基于matlab计算不等间距样本的一阶和二阶导数

简介: 基于matlab计算不等间距样本的一阶和二阶导数

✅作者简介:热爱科研的Matlab仿真开发者,修心和技术同步精进,matlab项目合作可私信。

🍎个人主页:Matlab科研工作室

🍊个人信条:格物致知。

更多Matlab仿真内容点击👇

智能优化算法       神经网络预测       雷达通信      无线传感器        电力系统

信号处理              图像处理               路径规划       元胞自动机        无人机

⛄ 内容介绍

1.1语法    

  [dy1,dy2,x1c,dy1c]=导数NE(x,y)

1.2输入      

    x 作为列或行向量,超过 2 个值

    y 作为列向量或行向量,与 x 的值数量相同

1.3输出      

    dy1:一阶导数

    dy2:二阶导数

         x(1) 处的第一个 dy2 和 x(end) 处的最后一个 dy2 不存在并设置为 NaN

    x1c 和 dy1c 返回精确的一阶导数 dy1c(x1c)

    如果 x 或 y 是行向量,则所有输出都作为行向量返回。

    出错时,所有输出都返回 NaN。

1.4语法示例      

          对于一导数:

                      dy1=导数NE([1;3;4;5],[1;9;16;25]);

          对于二阶数:

                     [~,dy2]=导数NE([1;3;4;5],[1;9;16;25]);

          对于更准确的一导数:

                     [~,~,x1c,y1c]=derivativeNE([1;3;4;5],[1;9;16;25]);


