Supervised dimensionality reduction via sequential semidefinite programming

Many dimensionality reduction problems end up with a trace quotient formulation. Since it is difficult to directly solve the trace quotient problem, traditionally the trace quotient cost function is replaced by an approximation such that the generalized eigenvalue decomposition can be applied. In contrast, we directly optimize the trace quotient in this work. It is reformulated as a quasi-linear semidefinite optimization problem, which can be solved globally and efficiently using standard off-the-shelf semidefinite programming solvers. Also this optimization strategy allows one to enforce additional constraints (for example, sparseness constraints) on the projection matrix. We apply this optimization framework to a novel dimensionality reduction algorithm. The performance of the proposed algorithm is demonstrated in experiments by several UCI machine learning benchmark examples, USPS handwritten digits as well as ORL and Yale face data.     

Monitoring and analysis of mining 3D time-series deformation based on multi-track SAR data

Accurate monitoring of the developing process of a surface subsidence basin is the basis of building damage assessment and surface deformation prediction. In this paper, the Synthetic Aperture Radar (SAR) data of three different imaging geometries, TerraSAR, Radarsat-2, and Sentinel-1A, were exploited. Firstly, two-dimensional (2D) time-series deformation of the surface subsidence basin caused by 15,235 working face mining was obtained based on Multidimensional Small Baseline Subset (MSBAS) technology from 19 December 2015 to 5 March 2016. By comparing vertical deformation with levelling data, it is shown that the root-mean-square error of vertical deformation is 3.2 mm and the standard deviation is 1.9 mm when the ascending-descending track SAR data is available. Otherwise, the root-mean-square error of vertical deformation is 18.1 mm and the standard deviation is 11.6 mm. Because of the low precision of the north–south horizontal movement monitored by the SAR sensor, the vertical deformation acquired by MSBAS technology and the rules of the mining subsidence (horizontal movement is proportional to tilt) were combined to obtain the north–south horizontal movement which was proven to be reliable by comparing the 2D time-series deformation obtained by MSBAS technology. Then, the deformation of the railway in the surface subsidence basin was analysed based on the three-dimensional (3D) time-series deformation. The results show that the subsidence, tilt, and horizontal movement strongly influence the railway in the monitoring period, but will not affect the normal traffic. This experiment lays a technical foundation for preventing the occurrence of mining disasters and verifies the ability to monitor the deformation of buildings and structures by interferometry synthetic aperture radar technology.     

浅析煤气热值变化对轧钢加热炉节能降耗的影响和应对措施

针对安钢第二炼轧厂炉卷机组1#加热炉在生产使用期间,因高、焦炉混合煤气的热值、压力的波动变化对1#加热炉炉造成煤气消耗偏高的原因进行了系统分析,从设备性能改进、操作方式优化等多方面进行了施实,最终该炉取得了良好的节能降耗效果。     

双丙酮丙烯酰胺的合成及提取工艺的优化

用正交实验设计法对双丙酮丙烯酰胺合成过程的工艺条件进行了优化 ,较优条件为 :n (丙酮 )∶ n (丙烯腈 )∶ n (浓硫酸 ) =2 .4∶ 1∶ 1 .1 ,反应温度 50℃ ,反应时间 4 h。对产品的提取工艺进行了改进 :将反应液在减压蒸馏前用水萃取 ,盐析 ,碱洗 ,除去大部分杂质。碱洗所用的 1 5%氢氧化钠溶液∶有机层 (质量比 ) =2∶ 1 0。产物经减压蒸馏后无需重结晶 ,收率 54.5% ,熔点 53～ 56℃。     

尿素装置工艺冷凝液电导率超标的原因及解决措施

从物料变化、仪表、设备等方面分析解吸水解系统在投产期间存在的一系列问题,针对水解工艺冷凝液电导率超标的问题制定了相应对策,并加以解决,实现了废水有效回收,达到了节能减排的目的。     

NaCl 胁迫对烤烟叶肉细胞超微结构的影响

为了探明NaCl 胁迫条件下烤烟叶片细胞超微结构的变化规律,利用透射电子显微镜对NaCl 胁迫下烤烟叶肉细胞超微结构进行了观察,并分析了烤烟应对NaCl 胁迫时细胞结构的变化。结果发现,未胁迫处理的叶绿体结构完整,形态未随处理时间的增加而发生明显改变,线粒体内外膜完整,核液浓度较高,大液泡结构清晰,细胞质与细胞壁结合紧密;盐胁迫处理后,叶绿体基粒片层松散、变形,失去清晰结构,整体结构紊乱,与自然衰老的叶肉细胞相比,未见明显的被膜破裂,线粒体膜内陷、破裂,外膜部分受损,细胞核液浓度降低,出现明显的异染色质化,核膜部分破损、界限模糊,液泡所占细胞比例变小,发生质壁分离。     

基于改进卷积神经网络的多源数字识别算法

现有的数字识别算法多是对单一类型数字进行识别,无法应对识别多源数字。针对包含手写体数字与数码管数字的字符识别场景,提出一种基于改进卷积神经网络（CNN）的多源数字识别算法。首先,使用从数显仪表生产企业现场采集的样本,结合MINIST数据集,建立起包含手写体和数码管的混合数据集;然后,考虑更好的鲁棒性,提出一种改进的CNN,并用上述混合数据集对其训练,实现了一个网络识别多类型数字;最后,训练好的神经网络模型被成功应用于RoboMaster机甲大赛的多源数字识别场景中。测试结果表明,所提算法整体识别准确率稳定且较高,具有较好的鲁棒性和泛化能力。     

替代蓄电池的超级电容储能模块设计

本文介绍了一种替代12V蓄电池的超级电容储能模块,并设计了相应的充电和稳压输出电路。通过Matlab软件对充电电流进行计算并采用L4970A和MAX668芯片进行电路设计,保证电路的效率和输出的稳定可靠。随着超级电容性能的提升,这种模块有望在部分电路代替蓄电池。     

1 000MW核电机组首台国产化半速汽轮机特点及振动控制

宁德CPR1000型压水堆核电厂1号机组(容量为1 000MW)汽轮机是我国首台自行制造的核电半速汽轮机。论述了该型汽轮机的特点;结合这些特点提出从安装角度控制汽轮机振动的措施,包括基础二次埋件及垫铁安装、轴瓦平行度调整、通流间隙调整、高中压缸负荷分配、低压缸与凝汽器连接、基础弹簧释放、轴系找中心、靠背轮铰孔连接等工序的振动控制措施。该机组实际运行情况表明,实施这些措施后效果明显,轴振最大为40.0 μm,轴瓦振动速度最大为1.42 mm/s,2项指标均满足标准要求。     

山地小口径管道的非接触式磁力检测

为了验证非接触式磁力检测技术的有效性,对山地小口径天然气长输管道进行弱磁检测,得到管道上方3个方向的磁感应强度梯度?Bx,?By,?Bz,通过定性和定量分析识别出了磁异常管段,并确定了其应力风险等级.开挖后,利用金属磁记忆检测技术和超声测厚技术对非接触式磁力检测的结果进行了验证.工程应用结果表明,在非开挖条件下,非接触...