型组合式多维小量程测力平台,可实现扑翼飞行器六维气动力和气动力矩的测量。考虑到扑翼飞行器机翼上下扑动过程动态测力需求,应用Ansys Workbench有限元分析软件对测力平台进行了模态分析和频响分析,获得在工作频率下的频率响应,仿真结果表明测力平台的振动特性满足设计要求。     

扑翼飞行器气动力测量技术研究进展与展望

扑翼飞行器是一种仿照鸟类飞行的新概念小型无人飞行器,区别于传统固定翼和旋翼飞行器,它主要通过机翼扑动与空气相互作用来提供飞行动力,从而实现飞行器的姿态变动。扑翼飞行器气动特性测试的实质是揭示在非定常流场环境下,扑翼飞行器气动力的产生机制,以及相关扑翼飞行器设计参数对气动特性的影响。通过气动试验方法为扑翼飞行器飞行控制和结构优化等研制工作提供数据支持,将对新型扑翼飞行器理论研究以及飞控品质的提升起到巨大的推动作用。     

Bird migration flight altitudes studied by a network of operational weather radars

A fully automated method for the detection and quantification of bird migration was developed for operational C-band weather radar, measuring bird density, speed and direction as a function of altitude. These weather radar bird observations have been validated with data from a high-accuracy dedicated bird radar, which was stationed in the measurement volume of weather radar sites in The Netherlands, Belgium and France for a full migration season during autumn 2007 and spring 2008. We show that weather radar can extract near real-time bird density altitude profiles that closely correspond to the density profiles measured by dedicated bird radar. Doppler weather radar can thus be used as a reliable sensor for quantifying bird densities aloft in an operational setting, which—when extended to multiple radars—enables the mapping and continuous monitoring of bird migration flyways. By applying the automated method to a network of weather radars, we observed how mesoscale variability in weather conditions structured the timing and altitude profile of bird migration within single nights. Bird density altitude profiles were observed that consisted of multiple layers, which could be explained from the distinct wind conditions at different take-off sites. Consistently lower bird densities are recorded in The Netherlands compared with sites in France and eastern Belgium, which reveals some of the spatial extent of the dominant Scandinavian flyway over continental Europe.     

Handling dataset dependence with model ensembles for skin lesion classification from dermoscopic and clinical images

Dataset dependence affects many real-life applications of machine learning: the performance of a model trained on a dataset is significantly worse on samples from another dataset than on new, unseen samples from the original one. This issue is particularly acute for small and somewhat specific databases in medical applications; the automated recognition of melanoma from skin lesion images is a prime example. We document dataset dependence in dermoscopic skin lesion image classification using three publicly available medium size datasets. Standard machine learning techniques aimed at improving the predictive power of a model might enhance performance slightly, but the gain is small, the dataset dependence is not reduced, and the best combination depends on model details. We demonstrate that simple differences in image statistics account for only 5% of the dataset dependence. We suggest a solution with two essential ingredients: using an ensemble of heterogeneous models, and training on a heterogeneous dataset. Our ensemble consists of 29 convolutional networks, some of which are trained on features considered important by dermatologists; the networks' output is fused by a trained committee machine. The combined International Skin Imaging Collaboration dataset is suitable for training, as it is multi-source, produced by a collaboration of a number of clinics over the world. Building on the strengths of the ensemble, it is applied to a related problem as well: recognizing melanoma based on clinical (non-dermoscopic) images. This is a harder problem as both the image quality is lower than those of the dermoscopic ones and the available public datasets are smaller and scarcer. We explored various training strategies and showed that 79% balanced accuracy can be achieved for binary classification averaged over three clinical datasets.     

Biomedical Event Extraction Using a New Error Detection Learning Approach Based on Neural Network

Supervised machine learning approaches are effective in text mining, but their success relies heavily on manually annotated corpora. However, there are limited numbers of annotated biomedical event corpora, and the available datasets contain insufficient examples for training classifiers; the common cure is to seek large amounts of training samples from unlabeled data, but such data sets often contain many mislabeled samples, which will degrade the performance of classifiers. Therefore, this study proposes a novel error data detection approach suitable for reducing noise in unlabeled biomedical event data. First, we construct the mislabeled dataset through error data analysis with the development dataset. The sample pairs’ vector representations are then obtained by the means of sequence patterns and the joint model of convolutional neural network and long short-term memory recurrent neural network. Following this, the sample identification strategy is proposed, using error detection based on pair representation for unlabeled data. With the latter, the selected samples are added to enrich the training dataset and improve the classification performance. In the BioNLP Shared Task GENIA, the experiments results indicate that the proposed approach is competent in extract the biomedical event from biomedical literature. Our approach can effectively filter some noisy examples and build a satisfactory prediction model.     