⛄ 完整代码

%Demo for 'derivativeNE.m'%After 5 clicks on 'run' you get the results for sine curves%No plot for dy1c(x1c), since dy1c is nearly on top of dy1clcclose all%Create parbola or sineif exist('curveTypeCounter','var')  curveTypeCounter=curveTypeCounter+1;  if curveTypeCounter>10    curveTypeCounter=1;  endelse  clear  curveTypeCounter=1;endcurveType=(curveTypeCounter>5);%  curveType=1;%0=force parabola, 1=force sine%Create parbola or sinen=round(rand(1)*150+10);%Number of samples, min=10, max=160% n=20;%Force n samplesif curveType==0  %Parabola with spike at x=0  xr=rand(n,1)*2;  x=sort(xr)-1;  %create spike  indxL=find(x<-0.1,1,'last');  indxR=find(x>0.15,1,'first');  x=[x(1:indxL);-0.1;0;0.15;x(indxR:end)];  x(1)=-1;  x(end)=1;  y=x.^2;  y(indxL+2)=0.05;%Spike amplitude  legendText1=sprintf('y=x^{2} with spike at x=0');else  %Sine  xr=rand(n,1)*2*pi;  x=sort(xr);  x(1)=0;  x(end)=2*pi;  y=sin(x);  legendText1='y=sin(x)';end[dy1,dy2,x1c,dy1c]=derivativeNE(x,y);%Calculate derivatives%Plot resultsnl=char(10);%new line, compatible to older Matlab releaseshfig=figure(1);clfhfig.Position(3)=hfig.Position(4)*1.8;yyaxis leftp1=plot(x,y,'.-g');% original curvexlabel('x')if curveType==0  %Parabola  ylabel('y = x^2 and spike at x=0')  xticks([-1,-.5,-.1,0,.15,.5,1])  text(1.27,0.5,['The exact derivates' nl 'at x=-0.1, 0, 0.15' nl 'are not defined as numbers.' nl ...    'They are for the 1st derivative' nl 'Heaviside functions' nl ...    'and for the 2nd derivative' nl 'Dirac functions.'],'BackgroundColor',[1,.85,.85],'FontSize',8)  yyaxis right  xexact=-1:0.01:1;  dy1exact=2*xexact;  p2=plot(xexact(1:91),dy1exact(1:91),'color',[.7,.7,1],'LineWidth',4);%left part exact 1st derivative  hold on  indxx0=find(x==0);  dyL=(y(indxx0)-y(indxx0-1))/(x(indxx0)-x(indxx0-1));  dyR=(y(indxx0+1)-y(indxx0))/(x(indxx0+1)-x(indxx0));  plot(xexact([91,91,101,101]),[dy1exact(91),dyL,dyL,dyR],'-','color',[.7,.7,1],'LineWidth',4);%left part of spike exact 1st derivative  plot(xexact([101,116,116]),[dyR,dyR,dy1exact(116)],'-','color',[.7,.7,1],'LineWidth',4);%right part of spike exact 1st derivative  plot(xexact(116:end),dy1exact(116:end),'-','color',[.7,.7,1],'LineWidth',4);%right part exact 1st derivative  p3=plot(xexact([1,90]),[2,2],'-','color',[1,.7,.8],'LineWidth',4);%left part exact 2nd derivative  plot(xexact([92,115]),[0,0],'-','color',[1,.7,.8],'LineWidth',4);%left and right part of spike exact 2nd derivative  plot(xexact([117,end]),[2,2],'-','color',[1,.7,.8],'LineWidth',4);%right part exact 2nd derivative  %Plot Diracs  quiver(xexact(91),0,0,8,'-','Color',[1,0.7,0.7],'LineWidth',2.5,'MaxHeadSize',.06)  quiver(xexact(101),0,0,-6,'-','Color',[1,0.7,0.7],'LineWidth',2.5,'MaxHeadSize',.08)  quiver(xexact(116),0,0,8,'-','Color',[1,0.7,0.7],'LineWidth',2.5,'MaxHeadSize',.06)    p4=plot(x,dy1,'b-');%1st numerical derivative interpolated  % p4=plot(x1s,dy1c,'.y');%1st derivative centered  p5=plot(x,dy2,'+r-');%2nd numerical derivative  legend([p1,p4,p2,p5,p3],{sprintf('y=x^{2} with spike at x=0'),'1st derivative numerical','1st derivative exact',...    '2nd derivative numerical','2nd derivative exact'},'Location','northeastoutside')  ylim([-8,12])    %Calculate Mean Absolute Errors and max errors without range of spike  %1st derivative at x  x4MAEdy1=[x(1:indxx0-2);x(indxx0+2:end)];%     x for MAE  dy14MAE= [dy1(1:indxx0-2);dy1(indxx0+2:end)];% dy1 for MAE  dy1AD=abs(dy14MAE-2*x4MAEdy1);%                Absolute differences  [dy1Max,dy1Maxindx]=max(dy1AD);%               Max absolute difference of dy1-dy1exact  dy1MAE= sum(dy1AD)/numel(x4MAEdy1);%           MAE dy1    %1st derivative at x1c including spike  x4MAEdy1c=[x1c(1:indxx0-2);x1c(indxx0+1:end)];%    x for MAE  dy1c4MAE= [dy1c(1:indxx0-2);dy1c(indxx0+1:end)];%  dy1c for MAE  dy1cAD=abs(dy1c4MAE-2*x4MAEdy1c);%                 Absolute differences  [dy1cMax,dy1cMaxindx]=max(dy1cAD);%                Max absolute difference of dy1c-dy1exact  dy1cMAE= sum(dy1cAD)/numel(x4MAEdy1c);%            MAE dy1c    %2nd derivative  x4MAEdy2=[x(2:indxx0-3);x(indxx0+3:end-1)];%      x for dy2 MAE  dy24MAE=[dy2(2:indxx0-3);dy2(indxx0+3:end-1)];%   dy2 for MAE  dy2AD=abs(dy24MAE-2);%                            Absolute differences  [dy2Max,dy2Maxindx]=max(dy2AD);%                  Max absolute difference of dy2-dy2exact  dy2MAE=sum(dy2AD)/numel(x4MAEdy2);%               MAE dy2    %Output text for MAE  text(1.27,-6,['Mean Abs. Error     Max abs. error' nl ...     'dy1:  ' num2str(dy1MAE,'%.1e') '         ' num2str(dy1Max,2) ' @x=' num2str(x4MAEdy1(dy1Maxindx),2) nl ...     'dy1c: ' num2str(dy1cMAE,'%.1e') '         ' num2str(dy1cMax,2) ' @x=' num2str(x4MAEdy1c(dy1cMaxindx),2) nl ...    'dy2:  ' num2str(dy2MAE,'%.1e') '         ' num2str(dy2Max,2) ' @x=' num2str(x4MAEdy2(dy2Maxindx),2) nl ...    'Range around spike is excluded'], ...    'BackgroundColor',[0.98,.98,.98],'FontSize',8)else  %Sine  ylim([-1.05,1.05])  ylabel('y = sin(x)')  xticks(0:pi/2:2*pi+eps)  set(gca,'TickLabelInterpreter','latex','XTickLabel',{'0','$\frac{\pi}{2}$','$\pi$','$\frac{3\pi}{2}$','$2\pi$'})%   set(gca,'XMinorGrid','on')%Only for test  yyaxis right  xexact=linspace(0,2*pi,200);  p2=plot(xexact,cos(xexact),'color',[.8,.8,1],'LineWidth',4);%         exact 1st derivative  hold on  p3=plot(xexact,-sin(xexact),'-','color',[1,.8,0.8],'LineWidth',4);%   exact 2nd derivative  p4=plot(x,dy1,'b-');%                                                 1st numerical derivative interpolated  % p4=plot(x1s,dy1c,'y-');%1st derivative centered  p5=plot(x,dy2,'+r-');%                                                2nd numerical derivative  legend([p1,p4,p2,p5,p3],{'y = sin(x)','1st derivative numerical','1st derivative exact',...    '2nd derivative numerical','2nd derivative exact'},'Location','northeastoutside')  ylim([-1.05,1.05])  xlim([0,2*pi])    %Calculate Mean Absolute Errors and max errors   %1st derivative at x  dy1AD=abs(dy1-cos(x));%            Absolute differences  [dy1Max,dy1Maxindx]=max(dy1AD);%   Max absolute difference of dy1-dy1exact  dy1MAE= sum(dy1AD)/numel(x);%      MAE dy1    %1st derivative at x1c   dy1cAD=abs(dy1c-cos(x1c));%             Absolute differences  [dy1cMax,dy1cMaxindx]=max(dy1cAD);%     Max absolute difference of dy1c-dy1exact  dy1cMAE= sum(dy1cAD)/numel(x1c);%       MAE dy1c    %2n derivative  x4MAEdy2=x(2:end-1);%                x for dy2 MAE  dy24MAE=dy2(2:end-1);%               dy2 for MAE  dy2AD=abs(dy24MAE+sin(x4MAEdy2));%   Absolute differences  [dy2Max,dy2Maxindx]=max(dy2AD);%     Max absolute difference of dy2-dy2exact  dy2MAE=sum(dy2AD)/numel(x4MAEdy2);%  MAE dy2      %Output text for MAE    text(7.2,0,['Mean Abs. Error     Max abs. error' nl ...     'dy1:  ' num2str(dy1MAE,'%.1e') '         ' num2str(dy1Max,'%.1e') ' @x=' num2str(x(dy1Maxindx),2) nl ...     'dy1c: ' num2str(dy1cMAE,'%.1e') '         ' num2str(dy1cMax,'%.1e') ' @x=' num2str(x1c(dy1cMaxindx),2) nl ...    'dy2:  ' num2str(dy2MAE,'%.1e') '         ' num2str(dy2Max,'%.1e') ' @x=' num2str(x4MAEdy2(dy2Maxindx),2)], ...    'BackgroundColor',[0.98,.98,.98],'FontSize',8)endylabel('1st and 2nd derivatives')hold offgrid on
