FMT系统中的滤波器组设计

研究了新型多载波传输技术——滤波多音频(FMT)的滤波器组调制技术，通过滤波器的多相分解得到FMT系统的高效实现方法；对滤波器组中原型滤波器提出了主要性能指标和设计方法，分析了在无线信道中利用该设计方法设计出的原型滤波器应用于FMT仃系统的性能．     

基于Matlab的数字FIR滤波器设计

介绍了有限冲击响应数字滤波器的设计和利用窗函数设计FIR滤波器的方法，以及运用Matlab语言实现低通FIR滤波器的设计。     

广义共轭留数边界元法电除尘器电场解析

将广义共轭留数法引入于边界元法，改变矩阵方程与高斯消去法相结合的原传统结构，按照指数函数规律缩短解题时间。自由度数越多其效果越显著。仅就电除尘器电场和电势的小规模数值结果，与实验结果相符。证明这种方法是有效的，计算速度比传统的边界元方法快两倍。     

程控状态变量RC有源滤波器的优化综合技术

介绍了基于Butterworth传递函数的二阶状态变量RC有源滤波器的优化综合技术,进而给出了一种由单片机参与的能自动跟踪输入信号频率、选择合适滤波器截止频率的程控滤波系统实例.其特点是实现电路简洁,网络参数易确定,调试方便,截止频率解析式明确,且有较好的精度,其电路和设计方法在数据采集和工程测试等领域有较好的应用参考价值.     

An extended nonlinear state predictor for a class of nonlinear time delay systems

An extended nonlinear state predictor (ENSP) for a class of nonlinear systems with input time delay is proposed. Based on the extended Kalman filter (EKF), the ENSP first estimates the current states according to the previous estimations and estimation errors, next calculates the future state values via the system model, and then adjusts the values based on the current errors. After a state predictive algorithm for a class of linear systems is presented, it is extended to a class of nonlinear time delay systems and the detailed ENSP algorithm is further proposed. Finally, computer simulations with the nonlinear example are presented, which demonstrates that the proposed ENSP can effectively and accurately predict the future states for a class of nonlinear time-delay systems no matter whether the state variables change quickly or slowly.     

Analysis of carbon isotopes in airborne carbonate and implications for aeolian sources

Methods were developed to determine the mass ratios of carbon isotopes in trace amounts of aerosol carbonate. A Finnigan MAT 252 mass spectrometer fitted with an on-line Kiel device was to determine the ^13C/^12C ratio in CO2 produced from the carbonate. A study using these methods was conducted to characterize the carbonate carbon isotopes in aerosol samples collected in Xi‘an on dusty and normal days during March and April 2002. Results of the study demonstrate that insights into the origin of the dust can be deduced from its isotopic composition. That is, the δ^13C ofcarbonate for dust storm samples ranged from -1.4‰ to -4.2‰, and this is consistent with sandy materials in dust source regions upwind. In contrast, for non-dusty days δ^13C ranged from -7.5% to -9.3‰, which is more similar to fine particles emitted from local surface soils. Comparisons of dust storm aerosols with surface soils from source regions and with aerosol samples collected downwind indicate that the δ^13C values did not change appreciably during longrange transport. Therefore, carbon isotopes have the potential for distinguishing among source materials, and this approach provides a powerful new tool for identifying dust provenance.     

船用平台式INS光学标校初始对准方法研究

为了解决航海型平台式惯性导航系统（INS）非自主式初始对准问题，建立了不同观测量条件下统一的INS初始对准方程。通过系统可观测性分析，得出增加外部姿态角的观测信息将有效提高INS平台失准角和惯性元件误差的估计能力的结论。设计出一种独特的可测量低动态载体两维姿态的高精度光学测量系统（OCS）用以提高航海型平台式INS初始对准精度。建立OCS／GPS／计程仪（LOG）／INS初始对准卡尔曼滤波器，并进行了基于部分试验数据的数字仿真，与传统的基于位置和速度的GPS／LOG／INS初始对准方法相比，基于OCS的系统不仅缩短了对准时间，还有效改善了系统惯性元件的误差估计精度。     

多传感器信息融合ARMA信号Wiener滤波器

应用现代时间序列分析方法,基于自回归滑动平均(autoregressive moving average,ARMA)新息模型和增广的状态空间模型,在线性最小方差按标量加权最优融合规则下,对带白色和有色观测噪声的单通道ARMA信号,提出了多传感器最优分布式融合Wiener滤波器。给出了计算局部滤波器误差方差和互协方差的公式,它们可被用于计算最优加权系数。同单传感器情形相比,可提高融合滤波器的精度。一个带三传感器的跟踪系统的仿真例子说明其有效性。     

基于模糊神经网络的自适应滤波方法仿真研究

提出了一种模糊自适应卡尔曼滤波算法。该算法基于模糊规则，根据新息相关性，自适应调整测量噪声方差R，有效的解决了噪声的统计特性与实际不符时，滤波器发散的现象。同时，利用Elman网络作为误差估计器，补偿模糊自适应卡尔曼滤波器的估计误差。仿真结果表明，两种方法结合，可以有效地防止滤波器发散，缩小实际的滤波误差，提高滤波精度，实现滤波器参数的在线改进。     

中文垃圾邮件过滤系统的实现和评估

垃圾邮件是长期以来困扰电子邮件使用者的一个问题,反垃圾邮件技术除了可以抑制垃圾邮件,对反垃圾短信和垃圾VoIP电话等问题也有借鉴意义. 为此,对使用贝叶斯方法过滤垃圾邮件进行了介绍,阐述了中文垃圾邮件过滤系统的实现,并给出了评估结果. 结果表明,在过滤中计算最终概率的特征数目以及用于训练的样本个数都存在某个最优值,当用于训练的样本个数逐渐超过这个最优值时,过滤效果会略微下降并趋于一致.