蒙特卡羅-牛頓法進(jìn)行電路產(chǎn)品合格率估計(jì)和最優(yōu)中心設(shè)計(jì)
MCA-NEWTON METHOD FOR CIRCUIT YIELD ESTIMATION AND OPTIMIZATION
-
摘要: 本文利用蒙特卡羅法估計(jì)電路產(chǎn)品合格率對(duì)元件參數(shù)中心值的一、二階偏導(dǎo)數(shù),再與牛頓最優(yōu)化方法結(jié)合得到了一個(gè)電路中心設(shè)計(jì)的算法,為了提高偏導(dǎo)數(shù)估計(jì)值的精度,文中又給出了利用失敗樣本點(diǎn)的估計(jì)式。為了減少電路分析次數(shù),作者提出了適用于蒙特卡羅分析的采樣頻率按權(quán)排序法,理論分析和算例表明:本文算法是成功的,適用于非凸、非單連通可行域,可進(jìn)行中等規(guī)模電路的產(chǎn)品合格率估計(jì)和中心設(shè)計(jì)。
-
關(guān)鍵詞:
Abstract: An algorithm for circuit yield estimation and optimization is described. To obtain the first and second derivatives of yield with respect to center and get the optimal circuit center, the Monte Carlo analysis and Newton method are employed. By using fail samples and rearranging sample frequencies according to their weights, two new methods for improving estimation and reducing computation are presented. Both theoretical analysis and calculation examples show that this algorithm performs well and can be used for middle-scale circuit design. No matter the region of acceptability is convex set or not, the algorithm is available. -
R.K. Brayton, et al.,Proc. IEEE, 69(1981), 1334.[2]K.Singghal and J. F. Piney IEEE Trans. on CAS, CAS-28(1981), 692.[3]D. E. Hocevar, et al., ibid, CAS-30(1983 ), 180.[4]D. M. Himmelbiau著,張文燊等譯,實(shí)用非線性規(guī)劃,科學(xué)出版社,1983.[5]于驍勇,電子電路優(yōu)化設(shè)計(jì)的一個(gè)算法,中國(guó)電子學(xué)會(huì)CAS學(xué)會(huì)第五屆年會(huì)論文集(西安),1989,pp 364-8.[6]E. Wehrhahn and R. Spence, IEEE Proc. ISCAS84, pp. 1424-38.[7]R. S. Soin and R. Spence, IEE Proc., pt. G, 127 (1980) , 260.[8]D. Agnew, and P. E. Men, ibid., 127(1980). 270.[9]M. A. Styblinski, et al., IEEE Proc. ISCAS81, pp. 554-7.[10]K. S. Tahim and R. Spence, IEEE Trans. on CAS, CAS-26(1979), 768.[11]于驍勇,重慶大學(xué)碩士論文,1985. -
計(jì)量
- 文章訪問(wèn)數(shù): 1919
- HTML全文瀏覽量: 153
- PDF下載量: 394
- 被引次數(shù): 0
下載:
