Matlab Matrix calculation coming wrong

Hello,

I am working on a sensor data. I have three csv file from which I do voltage and strain calculation. After calculation of strain, when its comes to calculation of angle theta here is following condtion happening:

Condition 1: the one with correct calculations

If I take individual value(for example take value no 110 of from each variable i.e gauge1 , gauge2 and gauge3) and apply formula for performing the calculation of angle of strain I get the correct answer.

Conditon 2: the one with wrong calculated answer

However, If I apply this formula to entire matrix gauge1,2,3 I get differenent value for calculated angle. I know this one is wrong since I have calculated the values by hand and they don't match up with this condition.

I have attached csv file in order if one can help.

Thank you

Hiril

%%
clc;
clear;
% ADC and strain gauge settings
vref = 3300;                % ADC reference volts
ADC_resolution = 2^24;   % 2^23 24bit of resolution
A = 4;                      % if fsr = +/-2 
pga = 128;                    % ADS1234 amplifier gain setting
E = 3300;                   % Bridge excitation in millivolts
K = 2.04;                   % Gauge factor

% Importing adc values
ch1 = readmatrix('C:\Users\Hiril Patel\PycharmProjects\writetofile\correct strain gauge -0.8\ch0.csv');    
ch2 = readmatrix('C:\Users\Hiril Patel\PycharmProjects\writetofile\correct strain gauge -0.8\ch1.csv');
ch3 = readmatrix('C:\Users\Hiril Patel\PycharmProjects\writetofile\correct strain gauge -0.8\ch2.csv');


% Conversion to proportional millivoltage
FSR = vref/pga;
LSB = FSR/(ADC_resolution);
Channel1 = LSB*ch1;
Channel2 = LSB*ch2;
Channel3 = LSB*ch3;

Channel1mean = mean(Channel1(1:50));
Channel2mean = mean(Channel2(1:50));
Channel3mean = mean(Channel3(1:50));

Channel1 = Channel1 - Channel1mean;
Channel2 = Channel2 - Channel2mean;
Channel3 = Channel3 - Channel3mean;

% Strain calculation
gauge1 = (4/K)*(Channel1/E);
gauge2 = (4/K)*(Channel2/E);
gauge3 = (4/K)*(Channel3/E);

fig = figure;
subplot(3,1,1);
plot(gauge1,'.-');
title('Gauge1');
subplot(3,1,2);
plot(gauge2,'.-');
title('Gauge2');
subplot(3,1,3);
plot(gauge3,'.-');
title('Gauge3');
han=axes(fig,'visible','off'); 
han.Title.Visible='on';
han.XLabel.Visible='on';
han.YLabel.Visible='on';
ylabel(han,'Micro-strain');
xlabel(han,'Samples');


a = ((sqrt(2))/3)*(sqrt(((gauge1-gauge2).^2) + ((gauge2 - gauge3).^2) + ((gauge3-gauge1).^2)));
e1 = ((gauge1 + gauge2 + gauge3)/3) + a;
e2 = ((gauge1 + gauge2 + gauge3)/3) - a;

% Uptill here answers from calculation are correct.
%The problem starts after this point where entire matrix is considered for
%calculation

b = (sqrt(3))*(gauge3-gauge2);
c = ((2*gauge1) - gauge2 - gauge3);
T = b/c;
theta = (-0.5)*(atand(T));

figure;
if gauge1 > ((gauge2+gauge3)/2) 
    plot(theta);
elseif gauge1 < ((gauge2+gauge3)/2) 
    theta = theta - 90;
    plot(theta)
elseif gauge1 == ((gauge2+gauge3)/2) & gauge2<gauge1
    theta = - 45;
    plot(theta);
elseif gauge1 == ((gauge2 + gauge3)/2) & gauge2>gauge1 
    theta = 45;
    plot(theta);
else 
    plot(theta)
end

Answer 1

[Alex Hanes]

In Line 65, you are using mrdivide (/), when it seems you are trying to element-wise division using rdivide (./). mrdivide solves the system of linear equations for , which is why your array for T = b/c has dimension (630 x 630).

T = b./c; 

[Hiril Patel]

This worked, I didn't noticed it thank you. But with this done I got rise to another problem. The if statements at the end are not working. I made an array of variable with each set of all three gauges satisfying all the condition of if statement but 3 of 4 statement works but the 2 one doesn't (

elseif gauge1 < ((gauge2+gauge3)/2) 
    theta = theta-90;
    plot(theta)

). Here is the modification I made.

gauge1 = [700;300;500;500]

gauge2 = [700;300;300;700]

gauge3 = [300;700;700;300]

So each set of value satisfy the one of the if statment and the plots values of all the statement provided. I have one more question, how can I set condition if the value is negative then perform addition and if value is positive perform subtraction in the if statement mentioned above. I came up with this but its not helping.

    elseif gauge1 < ((gauge2+gauge3)/2) 
    theta = theta-(sign(theta))*90;
    plot(theta)

[Torsten]

You know that your if-statement means that the condition must be true for all elements of gauge in order to be evaluated as true, don't you ?

