基于自适应带宽核密度估计的压气机叶片加工误差统计分析方法

针对压气机叶片的加工误差统计分析建模问题,本文提出了一种自适应带宽的非参数核密度估计方法,实现了压气机叶片加工误差的概率密度建模。首先,基于拇指法则求解固定带宽核密度估计中的固定最优带宽作为起始带宽。然后,在固定最优带宽的基础上引入灵敏因子构建了自适应带宽函数,并通过自适应带宽函数对核密度估计进行调控。在此基础上,以核密度估计的精度均方差(MSE)与灵敏度S为目标函数,采用精英策略遗传算法对灵敏因子进行优化获得最优灵敏因子。进而计算出不同数据样本点对应的最优带宽,使得带宽能够根据数据样本疏密程度进行自适应调整。最后,对134组机械加工叶片的叶型截面的6种叶型误差进行统计建模,并采用交叉检验法验证了自适应带宽核密度估计的泛化性能。实验结果表明该方法的适用性好精度高,且拟合优度均在0.9以上,同时避免了传统固定带宽核密度估计的局部适应性问题。本文所提出的统计分析方法能够准确获得现有工艺能力下压气机叶片加工误差的分布特性,为叶片气动设计优化改进提供了一种有效手段。

基于自适应带宽核密度估计的载荷外推方法研究

为了快速准确地得到玉米收获机车架的载荷谱,针对载荷谱编制过程中传统的载荷外推方法的局限性,提出一种基于四叉树算法的自适应带宽核密度估计(Kernel density estimation)算法,用来进行载荷外推。将经过预处理的实测原始载荷数据进行雨流计数统计,得到载荷循环均幅值矩阵,将小于载荷循环最大幅值10%的载荷滤除,其余的载荷循环均幅值数据根据四叉树分割算法进行不同区域的分割,选择高斯核函数,根据拇指法则计算各个区域的局部最优带宽,并根据数据区域内数据点的密集程度对核密度估计的输入进行优化,减少了核密度估计的计算量,最后结合蒙特卡洛模拟算法进行载荷外推。采用玉米收获机车架实测载荷数据进行实例验证,结果表明,与传统的固定带宽、自适应带宽核密度估计的载荷外推方法相比,本文提出的方法大大提高了计算效率,其概率密度函数图与实际载荷分布更为接近;载荷循环均幅值频次分布相关系数更接近于1,均方根误差更小;载荷循环幅值累积频次曲线的决定系数均大于0.99。

基于7阶相关NRZ编码的D2D接口设计

本文提出了一种新型高带宽密度、低功耗的面向片上(Die to Die,D2D)互连的7阶相关非归零(Non-Return-to-Zero,NRZ)编码接口电路结构.为了进一步提高5阶相关NRZ编码在D2D互连中的信噪比和带宽密度,设计了基于发射矩阵和接收矩阵的编解码电路.基于发射矩阵,在发射端设计了基于电压模驱动的编码电路,有效降低了功耗;基于接收矩阵,在接收端设计了基于有源可调电感的解码均衡电路,提高了通信速率.同时,为了解决接收端时钟偏斜问题,还设计了误码校准电路.该接口电路采用28 nm CMOS(Complementary Metal Oxide Semiconductor)工艺设计,核心面积为3 mm^(2),可适用于10~50 mm的片上互连.后端仿真结果表明,在奈奎斯特频率为20 GHz、信道插损为-8 d B的条件下,接收端最窄眼宽为0.45 UI,误码率小于10^(-15),能耗效率为1.2 p J/b,带宽密度为448 Gbps/mm.

