PSO算法优化BP神经网络的EIT图像重建算法

针对阻抗断层图像重建这个严重病态的非线性的逆问题,提出一种基于PSO优化BP神经网络的EIT图像重建算法。该算法在基本BP算法的误差反向传播调整权值的基础上,再引入PSO算法进行权值修正。该算法不仅能很好地适应EIT的病态非线性特性,而且可以克服基本BP算法收敛速度慢和易陷于局部极值的局限。实验结果表明该方法速度快并且能够有效提高图像分辨率。     

Ruland法测定合成分子筛结晶度

本文介绍Ruland方法测量分子筛绝对结晶度的原理和实验结果.这种方法考虑了热振动引起的散射扩散和物质结晶部分晶格的不完善性,并利用标准的散射曲线对康普顿散射进行特殊的吸收校正,从而提高了测量结果的准确性.     

一种锥束CT散射场检测优化方法

针对采用射束衰减网格的散射校正方法需额外增加较大扫描量的问题,提出了基于局部极值点扫描的检测优化方法,首先采集少鼍用于计算散射图像的辅助投影图像,然后采用三次样条插值由少量不同扫描角度的散射图像计算出完整的圆周扫描散射图像集,最后通过与原方法获取的散射图像及散射校正切片图像的对比分析,评估该方法的可行性与可靠性.实验结果表明,该方法仅需获取原方法1/20的散射图像即可完成锥束CT的散射校正,使扫描效率提高了47.6%,同时将散射校正切片误差控制在1%之内.     

基于视觉的焊接图像重建算法研究

利用视觉测量获得焊接图像,如果转换成一维向量,不能精确的得到特征值和特征向量.为了克服上述缺点,提出了一种新的焊接图像重建的方法:基于视觉的焊接图像重建算法研究,这种方法基于二维矩阵而不是一维向量,焊接图像矩阵是基于原始焊接图像矩阵得到一个图像协方差矩阵.经过仿真表明本文算法的均方误差比较小,能够重建出清晰的图像,说明本算法的有效性.     

基于逆问题的电磁成像图像重建算法

应用理论分析方法对电磁成像(EMT)简化传感模型的逆问题进行了求解,对得到的非线性数学方程进行数值计算和求解,完成图像重建算法的构建和图像重建的仿真.仿真重建结果说明了基于逆问题的图像重建算法是成功的.     

基于极限学习机求解正问题的 ECT 图像重建

在电容层析成像(ECT)图像重建迭代类算法中,通常采用线性正问题求解,以加快重建速度,由此产生重建误差。针对这一问题,提出了基于极限学习机(ELM)的非线性ECT正问题求解方法,ELM网络输入为介电常数分布,其输出为预测的电容测量值。将该方法与传统的Landweber迭代算法相结合构成ELM-Landweber迭代算法进行图像重建。为使样本具有较好的代表性,物体分布位置及大小均随机生成,并计算相应的归一化电容值作为ELM网络训练及测试样本,对ELM-Landweber迭代算法进行了仿真与静态实验,并与传统Landweber迭代算法进行比较。实验结果表明,相较于传统Landweber迭代算法,采用ELM-Landweber迭代算法,其算法收敛速度显著提高,重建图像质量得到明显改善。训练样本的平均图像相对误差由0.728减小至0.504,测试样本的平均图像相对误差由0.596减小至0.475。     

粒子滤波算法在ECT图像重建中的应用

针对电容层析成像技术(ECT)的图像重建质量精度较低的问题,提出了一种基于粒子滤波的ECT图像重建方法。首先,分析了ECT图像重建基本原理,以系统状态估计的方式描述了ECT图像重建最优解的搜索过程,并建立了状态空间模型。然后,以线性反投影(LBP)算法的图像重建结果作为初始状态,利用测量信息对从状态空间中获取的随机样本进行最优加权,以获得重建图像的最小方差估计。最后,对5种不同的流型进行了仿真实验。实验结果表明,利用本文方法获得的重建图像误差平均值为42.93%,相关系数平均值为0.813 9,比LBP算法、Landweber迭代算法和IMN-SNOF算法得到的相应指标要好。本文方法是一种有效、精度较高的ECT图像重建方法,为ECT图像重建技术提供了新的途径和手段。     

非延拓有效消除边缘振荡的小波变换方法

针对小波变换进行图像压缩过程中重建图像易引起边缘振荡的问题,提出一种非延拓消除边缘振荡的方法.它通过提升算法与传统的非线性ENO-小波变换结合,把高频分量边缘的能量集中到低频分量上,用小波分量模的极大值对跃变点进行奇异性检测,有效地消除了重建图像引起的边缘振荡.仿真实验表明该方法简单有效.     

