Modeling and performance of HF/OTH radar target classificationsystems

The effects of a class of multipath propagation channels on the performance of a over-the-horizon (OTH) radar target classification system are considered. A Rician frequency-selective fading channel model is employed to characterize the effects of the multipath propagation medium and evaluate the performance of radar target classification systems. The performance of classification algorithms that employ relative amplitude, relative phase, and absolute amplitude measurements as features is investigated. Performance estimates of the various classification algorithms for interesting sets of channel parameters are obtained by means of Monte-Carlo simulations     

Radar target identification using the bispectrum: a comparativestudy

Radar target identification is performed using time-domain bispectral features. The classification performance is compared with the performance of other classifiers that use either the impulse response or frequency domain response of the unknown target. The classification algorithms developed here are based on the spectral or the bispectral energy of the received backscatter signal. Classification results are obtained using simulated radar returns derived from measured scattering data from real radar targets. The performance of classifiers in the presence of additive Gaussian (colored or white), exponential noise, and Weibull noise are considered, along with cases where the azimuth position of the target is unknown. Finally, the effect on classification performance of responses horn extraneous point scatterers is investigated     

Ship target recognition using low resolution radar and neuralnetworks

The classification of ship targets using low resolution down-range radar profiles together with preprocessing and neural networks is investigated. An implementation of the Fourier-modified discrete Mellin transform is used as a means for extracting features which are insensitive to the aspect angle of the radar. Kohonen's self-organizing map with learning vector quantization (LVQ) is used for the classification of these feature vectors. The use of a feedforward network trained with the backpropagation algorithm is also investigated. The classification system is applied to both simulated and real data sets. Classification accuracies of up to 90% are reported for the real data, provided target aspect angle information is available to within an error not exceeding 30 deg     

Netted radar sensing

Future radar applications are beginning to stretch monostatic radar systems beyond their fundamental sensitivity and information limits. Networks of smaller radar systems can offer a route to overcome these limitations; for example, networks of radar sensors can counter stealth technology whilst simultaneously providing additional information for improved target classification. More generally, multiple independent sensors can provide an energetically more efficient collector of radar scatter. The relative merits of non-coherent and coherent networks are introduced and the balance between increased performance, complexity, and cost is discussed.     

Radar target classification using doppler signatures of human locomotion models

The problem of target classification for ground surveillance Doppler radars is addressed. Two sources of knowledge are presented and incorporated within the classification algorithms: 1) statistical knowledge on radar target echo features, and 2) physical knowledge, represented via the locomotion models for different targets. The statistical knowledge is represented by distribution models whose parameters are estimated using a collected database. The physical knowledge is represented by target locomotion and radar measurements models. Various concepts to incorporate these sources of knowledge are presented. These concepts are tested using real data of radar echo records, which include three target classes: one person, two persons and vehicle. A combined approach, which implements both statistical and physical prior knowledge provides the best classification performance, and it achieves a classification rate of 99% in the three-class problem in high signal-to-noise conditions.     

双并联前向神经网络及其在飞行故障检测仿真研究中的应用

首次把飞行故障检测视为一个非线性数据分类问题,从而可望借助人工神经网络来处理。为了克服ＭＬＦＮＮ在数据分类中存在的学习慢与分类精度低,发展了由ＭＬＦＮＮ和ＳＬＦＮＮ并联并可接收编码输入的ＤＰＦＮＮ模型,还将训练ＭＬＰ的有关算法推广到ＤＰＦＮＮ情形。用计算机仿真了若干飞行故障模式并用于测试ＤＰＦＮＮ。     

Methodology of using artificial neural networks for imitating the loading of a single-rotor helicopter

A methodology of creating a loading imitation model of a single-rotor helicopter with artificial neural network algorithms is considered. Stages of creation, the structure, and an algorithm of searching for an optimal neural networks configuration of loading imitation model are presented.     

