复数FIR DF设计的神经网络优化方法

本文基于人工神经网络（ＡＮＮ）能量函数优化理论，提出了一种ＦＩＲ数字滤波器（ＤＦ）神经网络优化设计（ＮＮＯ）方法的理论框架。该理论将实数与复数ＦＩＲ ＤＦ设计工作统一起来。表征设计质量的加权均方误差被当作ＡＮＮ能量函数，以此导出ＦＩＲ－ＮＮＯ的Ｌｙａｐｕｎｏｖ方程，文中说明了算法实现的基本原则，并给出了两个实数线位和一个复数非线性相位ＦＩＲ ＤＦ设计实例。通过与其它几种方法的比较证明了该方法的有效     

一种基于复数相位屏的强激光大气传输仿真方案

针对传统实数相位屏法中只能表征湍流扰动而不能综合表征强激光在大气传输时导致的能量衰减、湍流扰动和非线性热晕效应等困难,提出一种基于复数相位屏表征的强激光大气传输仿真方案.该方案采用复数相位屏的实部和虚部分别表征衰减和湍流、热晕作用,其中基于能见度特征综合描述大气吸收和散射的衰减作用(复数相位屏的虚部),基于流体力学的密...     

基于Hanning Window设计FIR数字高通滤波器

基于实序列的傅里叶变换具有的特征,提出了一种研究线性相位FIR数字滤波器幅度函数的简洁方法。首先揭示了偶对称和奇对称的线性相位FIR数字滤波器幅度函数的特点;然后研究了矩形窗函数对线性相位理想数字低通滤波器幅频特性的影响,即时域加矩形窗,频域形成过渡带;最后介绍了利用窗函数法设计FIR数字滤波器时,对窗函数的要求,选择窗函数的依据及利用窗函数法设计FIR数字滤波器的步骤,并给出了基于Hanning Window来设计FIR数字高通滤波器的实例。     

线性相位FIR滤波器频域特性的教学探索

线性相位FIR滤波器的频域特性分析是"数字信号处理"课程中的一个重要内容。本文给出了一种与传统教材不同的线性相位FIR滤波器频率响应的分析方法。笔者以Ⅰ型线性相位系统的频率响应为基础,证明了Ⅱ型、Ⅲ型和Ⅳ型线性相位FIR滤波器的频率响应可表示为Ⅰ型线性相位系统的频率响应与简单函数的乘积。该方法特点是分析过程简捷,突出了不同类型线性相位FIR滤波器频域特性,便于学生掌握线性相位FIR滤波器频域特性。     

一维线性相位FIR数字滤波器的频域最小平方误差设计方法

本文提出了一种设计一维线性相位FIR数字滤波器的新方法。该方法采用频域的最小平方误差函数使所求的增益特性逼近所希望的增益特性 ,其计算公式非常简单 ,并可设计任意形状增益特性的FIR数字滤波器。设计实例表明提出的方法非常有效     

连续相位板优化设计

提出一种适用于连续相位板设计的改进Gerchberg-Saxton(GS)算法,该算法应用相位展开相关原理,结合滤波处理使GS算法设计的近场相位连续分布,同时提出在GS算法的远场迭代中采用经过修正的振幅函数,可使远场光斑的包络更接近目标函数。模拟实验表明,该方法解决了现有连续相位板设计中相位连续性与靶面光强控制效果难以同时优化的矛盾,可满足惯性约速聚变(ICF)对束匀滑元件相位连续性、光斑包络及能量集中度的要求。     

基于神经网络的4型FIR滤波器的优化设计

详细讨论了4型线性相位滤波器的幅频特性与正弦基函数神经网络算法的关系,分析了神经网络系统的稳定条件,给出了FIR滤波器优化设计实例。根据4型FIR滤波器的幅频响应特性,构造出一个相应的神经网络模型,并建立了FIR线性相位数字滤波器的神经网络算法。该算法通过训练神经网络权值,使设计的数字滤波器与希望得到的FIR线性相位滤波器的幅频响应之间的误差平方和最小化,从而获得FIR线性相位数字滤波器的脉冲响应。计算机仿真表明了该算法的有效性和优异性能。     

