基于图分割与协同训练的工业过程半监督软测量方法

在工业过程中，质量检测数据与运行数据的比例失衡导致数据驱动的软测量性能受到限制。基于协同训练算法的半监督建模方法利用海量运行数据获得具有高置信度的伪标签以扩充有效建模数据，然而协同训练建模过程中的特征分割会影响建模中两模型的差异性进而影响最终的模型性能。针对该问题，本文提出一种基于图分割与协同训练偏最小二乘（GSCO-PLS）的半监督软测量算法。首先计算辅助变量之间的相关性指标作为权重获得以特征为顶点的有权无向图，将协同建模中的特征分割转化为平衡最小割图分块问题，以获得两个互相关性最小的特征子集。在此基础上通过协同训练利用运行数据更新初始模型实现半监督建模，通过对初始模型特征分割的优化实现对协同训练半监督软测量方法精度的提升。最后通过TE仿真过程的结果说明本文所提算法的有效性。     

基于BP神经网络的球头铣刀铣削力建模与仿真

将BP神经网络的理论和算法应用于球头刀具铣削力建模的研究中.采用LM算法建立了铣削力预测的神经网络模型,模型中考虑了影响铣削力的加工参数,选取铣削力试验数据对神经网络模型进行训练,用训练好的神经网络模型对铣削力进行仿真.仿真结果表明,用BP神经网络方法建立的铣削力模型能够对铣削力进行准确的预测.     

GM(1,1)模型参数估计的函数变换法及其应用

灰色GM(1，1)模型为试验数据处理与在线监测提供了新的科学方法。为了克服传统GM(1，1)的不足，提出用相除法估计模型参数，建立拟合模型的方法。通过构造一个新数据列，分析新数据列的标准差系数，可以判定拟合精度的高低。选择变换函数，使原始序列变为新的数据序列，进而建立GM(1，1)模型参数估计的函数变换法，建立相应的拟合模型。给出相除法和函数变换法的直接建模的应用实例。实例表明两种方法简单实用。     

D-H法建立连杆坐标系存在的问题及改进

广泛使用的标准D-H法允许坐标系建立在轴线的延长线上,会导致不能针对所有关节连杆进行建模、建立的机器人模型与实体不一致,以及无法进行局部关节位置分析等问题。为了得到统一、直观、准确的模型建立方法,提出了一种统一的实体建模方法--坐标系固定在实体上的D-H表示法（简称CFDH法)。说明了CFDH法坐标系建立的方法、参数获取过程。以一个六自由度机械臂作为仿真对象,分别用D-H法和CFDH法对其进行建模,然后采用Newton-Raphson求逆解,比较两者计算复杂度并对两种建模方法进行整体对比。仿真和实验结果表明,CFDH法满足机器人运动学建模的需求,解决了D-H法存在的问题。     

间歇过程的神经网络模型

间歇过程通常具有非线性，时变和易燃易爆的特点，用常规的建模方法建立起模型比较困难，本文针对间歇聚丙烯过程，利用前馈神经网络建立其数学模型。首先根据实际系统的输入输出建立网络的结构。再用经验数据对网络进行训练，并用未参加训练的数据对网络进行测试，测试的最大误差是0.03MPa，这一误差在要求的范围之内。     

双臂协调机器人相对动力学建模

双臂机器人动力学建模是研究双臂协调运动的关键技术,难点在于同一系统中建立两机械臂之间联系。针对双臂协调机器人动力学建模问题,基于拉格朗日方程建立了两臂动力学模型一般形式,运用矢量解析法求解单臂角速度雅可比矩阵,然后结合虚位移原理,基于相对雅可比矩阵建立双臂协调机器人相对动力学模型,确立两机械臂末端相对作用力与关节参数之间关系。仿真和试验结果验证了矢量解析法与相对动力学模型的正确性,该模型能够求解两机械臂末端相对力,为分析双臂协调运动提供理论依据。     

粒子群优化模糊系统的铣削力建模方法

根据铣削力的特点,提出用模糊系统进行铣削力建模的新方法.根据铣削力的特点和研究目的,设计钛合金铣削试验,试验在数控铣床上进行,用测力仪进行铣削力测量,由试验得到铣削力的训练数据和测试数据.在分析基本粒子群算法优缺点的基础上,将梯度下降算法嵌入基本粒子群算法形成改进粒子群算法.通过训练数据分别用梯度下降算法、基本粒子群算法和改进粒子群算法训练模糊系统,改进粒子群算法的收敛效果优于梯度下降算法和基本粒子群算法.用回归分析对铣削力进行建模,回归函数分别取为指数形式与线性全因子多项式形式,这样得到两种铣削力经验公式.用测试数据对各方法得到的模型分别进行测试,改进粒子群算法训练模糊系统的预测效果优于其他方法.预测结果验证了用改进粒子群算法训练的模糊系统进行铣削力建模是可行的.     

