Gradient Boundary Detection for Time Series Snapshot Construction in Sensor Networks

In many applications of sensor networks, the sink needs to keep track of the history of sensed data of a monitored region for scientific analysis or supporting historical queries. We call these historical data a time series of value distributions or snapshots. Obviously, to build the time series snapshots by requiring all of the sensors to transmit their data to the sink periodically is not energy efficient. In this paper, we introduce the idea of gradient boundary and propose the gradient boundary detection (GBD) algorithm to construct these time series snapshots of a monitored region. In GBD, a monitored region is partitioned into a set of subregions and all sensed data in one subregion are within a predefined value range, namely, the gradient interval. Sensors located on the boundaries of the subregions are required to transmit the data to the sink and, then, the sink recovers all subregions to construct snapshots of the monitored area. In this process, only the boundary sensors transmit their data and, therefore, energy consumption is greatly reduced. The simulation results show that GBD is able to build snapshots with a comparable accuracy and has up to 40 percent energy savings compared with the existing approaches for large gradient intervals.     

Globally valid adaptive controllers of mechanical systems

The adaptive controller design of a large class of mechanical systems is addressed in this paper. The configuration space for the attitude of a vehicle is modelled as SO(3), namely the rotation matrix group. To avoid singularity problems, some studies adopt the representation of unit quaternions for modeling system dynamics and for the design of controllers. However, the complicated formulation of quatrains makes the application of this method limited to simple mechanical systems. Here, the strategy is to carry out the modeling and analysis in terms of the global representations of jet bundles of SO(3), while the unit quaternion is chosen to implement the controller for its least number of parameters (four) to represent SO(3) globally. Based on the transformation between the rotation matrix and the quaternion, the method explored in this paper gives rise to an adaptive controller in a clean and efficient way     

橡胶挤出机平均温差分析与计算

从理论上推导挤出机平均温差的计算公式，并指出与一般换热器平均温差计算的异同。     

蛋白质与原卟啉Ⅸ及复合物发光机理探讨

对原卟啉Ⅸ（ＰｒｏｔｏｐｏｒｐｈｙｒｉｎⅨ）与蛋白质结合物的吸收、荧光的激发、发射光谱等的研究，表明了在光作用下，其可见区发光（λｍａｘ６３５ｎｍ左右）与卟啉有关，而紫外区发光（λｍａｘ４２２ｎｍ左右）与蛋白质的色氨酸残基有关，其光谱特性与临床诊断选择的癌固有荧光特征峰基本相符。该研究结果为探讨癌的发光机理积累了有益资料．     

方程根数对反应系数法求解负荷的影响

房间围护结构热力系统传递矩阵元素Ｂ（Ｓ）的根数多少或根值大小，直接影响反应系数的精度。其值取多少，国内外没有统一标准。本文结合空调房间的下部送风，针对不同结构的房间进行探讨，发现最小根值取到（—３０）时，即足能满足一般精度要求。     

MCS接枝共聚物的结构和形态

将甲基丙烯酸甲酯（ＭＭＡ）、苯乙烯（Ｓ）在氯化聚乙烯（ＣＰＥ）存在下进行悬浮接枝共聚，获得ＭＣＳ树脂。本文着重考察了ＭＣＳ树脂的化学结构及胶粒形态。结果表明，确实发生了接枝共聚。ＣＰＥ用量、ＲＳＨ含量、引发剂浓度、反应温度、溶胀时间及聚合转化率对ＭＣＳ树脂的接枝率、接枝效率均有影响。ＭＣＳ树脂是接枝共聚物、ＭＭＡ／Ｓ共聚物与ＭＭＡ均聚物的混合物，橡胶相ＣＰＥ在ＭＣＳ树脂中的分散情况为“海岛结构”，     

宽带钢冷轧机颤振控制的数值模拟

本文以较为准确的轧机振动状态下的轧制压力模型为基础，对宽带钢冷轧机干摩擦阻尼减振进行了数值模拟的研究。研究结果与实际观测值相等。     

图像复原中的模糊参数估计

图像复原技术在航空拍摄和机器视觉中是提高图像质量的重要手段.针对有相对运动情况的图像存在的降质模糊,分析了运动模糊的数学原理,针对不同运动模糊采取了不同去模糊方式.即先判断图像存在的运动模糊方式,针对直线运动通过Hough变换法和误差参数法相结合的方法估计降质参数,而对于旋转运动则采用曲线拟合和极坐标转换相结合的方法来估计降质参数.通过维纳滤波方法复原图像,提高图像质量.利用提出的自适应方法对直线运动模糊(参数为(30,70°))和旋转运动模糊(参数为(128,128),20°)分别作了实验计算.对比实验表明,这种由粗及精的方法能准确估计模糊参数,与传统处理方法相比,更加便捷有效.     

改进BP网络在航材需求预测中的应用

针对航材备件需求预测问题,在对影响航材备件需求量的多个因素进行分析研究的基础上,运用改进BP神经网络算法进行预测的仿真实验。实验结果表明,改进BP神经网络能够对积累的历史数据进行充分的应用,并且有较高的预测准确性。