Application of particle filtering algorithm to image reconstruction of ECT

针对电容层析成像技术(ECT)的图像重建质量精度较低的问题,提出了一种基于粒子滤波的ECT图像重建方法.首先,分析了ECT图像重建基本原理,以系统状态估计的方式描述了ECT图像重建最优解的搜索过程,并建立了状态空间模型.然后,以线性反投影( LBP)算法的图像重建结果作为初始状态,利用测量信息对从状态空间中获取的随机样本进行最优加权,以获得重建图像的最小方差估计.最后,对5种不同的流型进行了仿真实验.实验结果表明,利用本文方法获得的重建图像误差平均值为42.93％,相关系数平均值为0.813 9,比LBP算法、Landweber迭代算法和IMN-SNOF算法得到的相应指标要好.本文方法是一种有效、精度较高的ECT图像重建方法,为ECT图像重建技术提供了新的途径和手段.     

基于改进灵敏度矩阵的ERT图像重建

电阻层析成像(ERT)重建普遍借助基于灵敏度矩阵的正问题模型来实现。但ERT灵敏度矩阵病态严重,且传感器中心区域的灵敏度值远低于边缘区域的灵敏度值,给高质量的图像重建带来困难。针对上述问题,提出一种改进灵敏度矩阵的方法。用一个专门设计的阈值滤波器,增强中心区域的灵敏度,降低灵敏度矩阵的病态程度。仿真与实验结果表明,改进的灵敏度矩阵可显著降低正问题模化误差,有效地提高重建质量,使重建图像与真实分布更接近,因此方法有较大的实用价值。     

AD590温度传感器的非线性补偿方法

针对集成温度传感器AD590应用中存在的非线性问题,介绍了AD590的特点、工作原理及非线性误差产生的原因,重点讨论了实际应用中AD590在环境温度采集时温度补偿的几种方法,实验结果表明这些方法成本低、误差小、精度高。     

基于RBF神经网络的热敏电阻温度传感器非线性补偿方法

针对热敏电阻温度传感器应用中存在的非线性问题,本文提出了应用RBF神经网络强非线性逼近能力进行非线性补偿的方法。文中介绍了非线性补偿的原理和网络训练方法。该方法不依赖于传感器的模型,而是根据传感器的输入和响应数据,建立补偿模型。结果表明,这种非线性补偿模型具有误差小、精度高、可在线标定、鲁棒性强,与基于BP神经网络的非线性补偿模型相比,网络训练时间短等优点,从而方便了热敏电阻温度传感器在测控系统中的应用。     

基于 V-ResNet 的电阻抗层析成像方法

电阻抗层析成像技术(EIT)因其非侵入和可视化等特性为人体肺部空间特性的监测提供了一种有效的方法。但是EIT的逆问题具有严重的非线性、病态性和欠定性,使得图像重建结果含有严重的伪影。针对上述问题,提出了一种由预映射、特征提取、深度重建以及残差去噪四个模块构成的V-ResNet的深度网络成像算法,实现对场域空间位置和电导率参数分布的重建。该算法有效地增加了前馈信息的多重传递并解决了深度网络的梯度消失问题,同时残差去噪模块有效地平滑了图像边界。采用相对误差(RE)和结构相似度(SSIM)来衡量成像质量,实验得出RE的平均值为0.14,SSIM平均值为0.96。仿真与实验结果表明,基于V-ResNet的成像算法与传统的成像算法相比,图像重建结果边界清晰,空间分辨率高。     

Identification of structural non-linearities using describing functions and the Sherman–Morrison method

In this study, a new method for type and parametric identification of a non-linear element in an otherwise linear structure is introduced. This work is an extension of a previous study in which a method was developed to localize non-linearity in multi-degree of freedom systems and to identify type and parameters of the non-linear element when it is located at a ground connection of the system. The method uses a describing function approach for representing the non-linearity in the structure. The describing function contains only the first harmonic terms. The Sherman–Morrison matrix inversion method is used in the present study to put the response expression in a form where the non-linearity term can be isolated. Using measured responses one can calculate the value of the describing function representation of the non-linear element and thus perform the identification. This new method can be used for type and parametric identification of a non-linear element between any two coordinates of the system. Case studies are given to demonstrate the applicability of the method.     

