论坛 › 信号处理方法 › 关于EMD分解的边界效应处理

sssbbbhhh1 发表于 2009-3-18 12:06

我调试了以上方法，发现几个问题

1左端处理
avg=(sum(abs(extrx1))+sum(abs(rightvalue)))/(length(extrx1)*2-2);
       error = rightvalue - extrx1;
中rightvalue应为leftvalue
2.error 不能作为变量名

peterjiasheng 发表于 2009-5-16 16:58

回复 7楼 cleverblue 的帖子

请问那个积分中值求均值的程序谁有，能发我一份吗？我的邮箱是peterjiasheng@163.com 不胜感激

peterjiasheng 发表于 2009-5-18 11:14

回复 7楼 cleverblue 的帖子

= compose(tmin,tmax,mmin,mmax);
       = mvtd(m,mvalue,indt,0);    % 用中值定理法求均值
请问这个引用compose 函数是不是有问题？我仿真怎么说是:Function 'compose' is not defined for values of class 'double'. 请问这是什么意思？谁能指导一下不？

peterjiasheng 发表于 2009-5-18 21:23

回复 7楼 cleverblue 的帖子

tmax=;
mmax=;

tmin=;
mmin=[-2.296846525428775,-4.141623383125126,-4.084400242074468,0.165722184124079,-2.581249662433192,-2.116472992139788,1.136570601261942,-3.045081330767765,-2.972607334128952,-1.925543895436031,-4.871520644665601,-0.521316458710067,-2.192876415860649,-1.861963514986633,1.50560196796016,-2.914292656518384,0.519140276285282,-3.883123848963708,-2.806325396027761,-2.237631600793283,-3.221226039291187,-1.071224806274728,-0.463477168999168,-1.660299945403862,-0.664397103160116,-0.526777625109889,-3.360709334153492,-2.280793256440808,-2.599278536768881,-3.583498810153754,-1.704885104834053,-0.629334466285749,-1.968077926144889,-0.755271962500725,-0.039147942604617,-2.810828312810431,-1.515612372912226,-2.587602470818654,-3.686622880335742,-0.992898634030865,-2.650747503176382,-1.059857342489654,-0.17242152090046,-2.133342172288208,-1.122432327791427,-1.849779184199289,-3.567453121163892,-2.467638879805008,-1.773115313930688,-3.191370307708022,-1.532393381456038,-0.239410434908048,-1.93102391459998,-0.725532981877858,-1.169369353871392,-3.110854081074671,-2.23262198890181,-2.097533990735767,-3.644475842453983,-2.120377601127208,-0.66299325932594,-2.143704482476587,-0.614079032953674,-0.614031647854722,-2.497890976565047,-1.701615968272349,-1.986760566989064,-3.817983650228092,-2.553637686498472,-1.28058664316398,-2.64853996479398,-0.873588261974826,-0.343924125988255,-2.000157010512718,-1.0932630027658,-1.544083183954068,-3.616255743283878,-2.655737881200884,-1.808347179470624,-3.26063830051846,-1.416267211320117,-0.448382168487173,-1.812790735921519,-0.656267804581154,-0.915896862890417,-3.123076577064205,-2.38988688574425,-2.078428391319418,-3.735177928240775,-2.02869869489549,-0.881850028828932,-2.016003576784173,-0.562557670202894,-0.388343471438504,-2.516540075926342,-1.835828563506895,-1.970311318325455,-3.890841372179616,-2.493810947919833,-1.474870604670952,-2.540261707017983,-0.838820338896716,-0.166740104751433,-2.012138332084731,-1.204415019661657,-1.516998345942193,-3.683751537147896,-2.613449487021327,-1.981328346299294,-3.164114276511798,-1.399836502872536,-0.303680664333675,-1.837874599006318,-0.746069480804387,-0.926225414265623,-3.168330811298507,-2.353297308007195,-2.206431940274037,-3.661880101337778,-2.020687925490941,-0.771780254833345,-2.044695323077893,-0.629092413112705,-0.397906901877186,-2.545752266250568,-1.813895590765952,-2.069884659645433,-3.844333038767063,-2.470203233444309,-1.387488094867794,-2.555824767738202,-0.897152642123674,-0.153787065247497,-2.047205964498041,-1.196905816530512,-1.611901670682193,-3.646196591169452,-2.586880495506465,-1.914720647936704,-3.175446974512419,-1.442574391930074,-0.295889165773106,-1.862858113645833,-0.738553757370997,-0.993466440201802,-3.143090804173905,-2.330644234655106,-2.700512798743145];
我产生的数据是这样的，用 = compose(tmin,tmax,mmin,mmax);好像不可以？通过不了，谁能帮我看看，是不是这个函数用错啦，请问谁知用哪个函数不？:loveliness:

nkdtxf 发表于 2010-5-16 11:19

请问哪位知道神经网络，或者支持向量机的延拓方法程序呢。谢谢了

aishuishou 发表于 2010-12-29 00:07

回复 9 # cleverblue 的帖子

为什么我写的延拓程序放到emd里面，只迭代一次就停止了，另外请问你怎么看GRILLING镜像法的：
function = boundary_conditions(indmin,indmax,t,x,z,nbsym)
lx = length(x);
if (length(indmin) + length(indmax) < 3) //这个你怎么看？在emd理论里没有啊？！
error('not enough extrema')
end

lanceinmf 发表于 2011-1-6 10:47

先学习学习了

younger1985 发表于 2011-1-9 11:00

回复 20 # nkdtxf 的帖子

其实你用神经网络或者支持向量机做的效果不见的好，你试试就知道了 ！哎，感觉好多理论科学都是实际都脱节！

younger1985 发表于 2011-1-9 11:08

回复 21 # aishuishou 的帖子

有吧 你参考一下浙江大学胡劲松博士论文里面就有这个

aishuishou 发表于 2011-1-9 19:40

回复 24 # younger1985 的帖子

哈哈，谢谢提醒啊！！！我当初看这篇论文感觉有点难，没细看，刚才有回去看了一遍确实，他在博士毕业论文里2.3.2节说了，至少两个极值点（和主题里论文的要求极值点大于2没有本质区别，因为因为emd工具箱处理是按至少三个极值点处理的，结果就是：两个极值点时最后的趋势想为y=0，而且程序未予显示，并且三个极值点的话就会产生一个单调趋势项），终于明白了！！！

nkdtxf 发表于 2011-1-11 09:57

谢谢，楼上，提醒我了，明白了，趋势项的理论依据了

hitdely 发表于 2011-2-23 18:29

回复 20 # nkdtxf 的帖子

请问神经网络，或者支持向量机的延拓方法程序你那有吗 可以分享一下吗 谢谢

yuwenamy 发表于 2012-2-27 19:42

去掉emd.m中的边界延拓部分的程序,怎样去啊 ！

yueemd 发表于 2012-3-13 16:37

恩啊，我也有跟ls一样的问题啊，不知道该怎么去掉emd中的延拓部分，希望大家帮忙解释一下。。。谢谢了

peterzitong 发表于 2012-6-25 15:58

回复 15 # cleverblue 的帖子

不知道楼主还在此论坛里面吗

查看完整版本: 关于EMD分解的边界效应处理