基于鲁棒迭代学习控制的永磁同步电机转矩脉动抑制

为了提高永磁同步电机转速伺服系统的性能,抑制转矩脉动对控制系统的影响,提出了滑模控制与迭代学习控制相结合的鲁棒迭代学习控制方法(RILC)。设计了迭代学习控制器抑制周期性转矩脉动,提出了滑模控制器提高系统的抗扰动性能,保证系统强鲁棒性及响应快速性。实验结果显示,电机以900r/min的速度运行时,采用鲁棒迭代学习控制可将速度脉动6次谐波幅值由0.89降低到0.56;加入0.5N·m的负载扰动后,转速波动最大值为22r/min,比PI-迭代学习控制法得到的值减小了1.8%。电机以60r/min运行时,采用鲁棒迭代学习控制可将速度脉动6次谐波幅值由4.87降低到0.45;加入0.5N·m的负载扰动的,转速波动最大值为24r/min,比PI-迭代学习控制法得到的值减小了23%。得到的结果表明,鲁棒迭代学习控制方法可有效抑制转矩脉动,同时可提高永磁同步电机转速伺服系统的鲁棒性和动态响应性能。     

Deep ensemble learning with non-equivalent costs of fault severities for rolling bearing diagnostics

Effective fault diagnostics on rolling bearings is vital to ensuring safe and reliable operations of industrial equipment. In recent years, enabled by Machine Learning (ML) algorithms, data-based fault diagnostics approaches have been steadily developed as promising solutions to support industries. However, each ML algorithm exhibits some shortcomings limiting its applicability in practice. To tackle this issue, in this paper, Deep Learning (DL) and Ensemble Learning (EL) algorithms are integrated as a novel Deep Ensemble Learning (DEL) approach. In the DEL approach, the training requirements for the DL algorithm are alleviated, and the accuracy for fault condition classifications is enhanced by the EL algorithm. The DEL approach is comprised of the following critical steps: (i) Convolutional Neural Networks (CNNs) are constructed to pre-process vibration signals of rolling bearings to extract fault-related preliminary features efficiently; (ii) decision trees are designed to optimise the extracted features by quantifying their importance contributing to the faults of rolling bearings; (iii) the EL algorithm, which is enabled by a Gradient Boosting Decision Tree (GBDT) algorithm and a Non-equivalent Cost Logistic Regression (NCLR) algorithm, is developed for fault condition classifications with optimised non-equivalent costs assigned to different fault severities. Case studies demonstrate that the DEL approach is superior to some other comparative ML approaches. The industrial applicability of the DEL approach is showcased via the case studies and analyses.     

基于多模态的在线序列极限学习机研究

单一模态包含的物体信息有限,导致在物体材质识别分类中表现不佳,而传统多模态融合方法在样本训练过程中需要输入所有数据。提出一种多模态的多尺度局部感受野在线序列极限学习机方法。对物体不同模态样本运用改进的特征提取框架,利用多尺度局部感受野感知样本信息提取特征,并将不同模态特征融合后通过在线序列极限学习机进行训练学习。在线序列极限学习机在训练过程中增量式地输入样本进行训练,当有新数据需要训练时无需对所有数据重新训练。在TUM触觉纹理数据库上进行验证,实验结果表明,多模态融合的分类精度高于单模态的分类精度,且改进的特征提取框架可以显著提升分类性能。     

基于随机游走的网络表示学习推荐算法

根据网络结构中的连接关系得到节点的向量表示,进而将节点的向量表示应用于推荐算法可有效提升其建模能力。针对推荐系统中的同质网络,提出结合随机游走的网络表示学习推荐算法。以DeepWalk算法为基础,在随机游走过程中根据节点重要性设定节点游走序列数,并设置终止概率以控制游走长度优化采样结果,在网络表示学习过程中将SkipGram模型融合节点属性信息,同时考虑上下文节点离中心节点的距离获得更准确的推荐结果。实验结果表明,该算法相比DeepWalk、Node2vec等算法具有更高的推荐准确度,并且较好地解决了冷启动问题。     