基于数据驱动建模的钣金装配过程误差分析

钣金装配过程的误差分析对于消除钣金装配质量故障具有重要意义。现有分析建模方法由于受钣金装配零件的材料、几何形状和装配工艺的限制,难于对钣金装配过程进行准确建模和误差分析。与分析建模方法不同,基于装配体关键产品特征的历史测量数据提出进行钣金装配过程误差分析的数据驱动建模方法。所提方法由工程经验和数学推导,建立钣金装配过程的多元一阶自回归模型和多元部分线性模型。基于极大似然估计方法和最小二乘核光滑估计方法给出所建立模型的参数和非参数估计。四元四工序典型汽车引擎盖的装配实例证明,所提方法在钣金装配误差分析过程中具有有效性。基于数据驱动的建模方法易于建模,分析结果准确可靠,可为钣金装配过程的误差分析提供新思路。     

无轴承永磁同步电动机径向力模型

将麦克斯韦张量法和磁路分析法相结合,提出了一种计算无轴承永磁同步电动机径向力的方法,建立了考虑转子偏心位移的径向力模型,利用有限元分析方法对表面贴装式无轴承永磁同步电动机进行仿真研究,结果表明,利用该模型能够获得较好的建模精度,验证了麦克斯韦张量法建模的正确性和可行性。     

基于混合驱动的进给系统数字孪生模型自适应更新法

为了解决传统建模方法搭建的数控机床进给系统模型形式单一且缺乏与物理空间的在线实时交互，导致仿真结果精度低的问题，提出一种基于机理模型和数据驱动模型混合驱动的进给系统数字孪生模型自适应更新方法。该方法针对进给系统多领域耦合的特点，利用物理建模工具Simulink/Simscape建立了机电一体化数字孪生机理模型，同时基于长短记忆神经网络建立了数字孪生数据驱动模型，二者混合构成进给系统数字孪生模型。在此基础上，采用带有遗忘因子的递推最小二乘算法更新模型，提高了模型的适应性。通过双轴进给系统的虚实同步运动实验验证了所提方法的有效性。     

Needle steering for robot-assisted insertion into soft tissue: A survey

Needle insertion is a common surgical procedure used in diagnosis and treatment.The needle steering technologies make continuous developments in theoretical and practical aspects along with the in-depth research on needle insertion.It is necessary to summarize and analyze the existing results to promote the future development of theories and applications of needle insertion.Thus,a survey of the state of the art of research is presented on algorithms of needle steering techniques,the surgical robots and devices.Based on the analysis of the needle insertion procedure,the concept of needle steering is defined as a kinematics problem,which is to place the needle at the target and avoid the obstacles.The needle steering techniques,including the artificial potential field method and the nonholonomic model,are introduced to control the needles for improving the accuracy.Based on the quasi-static thinking,the virtual spring model and the cantilever-beam model are developed to calculate the amount of needle deflection and generate the needle path.The phantoms instead of the real tissue are used to verify the models mentioned in most of the experimentations.For the desired needle trajectories,the image-guided robotic devices and some novel needles are presented to achieve the needle steering.Finally,the challenges are provided involving the controllability of the long flexible needle and the properties of soft tissue.The results and investigations can be used for further study on the precision and accuracy of needle insertion.     

穿刺针尖光纤力传感信号的小波变换分析

经皮软组织穿刺过程中,穿刺针尖会历经不同材料特性的器官组织,受到复杂变化的作用力。为了测量针尖处的作用力并鉴定穿刺路径上不同层次的软组织,研制了一种光纤力传感器,并进行了力信号分析。首先,基于法-珀干涉介绍了光纤力传感器的测量原理、光链路设计、及其与针的集成和标定方法。在穿刺力信号分析的基础上,提出利用Mallat算法对力信号进行小波变换,提取信号的特征模式,界定软组织的各层边界。最后,在猪的肝脏和肋条组织样本上进行了针穿刺实验验证。结果显示:光纤力传感器的可测力为0至3.20N,测量精度可达0.01N。得到的穿刺试验结果表明Mallat算法能够有效地区分软组织的类型。研制的光纤传感器满足穿刺力的测量范围、精度和可靠性等要求,所提出的小波变换算法可用于软组织边界的界定,有望用于机器人穿刺控制。     

Contributions in medical needle technologies—Geometry,mechanics, design,and manufacturing