function [dy1,dy2,x1c,dy1c]=derivativeNE(x,y)%Calculate the 1st and 2nd derivative for Not Equal spaced samples.%%INPUT%  x as column or row vector, more than 2 values%  y as column or row vector, same amount of values as for x%%OUTPUT%  dy1: 1st derivative, y'(x)%       %  dy2: 2nd derivative, y"(x)%       The first dy2 at x(1) and the last dy2 at x(end) don't exist and are set to NaN%  x1c and dy1c return a more accurate (about 3 times better) 1st derivative dy1c(x1c), but with new x-values%%  If x or y is a row vector, all outputs are returned as row vectors.%  On error NaN is returned%%Syntax examples:%  dy1=derivativeNE([1;3;4;5],[1;9;16;25]);           % 1st derivative y'(x)%  [~,dy2]=derivativeNE([1;3;4;5],[1;9;16;25]);       % 2nd derivative y"(x)%  [~,~,x1c,y1c]=derivativeNE([1;3;4;5],[1;9;16;25]); % 1st derivative more accurate (about 3 times), y'(x1c)%%Peter Seibold, August 2023%Check inputsif size(x,2)>1 || size(y,2)>1  %Either x or y is a row vector  isRowVector=true;  x=x(:);%convert to column vector  y=y(:);else  isRowVector=false;endif numel(x)~=numel(y) || numel(x)<3  dy1=NaN; dy2=NaN; x1c=NaN; dy1c=NaN;%Return all output as NaN  disp('x and y must have the same size and at least 3 elements!')  returnend%1st derivativedx=diff(x);dy=diff(y);dy1c=dy./dx;%          1st derivative for centered xx1c=x(1:end-1)+dx/2;%  center each x (shift each x by half distance to next x)%interpolate/extrapolate dy1c(x1c) to match x, result: dy1(x)y2=[dy1c(2);dy1c(2:end);dy1c(end)];   %repeat border values for extrapolationy1=[dy1c(1);dy1c(1:end-1);dy1c(end-1)];x1=[x1c(1);x1c(1:end-1);x1c(end-1)];x2=[x1c(2);x1c(2:end);x1c(end)];dy1=(y2-y1)./(x2-x1).*(x-x2)+y2;      %interpolate/extrapolate%2nd derivativey32x32=dy1c(2:end);                   %1st derivatives at even positionsy21x21=dy1c(1:end-1);                 %1st derivatives at odd positionsdx31=y32x32;                          %create vector with dummy values, with two elements less than number of x-elementsdx31(1:2:end)=x(3:2:end)-x(1:2:end-2);%x-difference odd pairsdx31(2:2:end)=x(4:2:end)-x(2:2:end-2);%x-difference even pairsdy2c=2*(y32x32-y21x21)./dx31;         %2nd derivatives dy2 at center of the two surrounding samplesx2c=dy2c;                             %create vector with dummy values, with two elements less than number of x-elementsx2c(1:2:end)=(x(1:2:end-2)+x(3:2:end))/2;%x for dyc at center of the two surrounding samples, (overwrite dummy values)x2c(2:2:end)=(x(2:2:end-2)+x(4:2:end))/2;%dito for even x centered, now we have all center x2c for 2nd derivatives%interpolate dy2c(x2c) to match x, result: dy2(x)if numel(x)>3  dy2=[NaN;interp1(x2c,dy2c,x(2:end-1),'linear','extrap');NaN];%interpolate and pad unavailable border valueselse  %Interpolation needs at least two dy2c samples  dy2=[NaN;dy2c;NaN];%    output only the single dy2(x)end%Convert back to user formatif isRowVector  %Either x or y input was a row vector  %Convert to row vector  dy1=dy1';  dy2=dy2';  x1c=x1c';  dy1c=dy1c';end