基于卷积长短时记忆网络的人体行为识别研究

人体行为识别利用深度学习网络模型自动提取数据的深层特征,但传统机器学习算法存在依赖手工特征提取、模型泛化能力差等问题。提出基于空时特征融合的深度学习模型（CLT-net）用于人体行为识别。采用卷积神经网络（CNN）自动提取人体行为数据的深层次隐含特征,利用长短时记忆（LSTM）网络构建时间序列模型,学习人体行为特征在时间序列上的长期依赖关系。在此基础上,通过softmax分类器实现对不同人体行为分类。在DaLiAc数据集的实验结果表明,相比CNN、LSTM、BP模型,CLT-net模型对13种人体行为的总体识别率达到了97.6%,具有较优的人体行为识别分类性能。     

一种基于稀疏字典和残余字典的遥感图像超分辨重建算法

为了解决低分辨率遥感图像超分辨率重建问题,提出一种基于稀疏字典和结构自相似性的遥感图像超分辨方法.首先,引入了稀疏字典学习方法,改善了字典的结构性,得到的字典具有较好的正则性与灵活性.此外,为了更好地重建高分辨率图像,学习初始稀疏字典对和残余稀疏字典对.初始稀疏字典对用于重建初始高分辨率遥感图像;初始高分辨率遥感图像相对于原始高分辨率图像失去了部分细节信息,用残余稀疏字典对对图像的残留信息进行重建.最后,根据遥感图像存在大量的结构相似性特性,利用非局部均值算法对重建图像进行修正.实验结果表明,本算法与其他算法相比,图像质量在主观和客观方面都有所提高,峰值信噪(PSNR)比达到24.690 5,SSIM达到0.736 3.     

Odor learning and foraging success in the parasitoid,Leptopilina heterotoma

A brief 2-hr experience with hostDrosophila larvae in artificial apple-yeast or mushroom microhabitats had three effects on the foraging behavior of femaleLeptopilina heterotoma (Hymenoptera: Eucoilidae) parasitoids under field conditions. First, experienced females released at the center of circular arrays of apple-yeast and mushroom baits were more likely to find a microhabitat over the course of a daily census than naive ones. Second, for those females that found a microhabitat, experienced ones found it faster than naive ones (i.e., experience reduced travel times). Third, females experienced with a particular microhabitat were more likely to find that micro-habitat than an alternative one. Learned preferences were retained for at least one day and possibly as many as seven. Results generally did not depend on the host species (D. melanogaster orD. simulans) with which females were given experience. Females tended to arrive at baits upwind of the point of release, suggesting that odor is involved in finding host microhabitats and, in particular, in learning to find them more effectively. The implications of these results for the application of semiochemicals in biological control are discussed briefly.     

基于一维卷积神经网络的癫痫发作检测

癫痫是一种最常见的危及生命且具有挑战性的神经系统疾病。癫痫脑电信号复杂多样,人工检测癫痫信号耗时耗力,误判率高,不同的医务人员检测出来的结果也不相同,而且临床的原始脑电数据经常会包含多种噪音和生理伪迹,干扰癫痫检测性能。因此,非常有必要进一步研究高效可靠的癫痫自动检测技术,从而减轻医护人员负担。本文针对来自中国301医院收集的临床原始脑电数据进行分析训练,引入了一种基于一维卷积神经网络具有连续双层卷积结构的模型,可以高效稳定地检测到癫痫信号。结果为灵敏度、特异性、准确率和F1-score分别达到96.8%、99.8%、99.6%和96.1%,而且利用GPU进行模型训练的运行时间比对比模型低2到3倍。结果表明,本文引入的基于一维卷积神经网络模型优于现有方法,在癫痫检测性能上高效稳定,对癫痫的辅助诊断具有重要意义。     

学习领域课程及实施对策分析

德国学习领域课程以工作过程结构理论为基础,是与建立在传统学科结构理论基础上的学科课程相对的一种课程模式。与学科课程相比,前者更符合职业学校学生的认知特点。根据学习领域课程开发与具体职业相结合这一特征,提出了学习领域课程开发的两个关键问题：一个是从职业活动形态向职业学校课程形态转化,一个是从职业学校课程形态向职业活动形态转化。认为课程开发的质量取决于对这两个关键问题的把握。要推动优秀课程的开发,就要从提高影响转化的两个重要因素——教师和实训设施入手。     

石油企业创建学习型组织初探

学习型组织是指通过培养组织的学习气氛,充分发挥员工的创造性而建立起来的一种有机的、高柔性、扁平的、符合人性的能持续发展的组织.石油企业在国际化进程中,面临着严峻的挑战,创建学习型组织是提高竞争力的有效途径.在实践中要改善组织成员的思维和行为模式,形成高效的组织学习能力.