This article summarizes the contributions in research on tissue cutting with needles. The geometry of the needle's cutting edges was analytically defined and expressions for the inclination and rake angles of hollow and solid needles and trocars have been derived. Based on the semi-empirical method, finite element model and the fracture mechanics approach, force models of needle insertion were developed. The relationship between the needle's tip geometry and insertion force was established and used in several applications. It was shown, for example, that the cutting edge of the lancet needle can be optimally designed to minimize insertion force or bevel length. The cutting mechanics in rotary needle insertion was investigated along with the exploration of improvements of needle biopsy performance by decreasing the needle cutting and friction forces. The deflections of the needle during insertion were measured to develop a strategy for guiding the needle to the right position in brachytherapy and drug delivery. From an overall perspective, fundamental advances and application problems based on the cutting mechanics of soft tissue for needle were highlighted to lay the foundation for developments of biomedical device and improvements of healthcare procedures.     

Dynamic soft tissue deformation estimation based on energy analysi

The needle placement accuracy of millimeters is required in many needle-based surgeries. The tissue deformation, especially that occurring on the surface of organ tissue, affects the needle-targeting accuracy of both manual and robotic needle insertions. It is necessary to understand the mechanism of tissue deformation during needle insertion into soft tissue. In this paper, soft tissue surface deformation is investigated on the basis of continuum mechanics, where a geometry model is presented to quantitatively approximate the volume of tissue deformation. The energy-based method is presented to the dynamic process of needle insertion into soft tissue based on continuum mechanics, and the volume of the cone is exploited to quantitatively approximate the deformation on the surface of soft tissue. The external work is converted into potential, kinetic, dissipated, and strain energies during the dynamic rigid needle-tissue interactive process. The needle insertion experimental setup, consisting of a linear actuator, force sensor, needle, tissue container, and a light, is constructed while an image-based method for measuring the depth and radius of the soft tissue surface deformations is introduced to obtain the experimental data. The relationship between the changed volume of tissue deformation and the insertion parameters is created based on the law of conservation of energy, with the volume of tissue deformation having been obtained using image-based measurements. The experiments are performed on phantom specimens, and an energy-based analytical fitted model is presented to estimate the volume of tissue deformation. The experimental results show that the energy-based analytical fitted model can predict the volume of soft tissue deformation, and the root mean squared errors of the fitting model and experimental data are 0.61 and 0.25 at the velocities 2.50 mm/s and 5.00 mm/s. The estimating parameters of the soft tissue surface deformations are proven to be useful for compensating the needle-targeting error in the rigid needle insertion procedure, especially for percutaneous needle insertion into organs.     

Dynamic Soft Tissue Deformation Estimation Based on Energy Analysis

The needle placement accuracy of millimeters is required in many needle-based surgeries. The tissue deformation, especially that occurring on the surface of organ tissue, affects the needle-targeting accuracy of both manual and robotic needle insertions. It is necessary to understand the mechanism of tissue deformation during needle insertion into soft tissue. In this paper, soft tissue surface deformation is investigated on the basis of continuum mechanics, where a geometry model is presented to quantitatively approximate the volume of tissue deformation. The energy-based method is presented to the dynamic process of needle insertion into soft tissue based on continuum mechanics, and the volume of the cone is exploited to quantitatively approximate the deformation on the surface of soft tissue. The external work is converted into potential, kinetic, dissipated, and strain energies during the dynamic rigid needle-tissue interactive process. The needle insertion experimental setup, consisting of a linear actuator, force sensor, needle, tissue container, and a light, is constructed while an image-based method for measuring the depth and radius of the soft tissue surface deformations is introduced to obtain the experimental data. The relationship between the changed volume of tissue deformation and the insertion parameters is created based on the law of conservation of energy, with the volume of tissue deformation having been obtained using image-based measurements. The experiments are performed on phantom specimens, and an energy-based analytical fitted model is presented to estimate the volume of tissue deformation. The experimental results show that the energy-based analytical fitted model can predict the volume of soft tissue deformation, and the root mean squared errors of the fitting model and experimental data are 0.61 and0.25 at the velocities 2.50 mm/s and 5.00 mm/s. The estimating parameters of the soft tissue surface deformations are proven to be useful for compensating the needle-targeting error in the rigid needle insertion procedure, especially for percutaneous needle insertion into organs.     