FIR数字滤波器设计的复数神经网络方法

ＦＩＲ滤波器的设计是数字信号处理中的一个重要问题,神经网络则在优化设计方面显示了其良好的应用前景,但目前大多数应用均限于实数神经网络的情形．本文针对ＦＩＲ数字滤波器的设计问题,构造了具有复数状态、输出、联接权和偏置输入的Ｈｏｐｆｉａｌｄ网络,给出了网络的计算能量函数,并证明了算法的收敛性．通过近似线性相位、非共轭对称频响ＦＩＲ滤波器的设计实例表明,本文的工作对推广复数神经网络的应用与寻求更为复杂的ＦＩＲ滤波器设计的新方法均作了有效的尝试．     

可变带宽FIR低通滤波器设计和RAG算法系数综合

提出使用简化加法器图算法综合可变带宽FIR数字滤波器。首先使用谱参数的方法建立可变带宽、线性相位的FIR低通数字滤波器的系统函数,通过使用加权最小均方的方法,得到了滤波器系数的最优表达式。然后基于可变滤波器结构为定系数FIR子滤波器线性组合的特点,提出使用筒化加法器图算法综合其硬件结构。该算法生成一种能最大程度地利用系数之间共享特性的加法器流图,使用较少的加法器个数和加法次数实现系数相乘。最后设计实例证明了可变带宽的有效性和该算法的高效性。     

基于MATLAB与CCS的FIR滤波器的C语言实现

为保证所设计的FIR数字低通滤波器具有严格的线性相位,在对几种FIR基本结构的比较之后,采用了线性相位FIR滤波器的直接型结构。使用Matlab内置函数计算出滤波器的系数和检验滤波器的频率响应特性。采用C语言实现数字滤波器的设计,并在集成开发环境代码调式器(Code Composer Studio,CCS)上进行仿真,仿真结果表明,所设计的数字低通滤波器能够满足系统实时性和不失真要求。     

A realization of a GaAs FET microwave active filter

A methodology for microwave active filter design has been presented by this author (see ibid., vol.37, no.9, p.1418-24, 1989). The present work serves to verify aspects of this design approach. A practical implementation of a network discussed in that work is given. First, an ideal preliminary design for a second-order bandpass filter is obtained. Next, the nonideal effects of the active devices and the lossy reactive elements are introduced, and then optimization is used to obtain the final design. Finally, the filter is constructed and tested. The measured response is found to agree quite closely with the design specifications. The filter section is constructed in hybrid form with lumped components, and uses two NEC 900100 GaAs FETs. The prototype accurately realizes a second-degree bandpass response with a center frequency of 6 GHz and a pole q of 2.5     

ARMA谱估计的高阶关联神经网络优化方法

本文基于全反馈高阶关联神经网络优化理论，提出了一种将神经网络优化方法应用于ARMA谱估计（ARMA－NNO法）的理论框架。该方法与迄今为止所见方法的区别在于，它直接面对ARMA扩展的Yule—Walker方程的非线性，同时估计出模型的AR和MA两部分参数。描述估计质量的加权均方误差被当作神经网络能量函数，从而导出了ARMA－NNO法的Lyapunov方程。文中讨论了此法的实现方案，给出了几个谱估计实例，通过与其它几种ARMA谱估计方法的比较，证明了它的有效性。     

Synthesis of Combline and Capacitively Loaded Interdigital Bandpass Filters of Arbitrary Bandwidth

Synthesis procedures are presented for commensurate linelength combline and capacitively loaded interdigital filters. A simple modification for using lumped capacitors is given and is found to yield excellent results for any practical bandwidth. The synthesis procedures for both filter types follow directly from the classical bandpass filter design approach using a transformed frequency variable. Two design examples are given to illustrate the design procedures and experimental results are presented for a 40 percent bandwidth combline filter. The problems of network sensitivity, use of network equivalents, and extension to other filter types are also discussed.     