Free vibration of a laminated composite shell of revolution: Effects of shear non-linearity

Effects of shear non-linearity on free vibration of a laminated composite shell of revolution are investigated using a semi-analytical method based on the Reissner–Mindlin shell theory. The coupling between symmetric and anti-symmetric vibration modes of the shell is considered in the shear deformable shell element employed in this study. The Hahn–Tsai non-linearly elastic shear stress–shear strain relation is adopted. Numerical examples are given for laminated composite circular cylindrical and conical shells with various boundary conditions. The numerical results indicate that shear non-linearity may reduce significantly the fundamental frequencies of cross-ply composite shells of revolution.     

电磁推力轴承刚度非线性的研究

分别用电磁力理论公式和整体模型的有限元方法计算分析了电磁推力轴承的位置刚度和电流刚度随推力盘偏移量x的变化规律及影响因素。结果表明,电磁推力轴承的位置刚度和电流刚度存在着强烈的位移非线性现象,转轴漏磁和材料磁饱和影响了非线性关系的变化趋势。用有限元分析可以得出与实际情况更为符合的刚度曲线,从而为设计电磁轴承的控制系统提供更可靠的依据。     

射频法重油含水率测量仪的研究

本文介绍基于射频阻抗理论对重油的含水量进行快速检测和温度补偿的方法,探讨传感器特性的非线性三维补偿和多参数校正技术在水分检测中的应用,对实验结果进行了误差分析,计算表明最大非线性误差为０．５％。     

A Cutter Orientation Modification Method for the Reduction of Non-linearity Errors in Five-Axis CNC Machining

In the machining of sculptured surfaces, five-axis CNC machine tools provide more flexibility to realize the cutter position as its axis orientation spatially changes. Conventional five-axis machining uses straight line segments to connect consecutive machining data points, and uses linear interpolation to generate command signals for positions between end points. Due to five-axis simultaneous and coupled rotary and linear movements, the actual machining motion trajectory is a non-linear path. The non-linear curve segments deviate from the linearly interpolated straight line segments, resulting in a non-linearity machining error in each machining step. These non-linearity errors, in addition to the linearity error, commonly create obstacles to the assurance of high machining precision. In this paper, a novel methodology for solving the non-linearity errors problem in five-axis CNC machining is presented. The proposed method is based on the machine type-specific kinematics and the machining motion trajectory. Non-linearity errors are reduced by modifying the cutter orientations without inserting additional machining data points. An off-line processing of a set of tool path data for machining a sculptured surface illustrates that the proposed method increases machining precision.     

Optimization strategy for the velocity distribution based on tool influence function non-linearity in atmospheric pressure plasma processing

Atmospheric pressure plasma processing (APPP) is proved to be potential in the fabrication of optical elements with high efficiency and near-zero damage. However, high convergence rate in the figuring process is hard to achieve because of the tool influence function (TIF) non-linearity. Directly solved dwell time map by conventional deconvolution methods does not consider the non-linear thermal effect, which leads to significant figuring error. In this paper, the optimization strategy for TIF non-linearity based on the velocity distribution in APPP is presented. The exponential model of TIF with non-linearity is established by trench experiments. A series of simulations are also conducted to analyze the thermal effect of non-linearity on the figuring process, indicating the TIF constantly changes with velocity distribution. Two evaluation parameters, relative balance factor and velocity concentration factor, are proposed to investigate the figuring capacity of calculated velocity distribution. With two evaluation parameters, the optimization strategy of velocity distribution based on TIF selection is proposed to suppress the non-linearity. Verification experiments are carried out to validate the two optimized TIFs. The results show that high convergence is achieved to be 72.41% and 82.81% for root-mean-square value respectively, which proves the feasibility of the proposed optimization strategy.     

A regression model for analysing the non-linearity of laser triangulation probes

A study on the non-linearity of laser triangulation probes is presented in this paper. The influences of design variables on non-linearity were studied by response surface methodology, and the significant factors were identified. A regression model was developed with which non-linearity can be conveniently evaluated in the design stage of laser triangulation probes. The sensitivities of non-linearity to the design variables are evaluated.