基于自适应核密度估计理论的抗差状态估计的性能分析及算例验证

针对基于自适应核密度估计的电力系统抗差状态估计新模型,提出了相应的求解策略,并通过其数学性质分析该方法对正常量测、可疑量测和不良数据的辨识性能。该算法对于正常量测具有与加权最小二乘法类似的收敛性好、渐进无偏的特点;对于不良数据,在迭代过程中通过自适应地降低其核密度带宽而等效降低其权重直至为零,相应减小或消除其影响而具有良好的抗差能力;对于可疑量测,在迭代中自适应核带宽逐渐减小,部分量测的等效权值逐步增大而被辨识为好数据,部分量测的等效权值减小而逐步被辨识为不良数据,从而实现了可疑数据的平滑逐步辨识,避免了"非好即坏"的判定而提高了对于不良数据的辨识能力。多个算例表明:新算法相较于现有其他抗差方法而言,具有较强的抗差能力,较高的计算精度及较快的计算速度。所提方法具有较强的工程应用前景。

基于自适应核密度估计理论的抗差状态估计

针对当前抗差状态估计存在的缺点,提出一种基于自适应核密度估计的电力系统抗差状态估计新算法。首先建立基于自适应核密度估计理论的电力系统状态估计新模型;进而提出该新模型核密度函数宽度的确定方法。为确定该新模型的自适应核密度带宽,基于概率密度估计的均方误差公式,结合电力系统状态各分量具有与局部量测强相关性的特点,即系统状态变量各分量与其直接量测的相关性明显强于所有非直接量测,推导新模型的状态变量分量的近似最优核带宽表达式;基于保证量测数据的冗余度、系统的可观性,及消除残差污染和残差淹没所导致的不良数据的漏辨识或正常数据的误辨识,提出新模型状态变量分量核密度带宽的约束条件,包括初始约束条件、修正约束条件及量测标准差约束条件;进而给出基于近似最优核密度带宽表达式及所有约束条件的核密度带宽综合确定方法。实际系统的应用表明:所建立的状态估计新模型,所推导的近似最优核函数带宽表达式,所提出的核密度带宽约束条件及核密度带宽综合确定方法是有效的。

基于核密度估计和分形编码算法的图像检索技术研究

为了提高基于分形压缩编码图像检索技术的应用价值,针对自然图像提出了一种联合参数的图像检索方法,从分形编码中提取鲁棒性(旋转、平移、缩放等不变性)索引,即从由值域块均值构成的解码近似图像中提取改进的Hu不变矩特征量作为检索索引,再与分形编码参数的核密度估计统计特征相结合,其中,核密度估计方法中采用可变带宽。然后采用2个索引的加权和来比较图像的相似度。实验结果表明,使用2个索引的加权和比使用单独索引具有更好的检索结果。

基于核密度估计的可变码率视频流量预测算法

为解决实时可变码率(VBR)视频在传输中难以准确预测流量的问题,提出一种变带宽核密度估计算法。该算法以核密度估计算法为基础,针对VBR视频的场景切换特性动态改变各样本点的带宽,加快算法收敛速度。仿真结果表明,与变步长归一化最小均方法相比,该算法能够减少10%左右的预测误差,保证流量预测具有较高的预测精度。

非参数逐步判别分析在脑中风分类诊断中的应用

本文采用不等带宽核密度估计的非参数逐步判别分析,使用SAS软件,对复旦大学附属中山医院78例脑中风样本病例,科学地进行了脑出血、脑缺血的分类诊断。经交叉证实法得判别正确率为:脑出血87 88%、脑缺血88 89%,总判别正确率88 46%,判别效果良好。应用78例训练样本建立的判别函数,在苏州大学附属医院2002年42例脑中风病人的分类诊断中,取得了成功的实际应用效果。

浙江近岸海域赤潮事件时空分布特征研究

利用浙江省海洋灾害公报中2006—2017年的赤潮灾害数据,分析了浙江近海赤潮的年、月和季节尺度的变化规律,同时分析了浙江近海赤潮的空间分布规律。将自适应带宽核密度分析方法应用于赤潮灾害研究,分析了浙江近海不同海域受赤潮影响的程度。研究结果表明:浙江海域赤潮高发于春末和夏季,5月是发生峰值期;从年际情况看,近年来赤潮发生的累计次数和面积呈上升趋势;浙江省赤潮主要发生在5个海域,按受赤潮影响的程度由高到低依次为舟山本岛、嵊泗列岛、温州、宁波和台州。

