风机SCADA数据清洗（缺失值填补+异常值处理） MATALB代码

%% 清空环境变量 warning off % 关闭报警信息 close all % 关闭开启的图窗 clear % 清空变量 clc % 清空命令行 %% 读取数据 res = xlsread('Wind_farm.xlsx'); res = res(5e3:1e4,:); res(:,10)=[]; %% 异常值处理 % 一行元素全部相同为-99 直接删除 same_pos = []; for i = 1:size(res,1) if all(res(i,1:end-1) == res(i,1)) same_pos = [same_pos;i]; end end res(same_pos,:) =[]; %% 缺失值处理 采用前面的值替代 missing = isnan(res); for i=1:size(res,1) for j=1:size(res,2) if missing(i,j) == 1 res(i,j) = res(i-1,j); end end end %% 离群值处理 [B,TF,L,U,C] = filloutliers(res(:,:),"clip","movmedian",15);%滑动窗线性插值 res_new = B; %% 绘制处理前特征与标签相关性热力图 R1 = corrcoef(res_new(:,[1,3,5,7,11])); %风速与功率相关性强 figure; pcolor(R1); shading flat; % 去掉网格线 colorbar; % 显示颜色条 % 美化图形 title('处理前相关性热力图'); xlabel('变量X'); ylabel('变量Y'); axis equal tight; % 等比例轴，紧凑范围 %% 相关性极差的也定义为异常值 % 处理：采用Ransac拟合后替代 x = res_new(:,7); y = res_new(:,end); xyPoints = [x y]; % RANSAC直线拟合 sampleSize = 25; % 每次采样的点数，直线为2 maxDistance = 350; % 内点到模型的最大距离 fitLineFcn = @(xyPoints) polyfit(xyPoints(:,1),xyPoints(:,2),1); % 拟合方式采用 polyfit，这里不可以用x,y替换xyPoints(:,1)，xyPoints(:,2) evalLineFcn = @(model, xyPoints) sum((y - polyval(model, x)).^2,2);% 距离估算函数 [modelRANSAC, inlierIdx] = ransac(xyPoints,fitLineFcn,evalLineFcn,sampleSize,maxDistance);% 执行RANSAC直线拟合，提取内点索引 modelInliers = polyfit(xyPoints(inlierIdx,1),xyPoints(inlierIdx,2),1);% 对模型内点进行最小二乘直线拟合 figure; plot(xyPoints(inlierIdx,1),xyPoints(inlierIdx,2),'.'); % 内点 hold on; plot(xyPoints(~inlierIdx,1),xyPoints(~inlierIdx,2),'ro'); % 外点 hold on; inlierPts = xyPoints(inlierIdx,:); x2 = linspace(min(inlierPts(:,1)),max(inlierPts(:,1))); y2 = polyval(modelInliers,x2); plot(x2, y2, 'g-'); % RANSAC直线拟合结果 hold off; title('最小二乘直线拟合 与 RANSAC直线拟合 对比'); xlabel('X(m)'); ylabel('Y(m)'); zlabel('Z(m)'); legend('内点','噪声点','RANSAC直线拟合','Location','NorthWest'); %% 残差-孤立森林 % 计算Ransac理论值 T_linear = (modelRANSAC(1)*res_new(:,7)+modelRANSAC(2)); for i = 1:size(T_linear,1) if T_linear(i,end)<0 T_linear(i,end)=0; end end residual_power = abs(res_new(:,end) - T_linear); % plot(T_linear) % 孤立森林判断异常值 [error_pos2] = iso_forest([res_new(:,[1,3,5,7]) residual_power]); %% 替代异常值 for i = 1:size(error_pos2,2) res_new(error_pos2{i,1},end) = T_linear(error_pos2{i,1}); end res_new(~inlierIdx,end) = T_linear(~inlierIdx); %% 绘制处理后特征与标签相关性热力图 R2 = corrcoef(res_new(:,[1,3,5,7,11])); figure; pcolor(R2); shading flat; % 去掉网格线 colorbar; % 显示颜色条 % 美化图形 title('处理后相关性热力图'); xlabel('变量X'); ylabel('变量Y'); axis equal tight; % 等比例轴，紧凑范围 %% 保存清洗好的数据 save res res_new;