Pseudo‐3D RSSI‐based WSN localization algorithm using linear regression

Receiver Strength Signal Indication based Wireless Sensor Networks offer a cheap solution for location‐aware applications. For a final breakthrough these systems need fast deployment and easy auto‐configuration. In this study, we use the real‐life iMinds test bed to expand a two‐dimensional localization algorithm to the pseudo third dimension with very low additional computational time. Our experiments show that this fast three‐dimensional algorithm has no outliers and avoids manual calibration. Our algorithm has lower position errors than a maximum likelihood algorithm with a mean square error cost function. Furthermore, with non‐parametric statistical tests, we show that our previously designed two‐dimensional preprocessing performs equally well in pseudo‐three dimensions: the preprocessing reduces the position error in a statistically significant way.Copyright © 2013 John Wiley & Sons, Ltd.     

InP/InGaAs探测器光响应度与波长关系的研究

对于正面光入射的InP/InGaAs探测器,入射光会在空气、增透膜、InP盖层和InGaAs层之间发生多次反射.为了研究其对光响应度的影响,我们测量发现:探测器的光响应度会随探测波长出现非平坦的峰谷曲线,并且通过对该曲线上峰谷波长的简单数学处理,可以提取出已封装器件的结构参数和材料参数,而且这些参数与实验曲线符合得很好.利用该方法可以简单方便地提取出已封装器件的实际结构参数和材料参数.     

BGA空洞控制的回归分析研究

阐述了一种基于回归算法的建模方式,通过假设论证的方法分析和推断出了各因子对BGA空洞大小的影响力及相关性,并对模型在现实意义上进行了一系列的分析,为BGA空洞的控制提供了一种科学的和量化可控的方法.     

一种改进的自适应增强-支持向量回归机的故障预测方法

针对支持向量回归(SVR)方法对突变故障预测精度较低的问题,提出了一种改进的自适应增强算法(AdaBoost)提升SVR故障预测性能。该方法通过AdaBoost算法获取训练样本中突变点的权重并构造加权支持向量回归机增强突变点的训练,以提高对突变故障预测精度。利用自适应权重裁减方法剔除权重较小的样本点,来提高算法的训练速度。将本文方法用于发动机磨损元素的时间序列预测中,一步预测相对误差达到了0.025. 实验结果表明该方法在保证预测精度的前提下有效地提高了故障预测速度。     

固体燃料超燃冲压发动机燃烧室的数值仿真

基于一种固体燃料超燃冲压实验发动机的实验数据,使用数值模拟软件分别对超燃冲压发动机燃烧室的初始状态以及启动后的燃烧流动进行数值模拟。采用用户自定义函数方式给定PMMA燃料进口边界。数值模拟结果显示:燃烧室流场特性分布符合理论分析;燃烧室固体燃料壁面的燃料退移速率与实验数据有一定差异,但是整个燃面沿轴向的燃速分布规律与实验值近似;沿轴向的燃面附近的压力分布与实验结果较为吻合。研究结果表明:该数值计算模型较为合理,对固体燃料超燃冲压发动机的理论研究具有一定参考价值。     

基于支持向量回归的图像超分辨率重建算法

文章提出一种新的基于支持向量回归(SVR)和稀疏表示的图像超分辨重建算法。SVR对输入数据有良好预测输出类别能力。图像统计表明,图像块可以从过完备字典中通过稀疏线性组合很好的表示。对一幅低分辨率输入图像,可以将图像超分辨问题视为在高分辨图像中估计其像素位置。与传统的支持向量回归方法相比,本文采用的特征是不同类型的图像块的稀疏表示。研究表明,稀疏表示作为特征对噪声有一定的鲁棒性。实验结果表明,本文方法与传统支持向量回归方法相比在图像重建质量上有一定的优势。     

基于数据驱动的500 kV高压并联电抗器过流误报警在线判别方法研究

为提升500 kV并联电抗器在线监测的准确性,提出了基于逻辑回归和主成分分析的并联电抗器过流误报警在线判别方法。首先,分析了并联电抗器所在母线电压互感器以及电抗支路电流互感器测量误差对过流报警条件概率的影响。其次,定义逻辑回归拟合函数的中心点及不确定域为故障检测的二维关键指标,并利用主成分分析法生成一维综合指标。最后,通过将检测日综合指标与设备正常运行工况下综合指标进行对比,对检测日过流报警的正确性进行判别。该方法只需使用高压并联电抗器所在母线的电压测量有效值和监控系统所报出的过流报警信号,无需额外增加测量设备。以某实际500 kV高压并联电抗器作为算例进行的仿真实验,验证了该方法的有效性。     

Efficient localization of target in large scale farmland using generalized regression neural network

Although simple to implement, the traditional trilateration technique is generally associated with significant location estimation errors because of highly nonlinear relationship between Received Signal Strength Indicator (RSSI) and distance. In case of agricultural farmland, there is always noise uncertainty in the RSSI measurements because of signal propagation issues such as NLOS, multipath propagation, and reflection. In the context of such environmental dynamicity, the localization algorithm must be efficient in terms of Localization Accuracy and Execution Speed to provide real‐time performance. The Generalized Regression Neural Network (GRNN) is a noniterative highly parallel neural architecture with the capability to get trained quickly using very few training samples. This paper introduces a range free GRNN localization algorithm as an alternative to the traditional range‐based trilateration technique for a large scale wheat farmland. This paper also presents the modified Optimal Fitted Parametric Exponential Decay Model (OFPEDM)‐based signal path loss model to deal with the issue of environmental dynamicity. The evaluation of localization performance of the trilateration and the proposed GRNN‐based approaches is carried out with the help of Wireless Sensor Network (WSN) using three path loss models, namely, Log Normal Shadow Fading (LNSM), Original OFPEDM, and proposed Modified OFPEDM. For all these implementations, the proposed GRNN algorithm demonstrates superior localization performance (localization accuracy of the order of few centimeters) over traditional trilateration irrespective of nonlinear system dynamics, path loss model, and environmental dynamicity. The execution speed of the proposed algorithm is of the order of few milliseconds.     

基于回归模型的高职院校生均成本研究

MATLAB是一种高度集成的计算机语言,具有强大的科学计算功能及灵活的程序设计流程。文中充分利用MATLAB的优势,通过非线性回归分析函数nlinfit建立了两个非线性回归模型,对我国高职院校的生均成本进行了一些研究。     

Multi-Fidelity High-Throughput Optimization of Electrical Conductivity in P3HT-CNT Composites

Combining high-throughput experiments with machine learning accelerates materials and process optimization toward user-specified target properties. In this study, a rapid machine learning-driven automated flow mixing setup with a high-throughput drop-casting system is introduced for thin film preparation, followed by fast characterization of proxy optical and target electrical properties that completes one cycle of learning with 160 unique samples in a single day, a > 10 ×  improvement relative to quantified, manual-controlled baseline. Regio-regular poly-3-hexylthiophene is combined with various types of carbon nanotubes, to identify the optimum composition and synthesis conditions to realize electrical conductivities as high as state-of-the-art 1000 S cm⁻¹. The results are subsequently verified and explained using offline high-fidelity experiments. Graph-based model selection strategies with classical regression that optimize among multi-fidelity noisy input-output measurements are introduced. These strategies present a robust machine-learning driven high-throughput experimental scheme that can be effectively applied to understand, optimize, and design new materials and composites.