基于ZigBee技术的室内定位系统算法研究及实现

随着当今社会城市化进程的快速发展,人们对于在复杂环境内的定位需求越来越迫切。由于建筑物的遮挡使得GPS定位在室内不容易实现。文中设计了一种基于ZigBee技术的室内定位系统（IPS）的解决方案。结合信号强度测距法（RSSI）与场景指纹定位法,提出RSS场景指纹定位法。该算法采用RSS“场景特征信息”作为定位场景的“指纹信息”,来建立“指纹信息”数据库,并通过TI公司的Z-stack的硬件平台实现了该算法。测试结果表明,该定位系统在覆盖超过20m＊20 m的面积里使用4个参考节点就能以2 m以内的平均定位误差实现室内定位,满足普通的定位需求。     

机械电器产品检验抽样方法探讨

根据机电产品生产特点,对稳定生产状态下的生产条件进行细分,提出不同生产条件下的生产质量验收抽样方法。     

注塑整圆内侧倒扣塑料件的分步抽芯法

针对具有整圆内侧倒扣特征的塑料件,设计了一种新型的注塑模具分步抽芯机构,并使用Pro/Mechanism软件对该机构进行了运动模拟仿真,证明该分步抽芯机构的设计是切实可行的。     

基于多尺度特征多对抗网络的雾天图像识别

目的 当前的大型数据集,例如ImageNet,以及一些主流的网络模型,如ResNet等能直接高效地应用于正常场景的分类,但在雾天场景下则会出现较大的精度损失。雾天场景复杂多样,大量标注雾天数据成本过高,在现有条件下,高效地利用大量已有场景的标注数据和网络模型完成雾天场景下的分类识别任务至关重要。方法 本文使用了一种低成本的数据增强方法,有效减小图像在像素域上的差异。基于特征多样性和特征对抗的思想,提出多尺度特征多对抗网络,通过提取数据的多尺度特征,增强特征在特征域分布的代表性,利用对抗机制,在多个特征上减少特征域上的分布差异。通过缩小像素域和特征域分布差异,进一步减小领域偏移,提升雾天场景的分类识别精度。结果 在真实的多样性雾天场景数据上,通过消融实验,使用像素域数据增强方法后,带有标签的清晰图像数据在风格上更趋向于带雾图像,总的分类精度提升了8.2%,相比其他的数据增强方法,至少提升了6.3%,同时在特征域上使用多尺度特征多对抗网络,相比其他的网络,准确率至少提升了8.0%。结论 像素域数据增强以及多尺度特征多对抗网络结合的雾天图像识别方法,综合考虑了像素域和特征域的领域分布差异,结合了多尺度的丰富特征信息,同时使用多对抗来缩小雾天数据的领域偏移,在真实多样性雾天数据集上获得了更好的图像分类识别效果。     

深度学习场景文本检测方法综述

自然场景文本检测对于机器理解场景等有着重要作用。近年来,随着深度学习的发展,自然场景文字检测方法也日新月异,取得了很好的检测效果。分析、总结了近年来基于深度学习的场景文字检测方法,将其归纳分类为基于回归、基于分割,以及两者混合三种类型,并对各类检测方法的优缺点进行了对比分析。介绍了场景文本检测性能指标及常用的公开数据集以及下载方式。对场景文字检测领域研究进行总结和展望,有望为深度学习场景文本检测方法提供新的研究方向。     

基于Z-buffer算法优化的大型变电站场景模型快速线消隐方法

针对现有三维模型消隐方法面向大规模三维场景模型应用中存在的计算复杂、耗时长等缺陷, 本文提出了基于改进 Z-buffer 算法对大型变电站场景消隐的快速可视化方法。首先,为了简化计算,将场景模 型数据整合并重构;其次,通过透视投影变换将变电场景模型像素化;进一步,基于 Z-buffer 算法高效的像素 化计算特性提出了快速模型筛选方法,从而得到变电场景的子模型遮挡关系。最后,实验中将所得遮挡关系列 表融合现有消隐算法,结果表明本文提出的方法能够大幅度提升消隐的运算性能。     

Investigation on full vehicle height control algorithm using feedback linearization method

Aiming to improve the control accuracy of the vehicle height for the air suspension system, deeply analyzing the processes of variable mass gas thermodynamics and vehicle dynamics, a nonlinear height control model of the air suspension vehicle was built. To deal with the nonlinear characteristic existing in the lifting and lowering processes, the nonlinear model of vehicle height control was linearized by using a feedback linearization method. Then, based on the linear full vehicle model, the sliding model controller was designed to achieve the control variables. Finally, the nonlinear control algorithm in the original coordinates can be achieved by the inverse transformation of coordinates. To validate the accuracy and effectiveness of the sliding mode controller, the height control processes were simulated in Matlab, i.e., the lifting and lowering processes of the air suspension vehicle were taken when vehicle was in stationary and driving at a constant speed. The simulation results show that, compared to other controllers, the designed sliding model controller based on the feedback linearization can effectively solve the "overshoot" problem, existing in the height control process, and force the vehicle height to reach the desired value, so as to greatly improve the speed and accuracy of the height control process. Besides, the sliding mode controller can well regulate the roll and pitch motions of the vehicle body, thereby improving the vehicle''s ride comfort.     

基于场景元模型的软件产品线追踪方法

针对软件产品线需求追踪方法进行了研究,基于软件产品线领域需求分析、设计、实现三个阶段的信息传播,提出了一种基于场景元模型的多层次软件产品线追踪模型,完整描述了追踪信息变化的整体框架;建立了需求—用例追踪矩阵、用例—动作追踪矩阵、动作—构件追踪矩阵;根据矩阵运算,描述了具体追踪实现过程,通过一个ATM案例验证了该方法对于软件产品线领域需求获取以及需求变更等问题的解决有较好的效果。     

Segmentation of the image with multi-visual features for a traffic scene

Scene segmentations based on the pixel classifying calculation are complicated, and they use insufficient features, thus resulting in a low accuracy, so a new model is proposed to overcome these shortcomings,which is to learn these geometric classes based on multi-visual features of super-pixels. First, various features are extracted from the super-pixels of an input image. These features are used for classifying the super-pixels. Then the difference between the adjacent super-pixels is calculated to predict their consistency. The initial classification result and the consistency are synthesized to the Markov Random Field energy function, which is then minimized based on the graph-cuts algorithm to get the final labels of the super-pixels. Experimental results prove the effectiveness of the multi-visual features and the optimization method proposed, with superior performance achieved for traffic scenes.     

Research and Application of Dynamic Equation for Full Frontal Impact

Full frontal impact theory needs researching and exploring to satisfy the primary safety design of occupant restraint system,avoiding the increasingly "engineering"trend in order to develop and design safety vehicle. After occupant restraint system is simulated by using linear elastic stiffness k,the occupant-vehicle frontal rigid barrier impact model is established. Dynamic equation of dummy chest coupling vehicle is built for full frontal impact based on ordinary vehicle deceleration by Hooke law,and the equation is solved by comparing coefficient and satisfying boundary qualifications. While relative vehicle characteristic parameters are kept unchanging,the actual vehicle deceleration is fitted to the simplified equivalent square wave( ESW),tipped equivalent square wave( TESW) and equivalent dual trapezoids wave( EDTW). Phase angle  and amplitude A of dynamic equations based on ESW,TESW and EDTW are calculated and deduced. The results show that: the dynamic equation of dummy chest coupling vehicle can be well utilized to instruct the primary safety design of full frontal impact for objective vehicle to satisfy chest deceleration demands and the equation based on TESW is best for this design.