最优知识传递宽残差网络输电线路螺栓缺陷图像分类

目的 输电线路螺栓图像具有分辨率低和视觉信息较差的特点,针对螺栓缺陷图像分类时教师网络（大模型）参数量大、学生网络（小模型）分类精度低的问题,提出了一种最优知识传递宽残差网络输电线路螺栓缺陷图像分类方法,来弥补使用大小模型进行分类任务的局限性。方法 首先改变大模型宽度,即拓宽网络特征表达维度来增加向小模型传递的螺栓缺陷知识和简化小模型结构至3个残差块;然后为了选出传递螺栓缺陷知识性能最优的大模型,提出知识偏差的概念来可视化大模型向小模型螺栓缺陷知识传递的程度,综合分析不同宽度大模型利用知识蒸馏算法与注意力转移算法分别指导小模型训练后的分类准确率,并用知识偏差来确定最优知识传递模型;最后将最优知识传递模型利用知识蒸馏算法与注意力转移算法相结合指导小模型训练,尽可能提升小模型的螺栓缺陷分类性能。结果 在自建螺栓缺陷图像分类数据集上进行了验证,结果表明宽度为5的大模型向小模型传递螺栓缺陷知识性能最优,使小模型的螺栓缺陷分类准确率提高了5.56%,小模型与最优大模型的分类准确率只差2.17%,知识偏差为0.28,且小模型的参数量仅为大模型参数量的0.56%。结论 本文提出的最优知识传递宽残差网络输电线路螺栓缺陷图像分类方法,弥补了大小模型螺栓缺陷图像分类的局限性,实现了精度与资源消耗的平衡。     

无人机遥感技术的水利水电工程施工现场监控

针对水利水电工程施工现场异常因素多、监控力度薄弱问题,本研究通过融合ARM控制技术和LoRa通信技术实现现场信息的采集控制,利用STM32F103VET6单片机计算单元实现不同种类的传感器电路信息的计算,通过STM32F103VET6芯片和Cortex-M3内核的处理器实现多种信息的采集控制,本研究还通过构建卷积分神经网络模型实现异常数据信息下的图像识别,提高了监控力度.通过试验,本文研究方法大大提高了恶劣天气下的图像识别能力.     

Guaranteeing performance for uncertain nonlinear systems with bounded state constraint and mismatching condition

The control design problem for the uncertain nonlinear system with bounded state constraint and mismatching condition is considered in this paper. The uncertainty in the system, which may be due to unknown system parameters and external disturbance, is nonlinear and time‐varying. The state of the system is constrained to be bounded. The system does not satisfy the (global) matching condition. A creative one‐to‐one state transformation is proposed by converting the bounded states into the unbounded ones. A step‐by‐step state transformation is proposed to convert the mismatched system into a matched system. The robust control is then proposed based on the transformed system. The control is demonstrated to be able to guarantee the uniform boundedness and uniform ultimate boundedness of the system in the presence of uncertainty, while the state constraint can be always guaranteed.     

Dynamic analysis of a novel time‐lag four‐dimensional fractional‐order financial system

In this paper, a novel four‐dimensional fractional‐order financial system (FFS) with time delay is presented. Unlike traditional bifurcation analysis of financial systems, the selection rules of two bifurcation points within the system are discussed. In addition, the motion state of the system in the vicinity of two bifurcation points are analyzed separately, such that the dynamic analysis of this novel nonlinear fourth‐dimensional FFS is more comprehensive. The detailed dynamical behaviors of this financial system, such as oscillation, stability, and bifurcation points, are deduced via rigorous mathematical analysis. Finally, some simulations are performed to verify the dynamic characteristics of the FFS around the two bifurcation points which satisfy the selection conditions of the bifurcation point.     

Adaptive Safety Motion Control for Underactuated Hovercraft Using Improved Integral Barrier Lyapunov Function

International Journal of Control, Automation and Systems - In this paper, we investigate the problem of safety motion control for an underactuated hovercraft from subject to safety constraint on...     

对混合属性数据表可行的差分隐私保护方法

为加强隐私保护和提高数据可用性,提出一种可对混合属性数据表执行差分隐私的数据保护方法。该方法首先采用ICMD（insensitive clustering for mixed data）聚类算法对数据集进行聚类匿名,然后在此基础上进行
-差分隐私保护。ICMD聚类算法对数据表中的分类属性和数值属性采用不同方法计算距离和质心,并引入全序函数以满足执行差分隐私的要求。通过聚类,实现了将查询敏感度由单条数据向组数据的分化,降低了信息损失和信息披露的风险。最后实验结果表明了该方法的有效性。     

基于Filter Wrapper模式的特征选择算法*

特征选择是数据挖掘、机器学习和模式识别中始终面临的一个重要问题。针对类和特征分布不均时,传统信息增益在特征选择中存在的选择偏好问题,本文提出了一种基于信息增益率与随机森林的特征选择算法。该算法结合Filter和Wrapper模式的优点,首先从信息相关性和分类能力两个方面对特征进行综合度量,然后采用序列前向选择（Sequential Forward Selection, SFS）策略对特征进行选择,并以分类精度作为评价指标对特征子集进行度量,从而获取最优特征子集。实验结果表明,本文算法不仅能够达到特征空间降维的效果,而且能够有效提高分类算法的分类性能和查全率。     

基于多重约束的OpenXML复合文档内容源数据区容错定界算法

在高误码率条件下,接收到的OpenXML复合文档往往因数据含错而无法打开,而在文档修复过程中,对信源中每个源文件内容源数据区的定界经常容易出错,针对该问题,提出了一种基于多重约束的OpenXML复合文档内容源数据区容错定界算法。通过对文档协议冗余的分析与归类,将内容源数据区定界问题转化为对起始位置序列的最佳估计问题。在通过适当放宽匹配条件完成粗定界的基础上,利用冗余信息中的约束关系构建代价函数对观测数据进行筛选,有效地剔除“虚警”情况,进而实现对内容源数据区起始位置序列的最佳估计。仿真结果表明,在参数设置合理的情况下,该方法具有较好的容错定界能力。     

In situ study of the mechanical properties of airborne haze particles

Particulate pollution has raised serious concerns regarding its potential impacts on human health in developing countries. However, much less attention has been paid to the threat of haze particles to machinery and industry. By employing a state-of-the-art in situ scanning electron microscope compression testing technique, we demonstrate that iron-rich and fly ash haze particles, which account for nearly 70% of the total micron-sized spherical haze particles, are strong enough to generate abrasive damage to most engineering alloys, and therefore can generate significant scratch damage to moving contacting surfaces in high precision machineries. Our finding calls for preventive measures to protect against haze related threat.