Image classification method on class imbalance datasets using multi-scale CNN and two-stage transfer learning

Neural Computing and Applications - Image classification tasks widely exist in many actual scenarios, including medicine, security, manufacture and finance. A major problem that hurts algorithm...     

Visualization-based disentanglement of latent space

Neural Computing and Applications - In recent years, selecting manipulation of data attributes by changing latent code using auto-encoder has received considerable scholarly attention . However,...     

A deep multi-source adaptation transfer network for cross-subject electroencephalogram emotion recognition

Neural Computing and Applications - In real-world application of affective brain–computer interface (aBCI), individual differences across subjects and non-stationary characteristics of...     

An integrated inbound logistics mode with intelligent scheduling of milk-run collection,drop and pull delivery and LNG vehicles

Although it provides a feasible inbound logistics solution for steady production and low inventory management, the Milk-run mode inevitably leads to a high transportation costs due to the features of small-batch and high-frequency delivery. In order to break through the defections of the existing inbound logistics mode, an integrated inbound logistics (IIL) mode with low-carbon and high efficiency is established. An intelligent scheduling method combines Milk-run collection with drop and pull delivery together. Moreover, the LNG vehicles are simultaneously used in the whole process. With AJ company’s auto-parts inbound logistics as a case, the IIL mode is formulated with a mixed integer mathematical model. The genetic algorithm coded with Matlab is used to find the optimal solution. The results show that when compared with the original Milk-run mode, the IIL mode brings massive reductions in driving mileage, wait time and waste gas emission. It can make significant benefits in both economic and social sense. Therefore, it is entirely reasonable for management of industries to believe that the IIL mode will be a feasible and promising alternative for inbound logistics.

     

A steel surface defect inspection approach towards smart industrial monitoring

Journal of Intelligent Manufacturing - With the advance in Industry 4.0, smart industrial monitoring has been proposed to timely discover faults and defects in industrial processes. Steel is widely...     

基于D-FNN的混沌铁磁谐振系统非线性补偿控制方法

针对电网混沌铁磁谐振系统产生的混沌现象,提出基于动态模糊神经网络的混沌铁磁谐振系统非线性补偿控制方法。该方法采用动态模糊神经网络来逼近系统的非线性部分,消除了过电压的混沌现象,将系统稳定到目标位置,实现了对系统的非线性补偿控制。Matlab仿真结果表明,基于动态模糊神经网络的非线性补偿控制方法控制结果正确,响应快速。     

基于Ethernet网络化测试系统的组建

随着现代测试技术、网络技术和计算机技术的发展,测试系统的体系结构和运行管理发生了根本性的变化,通过Ethernet技术对测试进行管理已经成为测试技术发展的趋势。对网络化测试的整体结构进行了详细分析,并给出了基于WinPcap技术进行网络通信层开发的关键技术,最后对测试服务器的功能进行了介绍。结果表明,该系统性能良好,工作稳定。为网络化测试系统的实现提供了一种解决方案,具有一定的实用性和推广价值。     

面向IC装备功能仿真平台的多模式测试方法

为了实现全面对IC装备以及装备功能仿真平台进行功能测试,通过分析装备命令响应特点,提出了包含单指令测试、逻辑指令序列测试以及随机指令序列测试的多模式测试方法;单指令测试验证设备以及由相关设备组成的系统的单指令响应正确性;逻辑指令序列从控制系统角度验证设备或系统的连续动态运行性能;随机指令序列测试验证设备或系统的动态随机响应性能;实际应用表明,提出的测试方法可以满足IC装备功能仿真平台的测试需求。     

基于三维依赖矩阵的故障预测策略设计

针对不同基于状态监测的故障预测方法难以在一个平台下集成和组合的问题,设计了基于故障、测试、衰退信息三维依赖矩阵的故障预测策略设计框架,在框架内给出了故障预测需求设计和预测用测试选择的基本方法,并在灰色预测和支持向量回归预测的基础上,通过融合粒子群和声算法给出了一种灰色支持向量回归智能组合衰退趋势预测分析方法,结合故障预测实例验证了所给方法的有效性和准确性.     

光电火控系统指向精度CCD非接触测量方法研究

为满足某光电火控系统中二维转台指向精度的高精度测量需求,提出了一种利用CCD图像的非接触测量方法;以静态指向精度为测量对象,构建了CCD图像测量系统,系统分辨力为0.44mrad,采用视频序列的数字图像自动拼接方法得到单幅宽视场全景图,通过图像快速匹配方法得到转台的方位像素坐标值,对转台的指向精度进行了测量;测量实验结果表明:转台的指向精度为0.5mrad,满足指向精度2mrad的设计要求。