Low-Density Parity-Check Codes for Noncoherent UWB Communication Systems

In order to guarantee reliable data transmission, powerful channel coding techniques are usually required in noncoherent ultra-wideband(UWB) communication systems. Accordingly, several forward error correction(FEC) codes, such as Reed-Solomon and convolutional codes have been used in noncoherent UWB systems to improve the bit error rate(BER) performance. In this paper, low-density parity-check(LDPC) codes are further studied as more powerful FEC candidates for noncoherent UWB systems. Two LDPC codes and the corresponding decoding procedures are presented for noncoherent UWB systems. Moreover, performance comparison between the LDPC codes and other FEC codes are provided for three major noncoherent UWB communication systems, namely, noncoherent pulse position modulation(NC-PPM), transmitted reference(TR) and transmitted reference pulse cluster(TRPC). Both theoretical analysis and simulation results show that the two investigated LDPC codes outperform other existing FEC codes with limited penalty in terms of complexity and therefore they are promising FEC candidates for noncoherent UWB systems with low-cost and low-power consumption.

Spectrum Allocation for Cognitive Radio Networks with Non-Deterministic Bandwidth of Spectrum Hole

The spectrum allocation for cognitive radio networks(CRNs) has received considerable studies under the assumption that the bandwidth of spectrum holes is static. However, in practice, the bandwidth of spectrum holes is time-varied due to primary user/secondary user(PU/SU) activity and mobility, which result in non-determinacy. This paper studies the spectrum allocation for CRNs with non-deterministic bandwidth of spectrum holes. We present a novel probability density function(PDF) through order statistics as well as its simplified form to describe the statistical properties of spectrum holes, with which a statistical spectrum allocation model based on stochastic multiple knapsack problem(MKP) is formulated for spectrum allocation with non-deterministic bandwidth of spectrum holes. To reduce the computational complexity, we transform this stochastic programming problem into a constant MKP through exploiting the properties of cumulative distribution function(CDF), which can be solved via MTHG algorithm by using auxiliary variables. Simulation results illustrate that the proposed statistical spectrum allocation algorithm can achieve better performance compared with the existing algorithms when the bandwidth of spectrum holes is time-varied.

空间相对危险度估计方法在传染病风险评估中的应用

介绍可变带宽核密度算法、疾病空间相对危险度估计方法以及空间危险度统计学检验等在疾病风险评估中应用原理，利用2013年云南省鲁甸及周边县区感染性腹泻病网络直报数据中其他感染性腹泻病的空间相对危险度进行估计并绘制疾病图。结果显示其他感染性腹泻病高风险热点区域主要集中于研究地区的东南部大片区域内，表明基于可变带宽的核密度疾病空间风险度估计方法结合疾病制图技术，可为确定重点防控人群和区域提供直观可视化的工具。

GaN-based LEDs for light communication

Rapid improvement in the efficiency of GaN-based LEDs not only speed up its applications for general illumination, but offer the possibilities for data transmission. This review is to provide an overview of current progresses of GaN-based LEDs for light communications. The modulation bandwidth of GaN-based LEDs has been first improved by optimizing the LED epilayer structures and the modulation bandwidth of 73 MHz was achieved at the driving current density of 40 A/cm2 by changing the multi-quantum well structures. After that, in order to increase the current density tolerance, different parallel flip-chip micro-LED arrays were fabricated. With a high injected current density of ~7900 A/cm2, a maximum modulation bandwidth of ~227 MHz was obtained with optical power greater than 30 mW. Besides the increase of carrier concentrations, the radiative recombination coefficient B was also enhanced by modifying the photon surrounding environment based on some novel nanostructures such as resonant cavity, surface plasmon, and photonic crystals. The optical 3 dB modulation bandwidth of GaN-based nanostructure LEDs with Ag nanoparticles was enhanced by 2 times compared with GaN-based nanostructure LEDs without Ag nanoparticles.Our results demonstrate that using the QW-SP coupling can effectively help to enhance the carrier spontaneous emission rate and also increase the modulation bandwidth for LEDs, especially for LEDs with high intrinsic IQE. In addition, we discuss the progress of the faster color conversion stimulated by GaN-based LEDs.