It would of course help to have your data.
However this call:
std_y = std(P_all,0,2);
implies to me that your data are column vectors, not row vectors.
If that is correct, then this would be more appropriate:
patch([x; flip(x)], [y1; flip(y2)], 'g')
patch([x; flip(x)], [y2; flip(y3)], 'g')
It is the same essential idea, simply recognising the orientation of the data.
The,flip works regardless of the orientation, while fliplr simply filps the column vectors in this instance (although is appropriate if that is what you want to do).
EDIT — (8 Mar 2023 at 2:28)
With the supplied data (they are column data) —
LD = load(websave('Mathworks','https://www.mathworks.com/matlabcentral/answers/uploaded_files/1317410/Mathworks.mat'))
CAD = LD.CAD;
P_all = LD.P_all;
Pav = LD.Pav;
x = (-360:0.5:359.5)';
mean_y = Pav;
std_y = std(P_all,0,2);
y1 = mean_y;
y2 = mean_y - std_y;
y3 = mean_y + std_y;
figure
plot(x,y1)
hold on
patch([x; flip(x)], [y1; flip(y2)], 'g')
patch([x; flip(x)], [y2; flip(y3)], 'g')
hold off
xlabel('CAD')
figure
plot(x,y1)
hold on
patch([x; flip(x)], [y1; flip(y2)], 'g')
patch([x; flip(x)], [y2; flip(y3)], 'g')
hold off
xlabel('CAD')
ylim([50 80])
title('Zoom to show detail')
The standad deviation is quite small, so the lower (optional) plot zooms in to show the details.
.
