关键词： Embedded machine learning   real-time   tensorial kernel   tactile sensors   neural networks   singular value decomposition   FPGA  

摘要： This paper presents a novel hardware architecture of the Tensorial Support Vector Machine (TSVM) based on Shallow Neural Networks (NN) for the Single Value Decomposition (SVD) computation. The proposed NN achieves a comparable Mean Squared Error and Cosine Similarity to the widely used one-sided Jacobi algorithm. When implemented on an FPGA, the NN offers 324x faster computations than the one-sided Jacobi with reductions up to 58% and 67% in terms of hardware resources and power consumption respectively. When validated on a touch modality classification problem, the NN-based TSVM implementation has achieved a real-time operation while consuming about 88% less energy per classification than the Jacobi-based TSVM with an accuracy loss of at most 3%. Such results offer the ability to deploy intelligence on resource-limited platform for energy-constrained applications.

关键词： 哈齐高铁   精密控制网复测   数据处理  

摘要： 本文介绍某高寒铁路精密控制网复测的技术要求,结合项目特点及测量数据变化规律分析,给出了高寒地区铁路复测开始的时间;详细描述了高寒高铁精密控制网复测的数据处理流程,给出了超限点的分析更新方法,让超限点得到最优的更新成果.

关键词： Tectonic geomorphology   River basins   Morphometry   Lithological control  

摘要： Morphometric properties of drainage basins, expressed as indices related to their shapes, altitude variability, river networks and stream profiles, are frequently used as proxy information concerning the relative intensity of uplift. Research procedures lead towards division of an area into classes of relative tectonic activity (various rates of surface uplift) and are typically based on compound indices that jointly consider various individual parameters. Given that index approaches are quantitative in nature and comprise large sets of data, their statistical treatment should become an inherent component of the workflow, whereby the results of the statistical analyses should guide the conclusions. However, various data handling issues seem to be seldom addressed or are given inadequate attention. They include selection of variables, correlation analysis, principles of classification in terms of relative tectonic activity, validity of classifications, and criteria for using specific statistical procedures. In this paper we demonstrate the need for rigorous statistical data analysis in a morphometric study, focused on an intraplate mountainous terrain in Central Europe (Orlickie-Bystrzyckie Mountains Block), subject to low or moderate uplift. We analysed 56 drainage basins within the footwall blocks of several prominent fault-generated escarpments. Following literature guidance, the hypsometric and shape characteristics of the drainage basins were considered, whereas clustering and pure-rank based approaches were tested to distinguish classes of relative tectonic activity. The results were compared with indicators of tectonic activity calculated for the escarpments themselves and using DEM-derived terrain attributes indicating enhanced erosion. In addition, we used statistical tests to identify possible lithological influence on the variability of morphometric parameters, which in this study was found to be of secondary importance. Statistical considerations neither

关键词： Upper-Division Undergraduate   Chemoinformatics   Interdisciplinary/Multidisciplinary   Computer-Based Learning   Chemometrics   Computational Chemistry  

摘要： Machine learning, a subdomain of artificial intelligence, is a widespread technology that is molding how chemists interact with data. Therefore, it is a relevant skill to incorporate into the toolbox of any chemistry student. This work presents a workshop that introduces machine learning for chemistry students based on a set of Python notebooks and assignments. Python, one of the most popular programming languages, is open source, free to use, and has plenty of learning resources. The workshop is designed for students without previous experience in programming, and it aims for a deeper understanding of the complexity of concepts in programming and machine learning. The examples used correspond to real data from physicochemical characterizations of wine, a content that is of interest for students. The contents of the workshop are introduction to Python, basic statistics, data visualization, and dimension reduction, classification, and regression.

关键词： 光固化   位置敏感器件   液位检测   积分放大器   分段滤波  

摘要： 精确测量光固化成型设备中光敏树脂的液面位置是确保成型精度的关键,根据激光三角法原理设计基于PSD的液位检测系统,采用点激光和一维位置敏感器件组成液位检测系统。通过测量点激光在PSD上的位移量来获取液面位置的变化量,在分析PSD误差特点基础上,通过优化结构设计、信号采集、检测系统标定和实时数据处理4个方面来提高检测精度。实际测试表明:液位检测系统量程中间位置的误差小于0.27%,两边误差小于0.4%,可满足系统测量精度要求。

关键词： 高层建筑物   沉降监测   回归分析   数据处理  

摘要： 针对建筑物变形监测的数据处理问题,本文基于Matlab提出了几种拟合模型对所得的沉降数据进行分析、预测。本文以某高层建筑物变形监测数据为例,经基准点检查和三点修匀法对原始数据处理以后,通过模型预测分析得出高层建筑物沉降特征与规律。实验结果表明,使用指数函数模型拟合出的实际效果能够更加准确地分析高层建筑物的沉降变形。

关键词： 道路工程   探地雷达   无损检测   路面隐性病害   数据处理  

摘要： 概述了探地雷达现场检测布置方式和关键参数的设置,并基于专业Reflexw数据处理软件,对原始数据滤波处理方法进行了详细探究,结合工程实例中积累的大量雷达检测图像,着重对沥青路面典型隐性病害特征进行分析,并通过现场取芯,验证了探地雷达识别的准确性。研究结果表明,使用探地雷达可有效实现对路面结构内部隐性病害的无损探测。

关键词： 频率   无线电监测   数据处理分析   GIS  

摘要： 为监测民航空管设备电磁环境,该文基于无线电开场测试方法,研发设计了空管无线电监测与分析系统。该系统对空管设备环境的电磁信号进行采集与测量,计算频域、时域的参数特性,结合地理信息等数据融合,展示信号定位结果,分析设备台站状态,经数据比对反映出设备性能及周围环境的变化趋势。系统的应用可为民航空管部门提供设备运维、频率规划和台址选择等方面的参考依据。