【正文】 xpansion ponents, the resulting stability constraint bees a linear equality constraint. However, through a large number of simulations, this constraint could be found to be invalid in some situations. As an attempt to solve this problem, a new stability constraint is to be proposed, which is also based on AP. Unlike that in ,the stability constraint is approximated in a quadratic form, which is then bined with theaterative design will be shown that this stability constraint is imposed on an integral over the whole frequency band. Hence, neither a large number of constraints on densegrid frequency points nor multiple exchange algorithms are needed.In practice, two optimal criteria, namely, minimax, and(weighted)least squares , ,are most frequently used in IIR filter other criteria, such as equiripple pass band and peakconstrained least squares stop band ,leastpower error minimization , and weighted integral of the squared error(WISE) ,are also applied in IIR digital filter design. In this paper, we will focus on IIR filter design in the weighted least squares(WLS)sense. Convex optimization methods, such as secondorder cone programming(SOCP) , and SDP , have been widely applied in designing FIR and IIR digital filters[1],[7],[9].Compared with LP and convex QP problems, SOCP problems can incorporate more general constraints besides linear equalities and inequalities. As a matter of fact, LP and convex QP problems are special cases of SOCP problems. Therefore, all the LP and convex QP problems can also be solved by the SOCP algorithms. Although SOCP problems are less general than SDP problems, the putation plexity per iteration required in the SOCP algorithms is much less than that in the SDP algorithms, particularly when the dimensions of secondorder cone(SOC)constraints are large. Therefore, we will cast the IIR filter design problem into the SOCP form. In [26],we have presented a preliminary version of this paper. In order to incorporate the quadratic peak constraints, the design method shall be implemented in this paper as an iterative SOCP procedure instead of the iterative QP procedure adopted in . Accordingly, the quadratic cost function in is cast into an SOC constraint. Since the stability constraint derived from AP cannot be directly incorporated in the design procedure, an approximated stability constraint is further proposed in this paper. Two new examples and further discussion are included in this paper.This paper is organized as follows. In Section II, an iterative SOCP method without any stability constraint is introduced to design IIR digital filters. Then, peak constraints are incorporated as SOC constraints. In Section III, a new stability constraint based on AP is developed. Design examples are presented in Section IV. Finally, conclusions are drawn in Section V.In this paper, an IIR digital filter design method with a new Pbased stability constraint has been presented. As ared with the previous design methods using the reweighting echnique, the design problem is formulated as an iterative OCP problem, which can handle both linear and SOC contraints. Therefore, some more plicated constraints, such Squadratic peak error constraints, can be further in corporated. another advantage of the design method is that the stability constraintisde duced from AP, which is both sufficient and necessary for stability. Finally, the proposed stability