Macroscopic and microscopic observations of needle insertion into gels

Needle insertion into soft tissue is one of the most common medical interventions. This study provides macroscopic and microscopic observations of needle-gel interactions. A gelatin mixture is used as a soft-tissue simulant. For the macroscopic studies, system parameters, such as insertion velocity, needle diameter, gel elasticity, needle tip shape (including bevel angle) and insertion motion profile, are varied, while the maximum insertion force and maximum needle deflection are recorded. The needle tip and gel interactions are observed using confocal microscopic images. Observations indicate that increasing the insertion velocity and needle diameter results in larger insertion forces and smaller needle deflections. Varying the needle bevel angle from 8 degrees to 82 degrees results in the insertion force increasing monotonically, while the needle deflection does not. These variations are due to the coupling between gel rupture and tip compression interactions, which are observed during microscopic studies. Increasing the gel elasticity results in larger insertion forces and needle deflections. Varying the tip shapes demonstrates that bevel-tipped needles produce the largest deflection, but insertion force does not vary among the tested tip shapes. Insertion with different motion profiles are performed. Results show that adding I Hz rotational motion during linear insertion decreases the needle deflection. Increasing the rotational motion from I Hz to 5 Hz decreases the insertion force, while the needle deflection remains the same. A high-velocity (250 mm/s and 300 mm/s) tapping during insertion yields no significant decrease in needle deflection and a slight increase in insertion force.     

最小二乘支持向量机模型的简化结构研究

针对LS-SVM模型复杂度高的问题．本文提出了一种简化模型的思路。在保证LS-SVM模型性能不变的前提下,通过该模型的少量输入输出样本,利用LS-SVM建模进一步拟合该模型。使得到的新模型的复杂度降低。仿真试验表明。本文给出的模型简化方法是有效的。     

基于sEMG信号的关节力矩NARX预测模型

为解决利用力矩传感器控制肌力训练设备所带来的滞后性,利用表面肌电信号(sEMG)超前于运动的特性,设计了基于一组拮抗肌表面肌电信号的关节力矩预测模型。首先搭建康复训练设备为信号采集和实验验证提供条件。将sEMG经过预处理,选择sEMG信号的方差特征作为神经网络输入,利用带有外部输入的非线性自回归(NARX)模型的动态循环神经网络,分别建立了基于关节力矩实际值的超前多步(MSA)预测模型和基于模型预测输出(MPO)的预测模型,通过等张和等长测试实验,比较了MSA和MPO模型的力矩预测性能。实验结果表明,两种模型输出预测值和实际值之间都有极强关联性(皮尔逊相关系数均大于0.95)。随着超前预测的步数增加,MSA模型的预测精度降低,但是超前预测的时间增大。在等张和等长测试中,当超前步数分别小于29和35时,MSA预测精度显著高于MPO(p<0.05),但MPO模型在成本和体积上更具优势。综上所述,两种模型均可以准确预测关节力矩,在实际康复训练设备控制中,可根据应用需求选择不同的力矩预测模型。     

轴向超声振动辅助磨削的磨削力建模

以单颗磨粒为对象,分析轴向超声振动下磨粒的运动特性。在此基础上,将磨削力分为切屑变形力和摩擦力两部分,分别分析了轴向超声振动对切屑变形力和摩擦力的影响。在切屑变形力方面,轴向超声振动改变了磨粒运动方向与主切削方向间的夹角;在摩擦力方面,轴向超声振动降低了磨粒与工件间的摩擦因数。基于此建立了轴向超声振动辅助磨削的磨削力模型。通过对21Ni Cr Mo5H进行了轴向超声振动辅助磨削的磨削力试验,确定了模型中的常数,并验证了所建模型的正确性。建立的磨削力模型是轴向超声振动辅助磨削的磨削力预测的一种有效方法,对轴向超声振动辅助磨削机理的认识具有较大意义。     

Design and evaluation of a novel biopsy needle with hemostatic function

Biopsy is a method commonly used for early cancer diagnosis. However, bleeding complications of widely available biopsy are risky for patients. Safer biopsy will result in a more accurate cancer diagnosis and a decrease in the risk of complications. In this article, we propose a novel biopsy needle that can reduce bleeding during biopsy procedures and achieve stable hemostasis. The proposed biopsy needle features a compact structure and can be operated easily by left and right hands. A predictive model for puncture force and tip deflection based on coupled Eulerian–Lagrangian (CEL) method is developed. Experimental results show that the biopsy needle can smoothly deliver the gelatin sponge hemostatic plug into the tissue. Although the hemostatic plug bends, the overall delivery process is stable, and the hemostatic plug retains in the tissue without being affected by the withdrawal of the needle. Further experiments indicate that the specimens are well obtained and evenly distributed in the groove of the outer needle without scattering. Our proposed design of biopsy needle possesses strong ability of hemostasis, tissue cutting, and tissue retention. The CEL model accurately predicts the peak of puncture force and produces close estimation of the insertion force at the postpuncture stage and tip position.