Quantifying year-round nocturnal bird migration with a fluid dynamics model

To understand the influence of biomass flows on ecosystems, we need to characterize and quantify migrations at various spatial and temporal scales. Representing the movements of migrating birds as a fluid, we applied a flow model to bird density and velocity maps retrieved from the European weather radar network, covering almost a year. We quantified how many birds take-off, fly, and land across Western Europe to (1) track bird migration waves between nights, (2) cumulate the number of birds on the ground and (3) quantify the seasonal flow into and out of the study area through several regional transects. Our results identified several migration waves that crossed the study area in 4 days only and included up to 188 million (M) birds that took-off in a single night. In spring, we estimated that 494 M birds entered the study area, 251 M left it, and 243 M birds remained within the study area. In autumn, 314 M birds entered the study area while 858 M left it. In addition to identifying fundamental quantities, our study highlights the potential of combining interdisciplinary data and methods to elucidate the dynamics of avian migration from nightly to yearly time scales and from regional to continental spatial scales.     

柔性扑翼气动性能的数值研究

国内外对扑翼飞行的气动特性进行了大量研究,这些研究大多基于简谐扑动的刚性翼,然而大量观察发现鸟或昆虫飞行时,翅膀存在明显的柔性变形,这种变形对其气动性能具有显著的影响。该文针对一简化的二维柔性扑翼模型,采用数值求解N-S方程并耦合扑翼柔性变形方程的计算方法,研究了扑翼柔性变形对其气动性能的影响。结果显示扑翼的柔性变形改变了扑翼周围的涡结构,从而影响扑翼的气动性能;适当的柔性变形能延迟前缘涡的脱落,从而有效地改善扑翼的推进效率,但同时减弱了扑翼在低雷诺数环境中产生高升力的尾迹捕捉机制。     

堆叠自编码器在样本不充足下的轴承故障诊断方法

深度学习作为一种实用的大数据处理工具,在机械智能故障诊断领域也受到广泛关注,许多研究者已经成功地将深度学习模型应用于故障诊断领域.但这些研究往往忽略了两个重要的问题:(1)当原始训练数据集不足时,模型训练过程不理想;(2)网络模型的学习内容不明确.为了克服上述不足,提出一种新的数据增强的堆叠自编码器(DESAE)框架,...     

一种超低空飞行的仿生扑翼飞行器的设计及分析

为研制既具备一定的负载能力,又具有高隐蔽性的飞行器,依据鸟类的飞行方式,设计了一种可以超低空飞行的仿生扑翼飞行器.首先,计算了扑翼飞行器传动机构的自由度,从原理上确定了设计方案的可行性,并确定了飞行器各个构件的尺寸;其次,利用设计软件Creo绘制飞行器三维模型,通过运动仿真得出飞行器的扑动符合设计要求;然后,利用ADA...     

Analyzing Cross-domain Transportation Big Data of New York City with Semi-supervised and Active Learning

The majority of big data analytics applied to transportation datasets suffer from being too domain-specific, that is, they draw conclusions for a dataset based on analytics on the same dataset. This makes models trained from one domain (e.g. taxi data) applies badly to a different domain (e.g. Uber data). To achieve accurate analyses on a new domain, substantial amounts of data must be available, which limits practical applications. To remedy this, we propose to use semi-supervised and active learning of big data to accomplish the domain adaptation task: Selectively choosing a small amount of datapoints from a new domain while achieving comparable performances to using all the datapoints. We choose the New York City (NYC) transportation data of taxi and Uber as our dataset, simulating different domains with 90% as the source data domain for training and the remaining 10% as the target data domain for evaluation. We propose semi-supervised and active learning strategies and apply it to the source domain for selecting datapoints. Experimental results show that our adaptation achieves a comparable performance of using all datapoints while using only a fraction of them, substantially reducing the amount of data required. Our approach has two major advantages: It can make accurate analytics and predictions when big datasets are not available, and even if big datasets are available, our approach chooses the most informative datapoints out of the dataset, making the process much more efficient without having to process huge amounts of data.     

