基于能量-峰值控制算法的建筑结构重叠分散控制研究

分散控制策略为解决大型土木结构的振动控制问题提供了新的途径。本文将重叠分散控制方法、线性矩阵不等式方法、静态输出反馈法、能量-峰值控制算法相结合,提出了重叠分散静态反馈能量-峰值控制方法。依据建筑结构的链型拓扑结构特点,建立了建筑结构地震振动响应的简化动力学模型。在求解每个重叠子系统的控制增益矩阵时,通过对状态矩阵添加一个微小的扰动矩阵,解决了在求解相应的线性矩阵不等式(LMI)时所遇到的计算困难问题。文中对美国加利福尼亚州规范(SAC)设计的20层Benchmark钢结构模型进行了数值模拟与计算。结果表明:全状态集中控制、速度反馈集中控制、重叠分散速度反馈控制、多重叠分散速度反馈控制的平均控制率分别达到了66.8%、61.8%、44.9%、31.8%,采用这些控制方法都获得了较好的控制效果;重叠分散控制策略增加了控制器设计的灵活性。

具有离散和分布时滞系统的鲁棒有限时间能量-峰值控制

研究了不确定线性时滞系统在有限时间能量-峰值性能指标约束下的鲁棒控制问题.系统的系数矩阵包含范数有界的不确定性,状态包含离散时滞和分布时滞.引入一个含参数的Lyapunov泛函,通过检验它的指数增长情况,为状态和输出定界,据此得到了闭环系统对于所有允许的不确定性有限时间有界以及满足预先指定的能量-峰值性能指标的充分条件,同时给出了状态反馈控制率.这些条件是有特征值约束的线性矩阵不等式,通过把特征值约束转换成非线性矩阵不等式,设计了锥补偿线性化(CCL)算法来求解这样的矩阵不等式,最后用数值算例验证了所得方法的有效性.

采用积分二次相位函数的雷达脉内调制类型识别

为了在低信噪比条件下获得较好的识别准确率,提出了一种基于积分二次相位函数(IQPF)的识别分类方法。根据不同信号的调频斜率的正负值完成信号的预分类,然后再利用不同信号的IQPF峰值-能量比以及功率谱估计实现信号的自动化分类。给出了识别流程图并进行了仿真实验,结果表明,该方法在信噪比大于-3dB时,其识别正确率达到90%以上,证明了方法的有效性和可行性。相关研究为雷达脉内调制类型的识别提供了一种切实可行的方法,具有一定的应用价值。

复杂场景下的改进核相关滤波跟踪算法

为解决视觉目标跟踪的遮挡、尺度变化及背景杂波问题,在核相关滤波算法基础上,引入平均峰值相关能量遮挡判据,提出一种自适应融合多特征的抗遮挡核相关滤波算法(AMFKCF)。初始化目标特征及尺度因子,将提取、融合的目标多个特征和尺度因子训练位置和尺度滤波器,得到目标的中心位置响应,根据遮挡判据,引入卡尔曼位置滤波器,对未遮和遮挡的目标中心位置进行补偿。AMFKCF算法与主流算法在CVPR 2013 Benchmark数据集中进行对比,结果表明,AMFKCF算法与主流算法相比精度提高了0.115,成功率提高了0.083,中心位置误差提高了14.67个像素,距离精度提高了9.75个百分点。能够较好地解决遮挡、尺度变化、背景杂波等问题,且兼具核相关滤波算法的优点。

一种抗遮挡的快速多尺度相关滤波跟踪算法

针对视频目标跟踪过程中,目标发生尺度变换或被遮挡问题,文中提出一种抗遮挡的快速相关滤波目标跟踪算法,该算法主要由位置滤波器、快速尺度滤波器和卡尔曼滤波器3部分组成。首先,初始化位置滤波器和快速尺度滤波器;其次在跟踪时,用平均峰值相关能量(average peak-to correlation energy, APCE)遮挡判据判断目标是否被遮挡,若无遮挡,用位置、尺度滤波器及特征估计目标位置,当目标发生遮挡时,用卡尔曼滤波器估计目标遮挡的当前位置,代替位置滤波器的估计值;最后,计算相关跟踪结果。实验结果表明,在CVPR 2013 Benchmark公开数据集上进行测试时,该算法与主流算法相比,处理遮挡和尺度变换属性的视频序列的精度提高了0.047,成功率提高了0.067,中心位置误差提高了21.128个像素,距离精度提高了0.153。该算法能够较好地解决遮挡、尺度变化问题,且具有良好的跟踪速度。

Optimizing crashworthiness design of square honeycomb structure

To provide theoretical basis for square honeycombs used as crashworthy structures, energy-absorption properties of metal square honeycombs and the size optimization were performed. Specific energy absorption(SEA) was defined as the energy absorbed by the honeycomb structure per unit volume. This parameter was often used for determining the crashworthiness of thin-walled structures. In order to find the most optimized metal square honeycomb structure with the maximum SEA and the lowest peak stress, the cell length and the foil thickness of the metal honeycombs were optimized, with a low peak stress and a high SEA set as the two primary objectives. The pre-processing software Patran was used to build FE models, and the explicit solver LS-DYNA was employed to perform the crashworthiness analyses. The results show that the square honeycomb exhibits good energy absorption performance in some cases. The geometry is effective using 16.8% less buffer structure volume than the hexagonal honeycombs with a peak stress limitation of 1.21 MPa.

Peak Load Evaluation Based on the Accumulated Annual Energy

This paper presents a nonlinear approach to estimate the consumed energy in electric power distribution feeders. The proposed method uses the statistical solution algorithm to analyze the active energy monthly consumption, which enables one to estimate the energy consumption during any period of the year. The energy readings and the normalized accumulated energy profile are used to estimate the hourly consumed active power, which can be used for future planning and sizing the equipment of the electrical system. The effectiveness of the proposed method is demonstrated by comparing the simulated results with that of real measured data.

Comparison between the time-integrated spectrum and the peak time spectrum of gamma-ray bursts and possible implications

The mono-frequency peak luminosity and the corresponding photon energy of the time-integrated （L~, Eg） and peak time （L~, ffp） vfv spectra were derived for a sample of 38 redshift-known Fermi GRBs by fitting the spectra with the Band function. It was found that Ept is generally consistent with Ep, and Lp is averagely three times larger than Lp. The slope of the Lp1 -Etp relation was consistent with that of the Lps - Eps relation. The photon indices in the peak time spectrum, particularly, the index of the low energy end was, were statistically larger than that in the time-integrated spectrum. These results indicate that Lp and Ep are dominated by Lp1 and Ept, respectively. The difference of the spectral indices between the time-integrated and peak time spectra may be because of the overlap effect of a series of time-resolved spectra within a GRB. Our simulations, which were based on the observed spectral evolution and correlation between the energy flux and the peak energy within individual GRBs support our speculations. The Lt - Et, relation may be less contaminated by the overla!a effect, and it would may be an intrinsic feature of radiation ohvsics.