⛄ 运行结果


⛳️ 代码获取关注我

❤️部分理论引用网络文献,若有侵权联系博主删除
❤️ 关注我领取海量matlab电子书和数学建模资料

🍅 仿真咨询

1 各类智能优化算法改进及应用

生产调度、经济调度、装配线调度、充电优化、车间调度、发车优化、水库调度、三维装箱、物流选址、货位优化、公交排班优化、充电桩布局优化、车间布局优化、集装箱船配载优化、水泵组合优化、解医疗资源分配优化、设施布局优化、可视域基站和无人机选址优化

2 机器学习和深度学习方面

卷积神经网络(CNN)、LSTM、支持向量机(SVM)、最小二乘支持向量机(LSSVM)、极限学习机(ELM)、核极限学习机(KELM)、BP、RBF、宽度学习、DBN、RF、RBF、DELM、XGBOOST、TCN实现风电预测、光伏预测、电池寿命预测、辐射源识别、交通流预测、负荷预测、股价预测、PM2.5浓度预测、电池健康状态预测、水体光学参数反演、NLOS信号识别、地铁停车精准预测、变压器故障诊断

2.图像处理方面

图像识别、图像分割、图像检测、图像隐藏、图像配准、图像拼接、图像融合、图像增强、图像压缩感知

3 路径规划方面

旅行商问题(TSP)、车辆路径问题(VRP、MVRP、CVRP、VRPTW等)、无人机三维路径规划、无人机协同、无人机编队、机器人路径规划、栅格地图路径规划、多式联运运输问题、车辆协同无人机路径规划、天线线性阵列分布优化、车间布局优化

4 无人机应用方面

无人机路径规划、无人机控制、无人机编队、无人机协同、无人机任务分配

5 无线传感器定位及布局方面

传感器部署优化、通信协议优化、路由优化、目标定位优化、Dv-Hop定位优化、Leach协议优化、WSN覆盖优化、组播优化、RSSI定位优化

6 信号处理方面

信号识别、信号加密、信号去噪、信号增强、雷达信号处理、信号水印嵌入提取、肌电信号、脑电信号、信号配时优化

7 电力系统方面

微电网优化、无功优化、配电网重构、储能配置

8 元胞自动机方面

交通流 人群疏散 病毒扩散 晶体生长

9 雷达方面

卡尔曼滤波跟踪、航迹关联、航迹融合
相关文章
|
4月前
|
算法 计算机视觉 异构计算
基于FPGA的图像PSNR质量评估计算实现,包含testbench和MATLAB辅助验证程序
基于FPGA的图像PSNR质量评估计算实现,包含testbench和MATLAB辅助验证程序
|
5月前
|
计算机视觉
专题七MATLAB符号计算-2
专题七MATLAB符号计算
37 0
|
5月前
专题七MATLAB符号计算-1
专题七MATLAB符号计算
39 0
|
12天前
|
传感器 存储 数据库
matlab测量计算信号的相似度
matlab测量计算信号的相似度
14 0
|
7月前
|
机器学习/深度学习 传感器 算法
【TGV 正则器的快速计算方法】通过FFT的总(广义)变化进行图像去噪(Matlab代码实现)
【TGV 正则器的快速计算方法】通过FFT的总(广义)变化进行图像去噪(Matlab代码实现)
|
3月前
【数值分析】用幂法计算矩阵的主特征值和对应的特征向量(附matlab代码)
【数值分析】用幂法计算矩阵的主特征值和对应的特征向量(附matlab代码)
|
7月前
|
机器学习/深度学习 传感器 算法
【气动学】基于涡格法实现气动力计算附Matlab代码
【气动学】基于涡格法实现气动力计算附Matlab代码
|
8月前
|
算法
基于遗传算法和粒子群算法的潮流计算比较(Matlab代码实现)
基于遗传算法和粒子群算法的潮流计算比较(Matlab代码实现)
|
8月前
|
流计算
电力系统潮流计算(牛顿-拉夫逊法、高斯-赛德尔法、快速解耦法)【6节点 9节点 14节点 26节点 30节点 57节点】(Matlab代码实现)
电力系统潮流计算(牛顿-拉夫逊法、高斯-赛德尔法、快速解耦法)【6节点 9节点 14节点 26节点 30节点 57节点】(Matlab代码实现)
143 0
|
4月前
|
算法 计算机视觉
基于WTMM算法的图像多重分形谱计算matlab仿真
基于WTMM算法的图像多重分形谱计算matlab仿真

热门文章

最新文章