基于胶囊网络的汉字笔迹鉴定算法

由于采集脱机汉字手写样本时忽略了书写人的心理和生理等因素对书写活动的影响,因而传统笔迹鉴定算法的泛化能力较低。针对上述问题,提出基于胶囊网络的汉字笔迹鉴定算法,并构建了跟踪采集数据集以模拟复杂背景下产生的汉字。胶囊网络构建活动向量表示特定类型的实例化参数,通过动态路由算法将活动向量路由到下一层相应的胶囊中,使下一层胶囊得到更清晰的输入信号。分别采用5种算法对HWDB1.1数据集和跟踪采集数据集进行了测试,实验结果表明：本文算法的分类准确率比其他4种算法的都高,HWDB1.1数据集和跟踪采集数据集中算法的分类准确率分别为95.82%, 94.39%;本文算法具有较强的泛化性能,对训练样本数的依赖程度较低,弥补了卷积神经网络池化层的信息丢失缺陷。     

Imbalanced Data Classification Using SVM Based on Improved Simulated Annealing Featuring Synthetic Data Generation and Reduction

Imbalanced data classification is one of the major problems in machine learning. This imbalanced dataset typically has significant differences in the number of data samples between its classes. In most cases, the performance of the machine learning algorithm such as Support Vector Machine (SVM) is affected when dealing with an imbalanced dataset. The classification accuracy is mostly skewed toward the majority class and poor results are exhibited in the prediction of minority-class samples. In this paper, a hybrid approach combining data pre-processing technique and SVM algorithm based on improved Simulated Annealing (SA) was proposed. Firstly, the data pre-processing technique which primarily aims at solving the resampling strategy of handling imbalanced datasets was proposed. In this technique, the data were first synthetically generated to equalize the number of samples between classes and followed by a reduction step to remove redundancy and duplicated data. Next is the training of a balanced dataset using SVM. Since this algorithm requires an iterative process to search for the best penalty parameter during training, an improved SA algorithm was proposed for this task. In this proposed improvement, a new acceptance criterion for the solution to be accepted in the SA algorithm was introduced to enhance the accuracy of the optimization process. Experimental works based on ten publicly available imbalanced datasets have demonstrated higher accuracy in the classification tasks using the proposed approach in comparison with the conventional implementation of SVM. Registering at an average of 89.65% of accuracy for the binary class classification has demonstrated the good performance of the proposed works.     

Image Generation of Tomato Leaf Disease Identification Based on Small-ACGAN

Plant diseases have become a challenging threat in the agricultural field. Various learning approaches for plant disease detection and classification have been adopted to detect and diagnose these diseases early. However, deep learning entails extensive data for training, and it may be challenging to collect plant datasets. Even though plant datasets can be collected, they may be uneven in quantity. As a result, the problem of classification model overfitting arises. This study targets this issue and proposes an auxiliary classifier GAN (small-ACGAN) model based on a small number of datasets to extend the available data. First, after comparing various attention mechanisms, this paper chose to add the lightweight Coordinate Attention (CA) to the generator module of Auxiliary Classifier GANs (ACGAN) to improve the image quality. Then, a gradient penalty mechanism was added to the loss function to improve the training stability of the model. Experiments show that the proposed method can best improve the recognition accuracy of the classifier with the doubled dataset. On AlexNet, the accuracy was increased by 11.2%. In addition, small-ACGAN outperformed the other three GANs used in the experiment. Moreover, the experimental accuracy, precision, recall, and F1 scores of the five convolutional neural network (CNN) classifiers on the enhanced dataset improved by an average of 3.74%, 3.48%, 3.74%, and 3.80% compared to the original dataset. Furthermore, the accuracy of MobileNetV3 reached 97.9%, which fully demonstrated the feasibility of this approach. The general experimental results indicate that the method proposed in this paper provides a new dataset expansion method for effectively improving the identification accuracy and can play an essential role in expanding the dataset of the sparse number of plant diseases.     

Classification of audio radar signals using radial basis function neural networks

Radial basis function (RBF) neural networks are used to classify real-life audio radar signals that are collected by a ground surveillance radar mounted on a tank. Currently, a human operator is required to operate the radar system to discern among signals bouncing off tanks, vehicles, planes, and so on. The objective of this project is to investigate the possibility of using a neural network to perform this target recognition task, with the aim of reducing the number of personnel required in a tank. Different signal classification methods in the neural net literature are considered. The first method employs a linear autoregressive (AR) model to extract linear features of the audio data, and then perform classification on these features, i.e, the AR coefficients. AR coefficient estimations based on least squares and higher order statistics are considered in this study. The second approach uses nonlinear predictors to model the audio data and then classifies the signals according to the prediction errors. The real-life audio radar data set used here was collected by an AN/PPS-15 ground surveillance radar and consists of 13 different target classes, which include men marching, a man walking, airplanes, a man crawling, and boats, etc. It is found that each classification method has some classes which are difficult to classify. Overall, the AR feature extraction approach is most effective and has a correct classification rate of 88% for the training data and 67% for data not used for training.     