Design of Linear Phase Selective Comb-Line Filter

A multiwire approach has been used to develop design equations for linear phase selective comb-line filter. The filter under consideration consists of two rows of inductive resonators separated by a slotted coupling surface. Through the development process a multipath prototype network has been considered. A frequency transformation has been formulated to relate the multipath prototype network and the multiwire comb-line structure. Hence, once the element-values of the prototype network are chosen to meet prescribed requirements, the corresponding comb-line filter element values can be computed through the developed explicit formulas. To illustrate the design procedure a brief design example is presented.     

A design method for substrate integrated waveguide electromagnetic bandgap (SIW-EBG) filters

An efficient design method for substrate integrated waveguide electromagnetic bandgap (SIW-EBG) filters is proposed which provides direct dimensional synthesis approach for desired filter objectives without using network representations. The method is applied to the design of an X band SIW-EBG filter and its response is compared with HFSS (high frequency structure simulator) simulations for validation purposes. Fairly good agreement between the results shows the applicability of the proposed method for SIW-EBG filter design.     

An Analytic Design Method for Microstrip Tunable Filters

This paper describes an analytic design procedure for microstrip tunable filters. Step-impedance resonators are employed and loaded with varactors for achieving agility in the filter response. Fixed lumped capacitors are utilized as admittance inverters in order to minimize the number of varactors in the filter. An analytic approach for filter design makes it possible to achieve a tunable filter showing the same frequency response when the center frequency is adjusted. A two-pole microstrip tunable filter whose center frequency can be adjusted from 1.1 to 1.5 GHz is designed to demonstrate the validity of design theory. A prototype tunable filter operating from 2.1 to 2.7 GHz is also designed and measured. A good agreement between the measured and simulated results is shown. Finally, three- and four-pole tunable filters are designed to show straightforward application of the presented design method to higher order tunable filter design.      

Design of recursive polyphase networks with optimum magnitude and minimum phase

A novel approach to the design of recursive polyphase networks with application to multirate filtering and digital filter banks is proposed. To this end, Pelloni's method [4] is extended such that all zeros of the polyphase network are located on the z-plane unit circle and all poles are single, corresponding to an optimum magnitude response with minimum phase. An example, taken from transmultiplexing applications, shows that the proposed method results in substantially improved filter designs compared with the approaches of Bellanger et al. [1], Vary [6] and Pelloni [4].     

The Modern Network Theory Approach to Microwave Filter Design

A review of the modern network theory approach to microwave filter design based on Richards' transformation is given. Basic concepts are discussed, including the use of Kuroda's identities and the specialized theory of optimum filter forms. Applications of the theory presented to elliptic-function filters, special rejectiontype filters, and multiplexers are given. Throughout the presentation, emphasis is placed on basic methods and techniques rather than on detailed design procedures.     

General Theory and Design of Optimum Quarter-Wave TEM Filters

The exact synthesis and design of a broad class of quarter-wave TEM mode filters is presented in a three-step procedure. The theory is applicable to all microwave filter forms consisting entirely of a cascade of quarter-wave lines, quarter-wave stubs, and coupled quarter-wave lines. The exact design of many conventional filter forms that heretofore could only be designed using approximate methods is possible using the techniques and functions described. General approximating functions that give the "best" transmission response in a Butterworth or Chebyshev sense are derived. The use of a maximum number of available singularities to augment the filter skirt response leads to a minimum element filter termed "optimum multiple." An optimum multipole design insures the "best" possible response for the minimum element and all corresponding redundant element networks. The designer has the freedom of introducing sufficient redundancy to obtain a design that is practical to construct, but which still realizes the optimum response. Judicious choice of network form often results in improved performance and reduced size in comparison with many conventional filter forms. Design procedures are presented that allow the practical realization of distributed filter networks in the form of quarter-wave lines. A design example and experimental results are given to confirm the theory.