用于行人轨迹预测的场景限制时空图卷积网络

目的 针对行人轨迹预测问题,已有的几种结合场景信息的方法基于合并操作通过神经网络隐式学习场景与行人运动的关联,无法直观地解释场景对单个行人运动的调节作用。除此之外,基于图注意力机制的时空图神经网络旨在学习全局模式下行人之间的社会交互,在人群拥挤场景下精度不佳。鉴于此,本文提出一种场景限制时空图卷积神经网络（scene-constrained spatial-temporal graph convolutional neural network,Scene-STGCNN）。方法 Scene-STGCNN由运动模块、基于场景的微调模块、时空卷积和时空外推卷积组成。运动模块以时空图卷积提取局部行人时空特征,避免了时空图神经网络在全局模式下学习交互的局限性。基于场景的微调模块将场景信息嵌入为掩模矩阵,用来调节运动模块生成的中间运动特征,具备实际场景下的物理解释性。通过最小化核密度估计下真实轨迹的负对数似然,增强Scene-STGCNN输出的多模态性,减少预测误差。结果 实验在公开数据集ETH （包含ETH和HOTEL）和UCY （包含UNIV、ZARA1和ZARA2）上与其他7种主流方法进行比较,就平均值而言,相对于性能第2的模型,平均位移误差（average displacement error,ADE）值减少了12%,最终位移误差（final displacement error,FDE）值减少了9%。在同样的数据集上进行了消融实验以验证基于场景的微调模块的有效性,结果表明基于场景的微调模块能有效建模场景对行人轨迹的调节作用,从而减小算法的预测误差。结论 本文提出的场景限制时空图卷积网络能有效融合场景和行人运动,在学习局部模式下行人交互的同时基于场景特征对轨迹特征做实时性调节,相比于其他主流方法,具有更优的性能。     

结合Transformer与非对称学习策略的图像检索

目的 图像检索是计算机视觉领域的一项基础任务,大多采用卷积神经网络和对称式学习策略,导致所需训练数据量大、模型训练时间长、监督信息利用不充分。针对上述问题,本文提出一种Transformer与非对称学习策略相结合的图像检索方法。方法 对于查询图像,使用Transformer生成图像的哈希表示,利用哈希损失学习哈希函数,使图像的哈希表示更加真实。对于待检索图像,采用非对称式学习策略,直接得到图像的哈希表示,并将哈希损失与分类损失相结合,充分利用监督信息,提高训练速度。在哈希空间通过计算汉明距离实现相似图像的快速检索。结果 在CIFAR-10和NUS-WIDE两个数据集上,将本文方法与主流的5种对称式方法和性能最优的两种非对称式方法进行比较,本文方法的mAP(mean average precision)比当前最优方法分别提升了5.06%和4.17%。结论 本文方法利用Transformer提取图像特征,并将哈希损失与分类损失相结合,在不增加训练数据量的前提下,减少了模型训练时间。所提方法性能优于当前同类方法,能够有效完成图像检索任务。     

Optimal reduced-order interval observer design for uncertain continuous-time linear systems

This article investigates the reduced-order interval observer (R-IO) design technique for continuous-time linear systems with unknown external disturbances and measurement noises. First, we propose a coupled R-IO structure with more design degrees of freedom, and it can be directly applied not only to solve the difficulty of the error system cooperativity construction but to relax the constraint on the sensor measurement noises. Second, the R-IO existence condition is derived as a set of matrix equations (MEs), and a complete solution, explicitly showing the available design parameters, to such an R-IO is further obtained by solving the MEs. Third, using the solution, an integrated optimization indicator of the R-IO performance is built as the valid selection mechanism of these parameters. Finally, the efficiency of the obtained results is illustrated through a numerical example and a practical example.     

基于相机感知距离矩阵的无监督行人重识别

在跨场景、跨设备的行人重识别中虽然增加了可利用的行人数据,但由于行人姿态不同、部分遮挡现象,难以避免引入样本噪声,在聚类过程中易生成错误的伪标签,造成标签噪声,影响模型的优化。为减弱噪声影响,应用相机感知的距离矩阵对抗相机偏移引起的样本噪声问题,利用对噪声鲁棒的动态对称对比损失减少标签噪声,提出基于相机感知距离矩阵的无监督行人重识别算法。在聚类前通过更改度量行人特征相似度的距离矩阵,利用相机感知距离矩阵来增强类内距离度量准确性,减少由于拍摄视角不同对聚类效果造成的负面影响。同时,结合噪声标签学习方法,进行损失设计,提出动态对称对比损失函数,联合损失训练,不断精炼伪标签。在DukeMTMC-reID和Market-1501两个数据集上进行实验,验证了提出方法的有效性。     

CNN parameter design based on fault signal analysis and its application in bearing fault diagnosis

As a representative deep learning network, Convolutional Neural Network (CNN) has been extensively used in bearing fault diagnosis and many good results have been reported. In Prognostics and Health Management (PHM) field, the CNN’s input size is usually designed as a 1D vector or 2D square matrix, and the convolution kernel size is also defined as a square shape like 3 × 3 and 5 × 5, which are directly adopted from the image recognition. Though satisfying results can be obtained, CNN with such parameter specifications is not optimal and efficient. To this end, this paper elaborated the physical characteristics of bearing acceleration signals to guide the CNN design. First, the fault period under different fault types and shaft rotation frequency were used to determine the size of CNN’s input. Next, an exponential function was involved in fitting the envelope of decaying acceleration signal during each fault period, and signal length within different decaying ratios was used to define the CNN’s kernel size. Finally, the designed CNN was validated with the Case Western Reserve University bearing dataset and Paderborn University bearing dataset. Results confirm that the physics-guided CNN (PGCNN) with rectangular input shape and rectangular convolution kernel works better than the baseline CNN with higher accuracy and smaller uncertainty. The feasibility of designing CNN parameters with physics-guided rules derived from bearing fault signal analysis has also been verified.     