扑翼空气动力学研究进展与应用

昆虫、鸟类与蝙蝠等生物具有高超的飞行能力,是扑翼飞行器的主要模仿对象。近年来,扑翼空气动力学领域的研究取得了很大进展,该文主要对其主要研究成果进行综述,重点介绍扑翼空气动力学各研究方向的最新进展,包括昆虫、鸟类与蝙蝠扑翼的主要升力机制,翅膀形态学参数与微观结构、翅膀柔性与动态变形、翼-翼干扰、翼-身干扰、个体间干扰及地面效应等对扑翼气动特性的影响。同时还对仿生扑翼飞行器气动研究的新进展进行了介绍,并提出了扑翼空气动力学所面临的主要问题和挑战。     

Remote nocturnal bird classification by spectroscopy in extended wavelength ranges

We present optical methods at a wide range of wavelengths for remote classification of birds. The proposed methods include eye-safe fluorescence and depolarization lidar techniques, passive scattering spectroscopy, and infrared (IR) spectroscopy. In this paper we refine our previously presented method of remotely classifying birds with the help of laser-induced β-keratin fluorescence. Phenomena of excitation quenching are studied in the laboratory and are theoretically discussed in detail. It is shown how the ordered microstructures in bird feathers induce structural "colors" in the IR region with wavelengths of around 3-6 μm. We show that transmittance in this region depends on the angle of incidence of the transmitted light in a species-specific way and that the transmittance exhibits a close correlation to the spatial periodicity in the arrangement of the feather barbules. We present a method by which the microstructure of feathers can be monitored in a remote fashion by utilization of thermal radiation and the wing beating of the bird.     

An automatic channel selection method based on the standard deviation of wavelet coefficients for motor imagery based brain–computer interfacing

The redundant data in multichannel electroencephalogram (EEG) signals significantly reduces the performance of brain–computer interface (BCI) systems. By removing redundant channels, a channel selection strategy increases the classification accuracy of BCI systems. In this work, a novel channel selection method (stdWC) based on the standard deviation of wavelet coefficients across channels is proposed to identify Motor Imagery (MI) based EEG signals. The wavelet coefficients are calculated by employing a Continuous Wavelet Transform (CWT) filter bank to decompose each trial from the EEG channel. The wavelet coefficient's standard deviation values are obtained across the channels, and these values are then sorted to determine the EEG channels with the highest standard deviation values. The channels with the largest wavelet coefficient divergence are chosen. MI trials are then spatially filtered with the Common Spatial Pattern (CSP), and CWT filter bank-based 2D images are generated from the spatially filtered trials. These images are then classified using a unique nine-layered convolutional neural network (CNN) model that combines two feature maps acquired with differing filter sizes. The proposed framework (stdWC-CSP-CNN) is evaluated using kappa score and classification accuracy on two publically accessible datasets (BCI Competition III dataset IVa and BCI Competition IV dataset 2a). The suggested framework achieved a mean test classification accuracy of 88.8% for dataset IVa from BCI Competition III and 75.03% for dataset 2a from BCI Competition IV, according to the results. The proposed channel selection method outperforms the other channel selection methods examined, according to the results. By rejecting redundant channels, the whole framework can improve the performance of MI-based BCIs.     

A new twist on gyroscopic sensing: body rotations lead to torsion in flapping,flexing insect wings

Insects perform fast rotational manoeuvres during flight. While two insect orders use flapping halteres (specialized organs evolved from wings) to detect body dynamics, it is unknown how other insects detect rotational motions. Like halteres, insect wings experience gyroscopic forces when they are flapped and rotated and recent evidence suggests that wings might indeed mediate reflexes to body rotations. But, can gyroscopic forces be detected using only changes in the structural dynamics of a flapping, flexing insect wing? We built computational and robotic models to rotate a flapping wing about an axis orthogonal to flapping. We recorded high-speed video of the model wing, which had a flexural stiffness similar to the wing of the Manduca sexta hawkmoth, while flapping it at the wingbeat frequency of Manduca (25 Hz). We compared the three-dimensional structural dynamics of the wing with and without a 3 Hz, 10° rotation about the yaw axis. Our computational model revealed that body rotation induces a new dynamic mode: torsion. We verified our result by measuring wing tip displacement, shear strain and normal strain of the robotic wing. The strains we observed could stimulate an insect''s mechanoreceptors and trigger reflexive responses to body rotations.