Neural network learning of low-probability events

In the problem of stationary target identification (STI) via millimeter wave (MMW) seeker radars in heavy clutter environments, it is often necessary to use nonparametric identification procedures, as detailed parametric models of clutter and target returns are generally unavailable. Neural networks provide an attractive approach to perform nonparametric identification. However, when identifying low-probability events, the computational overhead associated with training a neural network can become excessive. This is because low-probability events must be adequately represented in the training sample. We present a modified backpropagation training algorithm based on a likelihood ratio weighting function (LRWF) to train the neural network using a much smaller training set than that required using the standard backpropagation algorithm This algorithm is closely related to the importance sampling technique used in digital communication systems to obtain probability of error estimates by using a much smaller number of simulation runs than what is required with standard Monte Carlo simulation. The modified backpropagation technique results in a significant reduction in computational overhead in training the network, resulting from a substantial reduction in the size of the training set required to achieve a given level of performance. We demonstrate the performance of the algorithm on simulated data for the STI problem in MMW radar     

Comparison of Two Target Classification Techniques

It has been shown that radar returns in the resonance region carry information regarding the overall dimensions and shape of targets. Two radar target classification techniques developed to utilize such returns are discussed. Both of these techniques utilize resonance region backscatter measurements of the radar cross section (RCS) and the intrinsic target backscattered phase. A target catalog used for testing the techniques was generated from measurements of the RCS of scale models of modern aircraft and naval ships using a radar range at The Ohio State University. To test the classification technique, targets had their RCS and phase taken from the data base and corrupted by errors to simulate full-scale propagation path and processing distortion. Several classification methods were then used to determine how well the corrupted measurements fit the measurement target signatures in the catalog. The first technique uses nearest neighbor (NN) algorithms on the RCS magnitude and (range corrected) phase at a number (e.g., 2, 4, or 8) of operating frequencies. The second technique uses an inverse Fourier transformation of the complex multifrequency radar returns to the time domain followed by cross correlation. Comparisons are made of the performance of the two techniques as a function of signal-to-error noise power ratio for various processing options.     

On the resolvability of sinusoidal parameter estimates [andapplication to SAR target features]

The resolvability of 2-D (two-dimensional) sinusoidal parameter estimates is studied. These sinusoids describe the target features in SAR (synthetic aperture radar) applications. We analyze the resolvability by considering the frequency estimates of the sinusoids. Our results may be used by target classification algorithms to better classify radar targets in SAR applications     

M-ary sequential hypothesis tests for automatic targetrecognition

Several forms of sequential hypothesis testing algorithms are described and their performance as classification algorithms for automatic target recognition is evaluated and compared. Several forms of parameteric algorithms, as well as a sequential form of a useful nonparametric algorithm are considered. The primary focus is the design of algorithms for automatic target recognition that produce maximally reliable decisions while requiring, on the average, a minimum number of backscatter measurements. The tradeoffs between the average number of required measurements and the error performance of the resulting algorithms are compared by means of Monte-Carlo simulation studies     

Application of neural networks for synthesizing flight control algorithms. I neural network inverse dynamics method for aircraft flight control

The application of artificial neural networks for aircraft motion control, in particular, for creation of nonlinear algorithms of the aircraft remote control system (RCS) is considered. Aircraft as a control object is represented as a multidimensional nonlinear dynamic system and nonlinear control methods are used to operate this system. The control loop is constructed using the method of inverse dynamics based on the feedback linearization principle. The nonlinear control law is represented as a neural network being learned (adjusted) by recorded or incoming measurements of motion parameters. Synthesis and testing of neural network control algorithms is performed with the fully nonlinear mathematical model of a maneuverable aircraft for three control channels. Simulation results of the closed system are presented.     

引入神经网络的交互式多模型算法

在交互式多模型算法中引入神经网络算法以改进目标跟踪的精度。利用神经网络算法对基于机动目标“当前”统计模型的均值和方差自适应滤波算法进行修改,提高该算法的性能,然后采用交互作用多模型算法跟踪机动目标,提高了机动目标的跟踪精度。     

Adaptive learning approach to landmine detection

We consider landmine detection using forward-looking ground penetrating radar (FLGPR). The two main challenging tasks include extracting intricate structures of target signals and adapting a classifier to the surrounding environment through learning. Through the time-frequency (TF) analysis, we find that the most discriminant information is TF localized. This observation motivates us to use the over-complete wavelet packet transform (WPT) to sparsely represent signals with the discriminant information encoded into several bases. Then the sequential floating forward selection method is used to extract these components and thereby a neural network (NNW) classifier is designed. To further improve the classification performance and deal with the problem of detecting mines in an unconstraint environment, the AdaBoost algorithm is used. We integrate the feature selection process into the original AdaBoost algorithm. In each iteration, AdaBoost identifies the hard-to-learn examples and a new set of features which provide the specific discriminant information for these hard samples is extracted adaptively and a new classifier is trained. Experimental results based on measured data are presented, showing that a significant improvement on the classification performance can be achieved.     