Optimized tracking control using reinforcement learning strategy for a class of nonlinear systems

This paper is to develop a simplified optimized tracking control using reinforcement learning (RL) strategy for a class of nonlinear systems. Since the nonlinear control gain function is considered in the system modeling, it is challenging to extend the existing RL-based optimal methods to the tracking control. The main reasons are that these methods' algorithm are very complex; meanwhile, they also require to meet some strict conditions. Different with these exiting RL-based optimal methods that derive the actor and critic training laws from the square of Bellman residual error, which is a complex function consisting of multiple nonlinear terms, the proposed optimized scheme derives the two RL training laws from negative gradient of a simple positive function, so that the algorithm can be significantly simplified. Moreover, the actor and critic in RL are constructed by employing neural network (NN) to approximate the solution of Hamilton–Jacobi–Bellman (HJB) equation. Finally, the feasibility of the proposed method is demonstrated in accordance with both Lyapunov stability theory and simulation example.     

一种改进残差深度网络的多目标分类技术

由于受场景、视角、光照、尺度变化以及局部变形等因素的影响,对重叠目标、拥挤目标、小目标的识别精度较低,提出了一种改进多支路的残差深度卷积神经网络来提高多目标识别的准确度。首先,在第一个卷积残差块layer1后保留恒等映射的同时,增加一个1×1的短接分支尽可能多的保留原始特征;再平行嵌入一个修改激活函数RELU6的空间_通道注意力机制模块（CBAM）;最后这三个特征图进行融合。融合后的特征层着重关注空间和通道中比较显著的信息,从而增强特征图的特征表达能力,以至于卷积神经网络(CNN)获得更多的判别特征,从而大大提高物体识别精度。在FashionMNIST和Cifar10两个数据集的对比性实验显示改进的resnet50算法是准确性-速度较为折中的目标识别模型。     

基于改进深度学习的风机油污识别

针对风机设备油液渗漏影响风机正常运行亟需解决的对风机设备油污的识别问题,提出了一种基于改进深度学习的风机油污检测方法。基于深度学习在目标检测中的应用特点,对目标检测网络YOLOv5n（You Only Look Once v5n）进行改进,将原网络中的非极大抑制（Non Maximum Suppression,NMS）替换为Soft-NMS,降低了网络的误检率,添加CA (Coordinate Attention)注意力机制,增强了模型对目标的定位能力,改进原网络损失函数为α-IoU（Alpha- Intersection over Union）损失函数,提高了边界框检测的准确度。实验结果表明:模型平均精度提升了8.1%,查全率提高了19.1%,网络推理速度提高了28.6%。改进后的模型能准确检测风机油污,有效解决了风机实际运行中油液渗漏所带来的问题。     

基于ARM内核单片机的体育比赛模拟控制系统

构建体育比赛模拟控制系统,提高对体育比赛的现场智慧化管理和信息调度能力,提出基于ARM内核单片机的体育比赛模拟控制系统设计方法。采用实时视频和图像检测方法,构建体育比赛现场赛事的模拟数据传输和图像分析模型,通过过程化的线性调度和赛事流程规划和控制,采用可视化的视频远程监控方法实现对体育比赛的现场赛事的视景仿真和动作特征点分析,采用赛事分布视频动态特征提取方法,实现体育比赛现场赛事管控特征分析。构建体育比赛现场的动态参数数据集预处理模块、信息传输模块、视频监测与回放模块和人机交互模块等,采用ARM内核单片机控制方法建立体育比赛模拟控制系统的核心处理器,构建体育比赛过程状态可视化云平台的服务器,通过五层结构体系设置的模拟控制系统的体系结构模型,结合总线传输控制技术实现控制系统硬件和软件设计。测试结果表明,设计的体育比赛模拟控制系统人机交互性和视频传输控制能力较好,控制收敛性较好,能实现体育比赛现场智能监控和动态控制。     

基于MultiResUNet-SMIS的皮肤黑色素瘤图像分割

针对现有的皮肤黑色素瘤病灶分割精度不高的问题,结合现有卷积神经网络方法提出皮肤黑色素瘤图像分割方法 MultiResUNet-SMIS.首先,依据皮肤黑色素瘤成像特点,引入不同空洞率的空洞卷积替换普通卷积,在参数量相同的前提下扩大感受野,使网络模型能够适用于多尺度病灶分割任务;其次加入空间和通道注意力机制以重新分配特征权重,扩大感兴趣特征影响,抑制无关特征;最后融合Focal loss与Dice loss提出一种新的loss函数FD loss用于计算回归损失,解决前景背景像素不均衡问题,进一步提高网络模型的分割精度.实验结果表明,MultiResUNet-SMIS在ISIC-2018数据集上的Dice指数、IoU指数以及Acc准确率分别达到了89.47%、82.67%、96.13%,与原MultiResUNet以及UNet、UNet++、DeepLab V3+等主流方法相比, MultiResUNet-SMIS在皮肤黑色素瘤图像分割中具有更好的效果.