自主虚拟人智能驾驶行为模型的研究和实现

描述了自主虚拟人的分层驾驶行为及对虚拟世界的感知.在对自主虚拟人的驾驶行为进行研究的基础上,结合导向行为与专家经验,建立了基于产生式规则的智能行为模型,并给出了规则的具体实现.通过虚拟现实技术对该智能模型进行了验证,证明了其可行性.      

新一代GNSS接收机的智能跟踪状态评估技术

新一代全球导航卫星系统的接收机需要根据用户所处环境,自适应地调整接收机内部参数以达到最稳健的导航效果。这就要求接收机在运行过程中通过评估跟踪状态去检测和分析用户场景。基于对传统接收机控制系统缺陷的分析,提出了智能接收机的概念、特点及系统框架;在此基础上设计、分析接收机的智能跟踪状态评估算法;最后通过仿真验证了所设计的算法可以有效地识别场景及场景间的切换。     

平面工艺功率型半导体晶体管铝迁移实验

在平面工艺制造的功率型半导体晶体管芯片表面,发现高强度电场条件下存在跨越0.05 mm半绝缘硅带的铝迁移现象,对此开展了一系列实验研究. 通过开帽物理观察与分析、低温测试与烘干验证、迁移物质分析对比,确定了迁移现象中的迁移物质,同时分别取证环境温度、湿度、芯片表面污染物以及器件局部结构与此迁移现象的关联性,以此提出了迁移机理假说,为在不同的诱发条件下对此迁移机理进行确认和量化研究提供指导.      

多阶段任务系统通用可靠性仿真模型

归纳了当前多阶段任务系统(PMS,Phased Mission Systems)可靠性建模的基本假设,指出了解析模型描述和求解能力的不足以及仿真模型存在的难以转换的缺点.应用着色Petri网(CPN,Colored Petri Net) Tools建立PMS的多层通用仿真模型:顶层控制模型将PMS各阶段的故障树参数化,作为CPN通用仿真模型的输入;单阶段处理模型根据顶层控制模型的输入判别该阶段顶事件是否发生;底事件处理模型通过比较随机数和故障发生概率的大小产生故障.通用仿真模型可以保持模型结构的稳定,同时又简化了仿真模型的生成.案例对比了CPN仿真模型与二元决策图(BDD,Binary Decision Diagrams)算法的可靠性计算结果,证明了模型的正确性;具有动态变化参数的PMS仿真则说明了模型的适用性.      

A method of estimating GPS instrumental biases with a convolution algorithm

This paper presents a method of deriving the instrumental differential code biases (DCBs) of GPS satellites and dual frequency receivers. Considering that the total electron content (TEC) varies smoothly over a small area, one ionospheric pierce point (IPP) and four more nearby IPPs were selected to build an equation with a convolution algorithm. In addition, unknown DCB parameters were arranged into a set of equations with GPS observations in a day unit by assuming that DCBs do not vary within a day. Then, the DCBs of satellites and receivers were determined by solving the equation set with the least-squares fitting technique. The performance of this method is examined by applying it to 361?days in 2014 using the observation data from 1311 GPS Earth Observation Network (GEONET) receivers. The result was crosswise-compared with the DCB estimated by the mesh method and the IONEX products from the Center for Orbit Determination in Europe (CODE). The DCB values derived by this method agree with those of the mesh method and the CODE products, with biases of 0.091?ns and 0.321?ns, respectively. The convolution method's accuracy and stability were quite good and showed improvements over the mesh method.     

Detection of aerosol pollution sources during sandstorms in Northwestern China using remote sensed and model simulated data

The present paper has used a comprehensive approach to study atmosphere pollution sources including the study of vertical distribution characteristics, the epicenters of occurrence and transport of atmospheric aerosol in North-West China under intensive dust storm registered in all cities of the region in April 2014. To achieve this goal, the remote sensing data using Moderate Resolution Imaging Spectroradiometer satellite (MODIS) as well as model-simulated data, were used, which facilitate tracking the sources, routes, and spatial extent of dust storms. The results of the study have shown strong territory pollution with aerosol during sandstorm. According to ground-based air quality monitoring stations data, concentrations of PM₁₀ and PM_2.5 exceeded 400?μg/m³ and 150?μg/m³, respectively, the ratio PM_2.5/PM₁₀ being within the range of 0.123–0.661. According to MODIS/Terra Collection 6 Level-2 aerosol products data and the Deep Blue algorithm data, the aerosol optical depth (AOD) at 550?nm in the pollution epicenter was within 0.75–1. The vertical distribution of aerosols indicates that the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) 532?nm total attenuates backscatter coefficient ranges from 0.01 to 0.0001?km^?1?×?sr^?1 with the distribution of the main types of aerosols in the troposphere of the region within 0–12.5?km, where the most severe aerosol contamination is observed in the lower troposphere (at 3–6?km). According to satellite sounding and model-simulated data, the sources of pollution are the deserted regions of Northern and Northwestern China.     

基于S型测试工作量函数的软件可靠性增长模型

软件测试工作量随时间的变化情况对于软件可靠性增长曲线的形状具有显著影响,因此将测试工作量函数(TEF,Testing Effort Function)引入软件可靠性增长模型(SRGM,Software Reliability Growth Model)中以提高软件可靠性评估性能.分析了软件结构特征及学习因素对实际测试过程的综合影响,研究增长速率为先增后减的S型增长趋势的TEF,提出两种S型TEF,即延迟S型TEF与变形S型TEF,并提出考虑延迟S型TEF的SRGM(DSTEF-SRGM)以及变形S型TEF的SRGM(ISTEF-SRGM).在两组真实失效数据集上,进行了这两种SRGM与经典SRGM及其它考虑TEF的SRGM的对比研究.结果表明,ISTEF-SRGM的模型拟合效果最优,同时验证了该模型具有优越的软件可靠性评估性能及模型适应性.      

基于静态输出反馈的大型飞机横航向控制综合

由于广泛采用轻质材料,大型飞机的结构柔性使得控制系统和结构动力学之间存在不可避免的耦合现象,控制器设计必须同时考虑结构柔性对系统反馈的贡献,并满足飞行品质、鲁棒性等目标.提出了一种利用刚体飞行动力学模型和包含弹性模态的高阶模型设计满足多种目标的控制器综合方法,首先由刚体模型设计静态输出反馈参数,满足动态响应、飞行品质和抑制阵风扰动等要求,在此基础上利用已获得的静态反馈参数针对高阶模型设计H_∞回路成形控制器,实现对结构模态的抑制,并满足系统的鲁棒性需求.仿真结果表明:设计方法能够充分利用包含弹性模态的大型飞机飞行动力学模型的特点,满足结构模态抑制和鲁棒稳定裕度等多种需求.      

基于神经网络的维修保障流程影响度分析方法

针对维修保障能力的主要影响因素分析及其影响度的量化问题,以维修保障业务流程作为分析对象,通过建立业务流程参数与装备使用能力的关系函数,建立影响因素和影响度的分析方法,并给出影响程度、关键要素、关键影响要素以及综合影响度定义.以流程数量、效率和稳定性作为分析指标,利用MATLAB软件基于神经网络实现对关系函数的逼近求解,为装备制造企业提供确定关键流程和主要影响因素的系统实现.以某航空装备制造企业维修保障业务流程为例给出具体应用,与专家评价方法相比该方法可获得更为客观的分析结果.