Time-varying autoregressive modeling of HRR radar signatures

A time-varying autoregressive (TVAR) model is used for the modeling and classification of high range resolution (HRR) radar signatures. In this approach, the TVAR coefficients are expanded by a low-order discrete Fourier transform (DFT). A least-squares (LS) estimator of the TVAR model parameters is presented, and the maximum likelihood (ML) approach for determining the model order is also presented. The validity of the TVAR modeling approach is demonstrated by comparing with other approaches in estimating time-varying spectra of synthetic signals. The estimated TVAR model parameters are also used as features in classifying HRR radar signatures with a neural network. In the experiment with two sets of noncooperating target identification (NCTI) data, about 93% of samples are correctly classified     

一种基于u检验的空海目标分类方法

阐述了对于机载雷达,测高精度不高,特别是对远距离目标的测高精度更差,因而利用机载雷达提供的高度信息进行空海目标分类存在很大的不确定性。为了能有效地利用目标高度信息进行空海目标分类,把空海目标分类问题看成是一个u检验问题。首先,给出了用于空海目标分类的判别函数;然后,给出了一种决策规则,并推导出决策门限的计算公式和空中目标误判为海面目标的概率的计算公式;最后,通过仿真表明该算法的简易性和有效性。     

受限资源下制导武器末制导机器视觉技术研究

针对精确制导武器末制导机器视觉技术应用需求,研究了基于卷积神经网络的、针对复杂背景及小目标的自主目标检测识别算法,并分别进行了网络性能评估和硬件资源需求定量评估。针对最优算法,提出了基于嵌入式受限资源下的高精度神经网络压缩算法,并对算法进行了普适性评估。基于GPU嵌入式平台,实现TensorRT路线网络优化,并在速度和精度两方面均衡考虑下,对裁剪与量化算法进行了详细实验验证。实验结果表明,高精度神经网络压缩算法在硬件资源受限条件下,可以有效提升推理速度,最终经算法优化后的网络结构,可以获得3倍以上的速度提升,网络精度损失小于5%。     

Performance complexity study of several approaches to automatictarget recognition from SAR images

A framework which allows for the direct comparison of alternate approaches to automatic target recognition (ATR) from synthetic aperture radar (SAR) images is described and applied to variants of several ATR algorithms. This framework allows comparisons to be made on an even footing while minimizing the impact of implementation details and accounts for variation in image sizes, in angular resolution, and in the sizes of orientation windows used for training. Alternate approaches to ATR are characterized in terms of the best achievable performance as a function of the complexity of the model parameter database. Several approaches to ATR from SAR images are described and the performance achievable by each for a range of database complexities is studied and compared. These approaches are based on a likelihood test under a conditionally Gaussian model, log-magnitude least squared error, and quarter power least squared error. All approaches are evaluated for a wide range of parameterizations and the dependence on these parameters of both the resulting performance and the resulting database complexity is explored. Databases for all of the approaches are trained using identical sets of images and their performance is assessed under identical testing scenarios in terms of probability of correct classification, confusion matrices, and orientation estimation error. The results indicate that the conditionally Gaussian approach outperforms the other two approaches on average for both target recognition and orientation estimation, that accounting for radar power fluctuation improves performance for all three methods, and that the conditionally Gaussian approach normalized for power delivers average performance that is equal or superior to all other considered approaches     

三种用于加工特征识别的神经网络方法综述

加工特征自动识别技术是智能化设计与制造的关键支撑，已有的实用性算法普遍存在学习能力差、识别范围有限和识别速度慢等共性问题。神经网络方法在计算机视觉和模式识别领域获得了巨大成功，其自学习与自适应能力和高速计算等优势也已在加工特征识别中得到初步的展现。对加工特征识别中具有应用潜力的三种不同的神经网络方法进行了研究，剖析了神经网络识别加工特征中的预处理与编码和神经网络结构设计等关键性问题，分析了不同神经网络方法的异同点，总结了当前神经网络识别加工特征的发展方向，为相关领域的研究提供一定的理论指导与技术支持。     

基于粗糙集与RBF神经网络的雷达辐射源信号识别研究

针对新体制、新用途雷达辐射源信号难以识别的特点,用粗糙集理论对雷达辐射源信号进行离散化、属性约简、规则提取,达到分类的目的。用粗糙K-均值聚类方法计算径向基神经网络（RBFNN）的聚类中心,然后用粗糙集理论约简得到的规则构建径向基神经网络对未知雷达辐射源信号进行识别。仿真结果表明,这种基于粗糙集与RBF神经网络的识别模型减少了识别冗余特征、简化了神经网络结构,能有效地识别雷达辐射源信号。