constraint is impose don the whole frequency band, which greatly facilitates the design progress. In order to incorporate it into the design procedure, the similar reweighting technique is employed, and the stability constraint is then bined with the iterative procedure. If the iterative procedure converges and parameters and are appropriately selected, the stability constraint can guarantee the stability of the designed filter. When the maximum pole radius is given, by adjusting the parameter , we can also control poles’ locations. The robustness and effectiveness often proposed approach have been demonstrated in this paper by four numerical examples. IIR數(shù)字濾波器設(shè)計(jì)的新的穩(wěn)定基于約束原理論證摘要：本文提出了一種加權(quán)最小二乘（英國威爾士）方法IIR數(shù)字濾波器的設(shè)計(jì)采用了新的穩(wěn)定的制約因素。本算法用了權(quán)值改造技術(shù)，迭代二階錐編程（ SOCP ）方法是用來解決設(shè)計(jì)問題，例如，一個(gè)新的穩(wěn)定的制約因素與預(yù)抄極半徑來自論點(diǎn)原則（美聯(lián)社），類似的技術(shù)部署權(quán)值改造技術(shù)概略穩(wěn)定約束的二次型。索引詞參數(shù)原則（美聯(lián)社） ，無限脈沖重響應(yīng)（ IIR號(hào)）數(shù)字濾波器， 權(quán)值改造技術(shù)，二階錐規(guī)劃（ SOCP ） ，加權(quán)最小二乘法（英國威爾士）逼近。1 簡介相比與FIR數(shù)字濾波器的設(shè)計(jì)，即一個(gè)飛行情報(bào)區(qū)挖意大利過濾器的過濾器規(guī)格滿足的目的是第一，和然后，近似程序可以保證穩(wěn)定的設(shè)計(jì)IIR數(shù)字濾波器， ， ，許多其他算法已建議設(shè)計(jì)IIR數(shù)字濾波器在一個(gè)直接的方式，一些迭代方法。10月:2007 25 。 5月修訂01,2008 。首次出版10月31,2008 。 3月出版的最新版本11,2009 。作者與新聞部的電氣與計(jì)算機(jī)工程工程，溫莎大學(xué)，溫莎，在N9B 3P4 ，加拿大，到目前為止，收斂這些方法不能得到嚴(yán)格保證，我們采用相同的迭代程序使用Steiglitz 麥克布賴德計(jì)劃，設(shè)計(jì)IIR數(shù)字濾波器。穩(wěn)定是一個(gè)重要的問題的IIR數(shù)字濾波器的設(shè)計(jì)。最近，一些積極的，實(shí)為基礎(chǔ)的，實(shí)為基礎(chǔ)的[ 13 ]站，性約束表示為線性不等式與尊重以分母系數(shù)，它可以很容易地納入線性規(guī)劃（唱片）和二次規(guī)劃（卡塔爾）問題. Rouch233。定理為基礎(chǔ)的穩(wěn)定的制約因素，這是性低于正實(shí)的站，性約束，，，它是用來識(shí)別一套積極系統(tǒng)規(guī)定參數(shù)，穩(wěn)定約束取代由數(shù)目有限的積極系統(tǒng)規(guī)定參數(shù)，在，通過定義凸穩(wěn)定域，設(shè)計(jì)問題可以制定作為迭代半定規(guī)劃（社民黨） ， ，一個(gè)穩(wěn)定的基礎(chǔ)上限制的論點(diǎn)原則（美聯(lián)社）的復(fù)雜的分析介紹了，由此產(chǎn)生的穩(wěn)定的制約因素成為線性平等條件. 然而 ，通過大量一些模擬，試圖解決這個(gè)問題，一個(gè)新的穩(wěn)定的制約因素是擬建議，這也是基于關(guān)于在穩(wěn)定的制約因素是約，在二次型， ，無論是大量的制約因素密集發(fā)車頻率點(diǎn)也不多交換算法是必要的。在實(shí)踐中，兩個(gè)最佳的標(biāo)準(zhǔn)，即極小和（加權(quán)）最小二乘法，，如等邊的通和高峰約束最小廣場阻，至少功率誤差最小，和加權(quán)積分的均方誤差（偉），我們將集中關(guān)于IIR濾波器設(shè)計(jì)中的加權(quán)最小二乘（英國威爾士）意義。凸優(yōu)化的方法，如二階錐規(guī)劃（ SOCP ），和社民黨，有被廣泛應(yīng)用于設(shè)計(jì)飛行情報(bào)區(qū)[和IIR數(shù)字過濾器。與唱片和凸資格問題，SOCP問題都可以納入更一般的限制是，唱片和凸資格問題是特殊情況SOCP問題。因此，所有的唱片和凸資格問題，也可用來幫助解決的SOCP SOCP問題不到一般比社民黨的問題，計(jì)算復(fù)雜度每次迭代所需的SOCP算法遠(yuǎn)遠(yuǎn)少于在社民黨的算法，尤其是當(dāng)層面二階錐（ SOC ） ，我們將投下IIR濾波器設(shè)計(jì)問題納入SOCP ，我們已經(jīng)提交了一份初步版本， 這次是為了將二次高峰期的限制，設(shè)計(jì)方法予以實(shí)施這一文件作為迭代SOCP程序，而不是量化參數(shù)迭代程序通過的。因此，二次成本函數(shù)在[ 26 ]，進(jìn)一步討論包括本文件中。，迭代SOCP方法沒有任何穩(wěn)定的制約因素是引入設(shè)計(jì)IIR數(shù)字濾波器 ，峰值限制被納入作為的SOC ，一個(gè)新的穩(wěn)定因素根據(jù)美聯(lián)社的是DSP例子中提出科IV. 最后，結(jié)論在第五節(jié).本文提出了一種IIR數(shù)字濾波器的設(shè)計(jì)方法，以新的P 基于穩(wěn)定約束已提交。作為復(fù)合的與以前的設(shè)計(jì)方法使用權(quán)值，設(shè)計(jì)的問題是制定一個(gè)迭代羥基的問題，它可以同時(shí)處理線性和SoC協(xié)商協(xié)議 。因此，一些更復(fù)雜的限制，例如波形的峰值誤差的限制，穩(wěn)定約束，這既是充分和NEC – 的穩(wěn)定。最后，擬議的穩(wěn)定的制約因素是強(qiáng)加所有的整個(gè)頻帶，從而大大促進(jìn)了設(shè)計(jì)的進(jìn)步。為了將其加入到設(shè)計(jì)過程中，類似的權(quán)值約束技術(shù)工作，和穩(wěn)定的制約因素是然后結(jié)合迭代程序。如果迭代過程收斂和參數(shù)，并有適當(dāng)?shù)倪x擇，穩(wěn)定約束才能保證穩(wěn)定的設(shè)計(jì)濾波器。當(dāng)最大極半徑是，通過調(diào)整參數(shù)，我們也可以控制桿的位置。魯棒性和有效性經(jīng)常提出的方法已被證明了的四個(gè)數(shù)值例子。 20 屆本科生畢業(yè)設(shè)計(jì)（論文）資料第三部分 過程管理資料 2009屆畢業(yè)設(shè)計(jì)（論文）課題任務(wù)書系(部)： 電子與通信工程系 專業(yè)： 通信工程 指導(dǎo)教師王路露職稱助教學(xué)生姓名何雙喜課題名稱IIR數(shù)字濾波器的仿真與實(shí)現(xiàn)