面向金属板材数控渐进分区分阶段成形的分区算法研究

金属板材数控渐进分区分阶段成形中的一个重要环节是需要对待成形模型、成形轨迹和支撑进行分区处理。为金属板材数控渐进分区分阶段成形工艺,综合考虑挤压工具长度、可用支撑坯料厚度、板材件的深度以及挤压工具与工件之间的干涉等因素,提出了将不等距分层和中间面逼近法相结合的挤压工具与工件之间干涉检查算法和分区位置确定算法,给出了根据分区位置将待成形件模型、成形轨迹和支撑进行相应分区处理的算法,并通过算法应用实例和分区分阶段成形仿真验证了上述算法的可行性。     

基于单片机的智能型光电编码器计数器

介绍了一种对光电编码器的输出脉冲进行处理的方法,给出了基于单片机的集倍频,鉴向,计数于一体的智能型计数器原理及其硬件电路和软件的设计。     

基于小波变换与均值滤波的成形力信号处理研究

成形力是渐进成形过程的重要参数,针对渐进成形力的检测过程中成形力信号含有脉冲干扰、白噪声等混合噪声,提出将小波变换与均值滤波相结合的去噪方法。描述了基于小波变换与均值滤波相结合的算法步骤,针对成形力信号对小波阈值去噪参数选取进行了研究,并通过实验对比说明了该方法的可靠性与有效性。     

一种新的类增量学习方法

提出一种新的基于超椭球的类增量学习算法。对每一类样本,在特征空间求得一个包围该类尽可能多样本的最小超椭球,使得各类样本之间通过超椭球隔开。类增量学习过程中,只对新增类样本进行训练。分类时,通过计算待分类样本是否在超椭球内判定其所属类别。实验结果证明,该方法较超球方法提高了分类精度和分类速度。     

Thermal Buckling Analysis of SMA Fiber-Reinforced Composite Plates Using Layerwise Model

Thermal buckling analysis of laminated smart composite plates subjected to uniform temperature distribution has been presented. Shape memory alloy (SMA) fibers whose material properties depend on temperature have been used as a smart material. A three-dimensional layerwise plate model has been employed in developing the system equations using variational approach. Finite-element method has been adopted for discretization of the laminate. Lagrangian interpolation functions have been used to approximate the displacement components along the thickness as well as in the in-plane direction. The actual variation of prebuckling stresses has been accounted for in the derivation of the geometric stiffness matrix of the laminates. An incremental load technique has been used in the analysis to take into account the nonlinearity in the material properties of the SMA arising due to their temperature dependence. The effects of thickness ratio, orthotropic ratio, fiber orientation, aspect ratio, stacking sequence and various boundary conditions on the critical buckling temperature have been examined in details. The results have been validated with those available in the literature.     

Seismic Safety Assessment of Gated Spillways

This paper presents an innovative three-dimensional performance-based earthquake engineering assessment methodology for gated spillways. Several functionality and structural stability limit states are defined. The functionality of the gates is tackled by assessing seismic demand-to-capacity indices along a success path leading to postearthquake gate opening. As an application example, the proposed methodology is applied to an existing gated spillway built in 1927. Finite element incremental dynamic analyses are performed for six sets of spectrum compatible ground motions with return periods ranging from 200?to?10,000?years. Linear transient dynamic analyses are first performed to compute internal force resultants, displacements, and distortions in structural components. For the concrete structures, the forces from the analyses are input in a postprocessor to compute sliding safety factors using a three-dimensional extension of the gravity method accounting for concrete cracking. The assessment of the hoist supporting structure and gates is done using current steel member demand-to-capacity checks. The actual failure mode hierarchy is identified and found unsatisfactory as the concrete structures and steel lifting tower anchors are expected to fail prematurely. Specific performance criteria and related ground motion return periods are proposed (1) to maintain normal gate opening; (2) to maintain the load bearing capacity of the steel towers and bridge; (3) for life safety of personnel; and (4) to avoid an uncontrolled release of the reservoir.     

基于滑动窗口的快速Learn++.NSE*

Learn++.NSE集成的单个基分类器需根据其在所有历经环境中的分类错误率加权计算投票权重,学习效率有待提高.因此,文中采用滑动窗口技术优化权重的计算过程,提出基于滑动窗口的快速Learn++.NSE算法(SW-Learn++.NSE).该算法仅考虑使用单个基分类器近期窗口内的分类准确率计算投票权重,提高集成学习的效率.实验表明,相比Learn++.NSE,在取得同等分类准确率的情况下,文中算法分类学习的效率更高.     

数控机床平动轴误差辨识及补偿研究

为了大幅提升数控机床平动轴的运动精度,从而满足当代数控系统对其高精度的要求,针对机床平动轴的空间几何误差开展了深入研究,提出了可以有效辨识平动轴空间几何误差的便捷方法,并基于齐次坐标变换原理建立了平动轴空间几何误差的辨识模型;针对机床平动轴定位误差的特性,提出了可以实现有效补偿的增量式误差补偿原理,并建立了相应的误差补偿模型。经验证,所提出的平动轴误差辨识法和增量式误差补偿原理不仅理论正确,而且可以大幅地提高机床平动轴的定位精度。     

预测增量动态逆无人机姿态控制

针对小型无人机模型参数精度不高的问题,提出一种预测增量动态逆姿态控制方法。该方法以角加速度作为姿态控制的反馈,从而降低了控制系统对模型参数的敏感性。根据无人机力矩方程,先将角加速度计算式写成增量形式;再根据非线性动态逆方法构建出角加速度为隐式输入、舵面偏转量增量为输出的直接关系式;在控制器中引入预测滤波器,对角加速度进行滤波和预测,提高反馈信号的精度和实时性。飞行实验表明方法是有效性的。     

基于自组织增量神经网络的码书产生方法在图像分类中的应用

针对基于码书模型的图像分类方法忽略图像的拓扑信息及增量学习导致分类精度有限的问题,提出了基于自组织增量神经网络(SOINN)的码书产生方法。首先回顾了常见的码书编码方式;其次改进了基本的码书模型,利用SOINN自动产生聚类数目和保留数据拓扑结构的两项能力,寻找更有效的单词和设计更有效的编码方式,产生更合适的码书。实验结果显示在不同样本数和不同规模码书下分类精确度相对同类算法有最高将近1%的提升。该结果表明基于SOINN的码书产生方法显著提高了图像分类算法的精度,该方法还可以更高效、更准确地运用于各种图像分类任务。