It is equivalent to

if (gauge1(1) < ((gauge2(1)+gauge3(1))/2)) && ...
   (gauge1(2) < ((gauge2(2)+gauge3(2))/2)) && ...
   (gauge1(3) < ((gauge2(3)+gauge3(3))/2)) && ...
   (gauge1(4) < ((gauge2(4)+gauge3(4))/2))
  theta = theta-(sign(theta))*90;
  plot(theta)

[Hiril Patel]

Actually no, so here is the thing in detail, I am measuring strain gauge output and there are different types of strain induced on the specimen and in order to identify where is the angle of the strain (theta) there is one of the 4 condition which needs to be met. SO each set of data from all the gauges should pass from the those condtion in order to identify what is the angle of strain.

I hope i have explained it clearly, but in case this is the literature that i am following:

check on page 5 below the equation 5: that states the reason of the condtion being present(but they are for rectangular rossette) and on the page six the condition mentioned are the ones that I am using(because I am using delta rosette)

https://intertechnology.com/Vishay/pdfs/TechNotes_TechTips/TN-515.pdf

The way i added data in this vairables

gauge1 = [700;300;500;500]

gauge2 = [700;300;300;700]

gauge3 = [300;700;700;300]

So the first element of the gauge1,2,3 satisfy first if statment and this gives me with the original value of theta which was derived from

b = (sqrt(3))*(gauge3-gauge2);
c = ((2*gauge1) - gauge2 - gauge3);
T = b./c;
theta = (-0.5)*(atand(T));

The second elements of gauge 1,2,3 satisfy second if statement(this is where I get wrong value if you run all three arrays using the formula above and pass them from my if statement then the plot gives me (-30.-30,-45,45) values of theta however this is not correct. The correct values should be (-30, -60, -45, 45), so clearly the I have made logical error of applying the condtions.

I hope I was able to explain clearly

[Torsten]

I hope I was able to explain clearly

Yes, but you didn't get my point (see below).

Answer 2

[Torsten]

% Strain calculation
gauge1 = [700;300;500;500];
gauge2 = [700;300;300;700];
gauge3 = [300;700;700;300];
fig = figure;
subplot(3,1,1);
plot(gauge1,'.-');
title('Gauge1');
subplot(3,1,2);
plot(gauge2,'.-');
title('Gauge2');
subplot(3,1,3);
plot(gauge3,'.-');
title('Gauge3');
han=axes(fig,'visible','off'); 
han.Title.Visible='on';
han.XLabel.Visible='on';
han.YLabel.Visible='on';
ylabel(han,'Micro-strain');
xlabel(han,'Samples');

a = ((sqrt(2))/3)*(sqrt(((gauge1-gauge2).^2) + ((gauge2 - gauge3).^2) + ((gauge3-gauge1).^2)));
e1 = ((gauge1 + gauge2 + gauge3)/3) + a;
e2 = ((gauge1 + gauge2 + gauge3)/3) - a;
% Uptill here answers from calculation are correct.
%The problem starts after this point where entire matrix is considered for
%calculation
b = (sqrt(3))*(gauge3-gauge2);
c = ((2*gauge1) - gauge2 - gauge3);
T = b./c;
theta = (-0.5)*(atand(T))

theta = 4×1

30.0000 30.0000 -45.0000 45.0000

figure;
for i = 1:numel(gauge1)
    if gauge1(i) > ((gauge2(i)+gauge3(i))/2) 
        disp(theta(i));
    elseif gauge1(i) < ((gauge2(i)+gauge3(i))/2) 
        theta(i) = theta(i) - 90;
        disp(theta(i))
    elseif gauge1(i) == ((gauge2(i)+gauge3(i))/2) & gauge2(i)<gauge1(i)
        theta(i) = - 45;
        disp(theta(i));
    elseif gauge1(i) == ((gauge2(i) + gauge3(i))/2) & gauge2(i)>gauge1(i)
        theta(i) = 45;
        disp(theta(i));
    else 
        disp(theta(i))
    end
end

   30.0000

   -60

   -45

    45

[Steven Lord]

for i = 1:numel(gauge1)
    if gauge1(i) > ((gauge2(i)+gauge3(i))/2) 
        display(theta(i));
    elseif gauge1(i) < ((gauge2(i)+gauge3(i))/2) 
        theta(i) = theta(i) - 90;
        display(theta(i))
    elseif gauge1(i) == ((gauge2(i)+gauge3(i))/2) & gauge2(i)<gauge1(i)
        theta(i) = - 45;
        display(theta(i));
    elseif gauge1(i) == ((gauge2(i) + gauge3(i))/2) & gauge2(i)>gauge1(i)
        theta(i) = 45;
        display(theta(i));
    else 
        display(theta(i))
    end
end

Another way to write this is to use logical indexing. Assuming that none of the gauge arrays contain NaN values:

% Compute the mean of gauge2 and gauge3 since we use it repeatedly
M = (gauge2 + gauge3)/2;
% Preallocate a new theta vector to store the transformed values
thetavalues = zeros(size(gauge1));

% Condition 1: gauge1 is greater than the mean
thetavalues(gauge1 > M) = theta(gauge1 > M);

% Condition 2: gauge1 is less than the mean
thetavalues(gauge1 < M) = theta(gauge1 < M) - 90;

% Condition3: we know gauge1 is equal to the mean. Use the sign function
s = sign(gauge2 - gauge1);
thetavalues(gauge1 == M) = s(gauge1 == M) * 45;

[Hiril Patel]

Indexing, didn't thought about that. thank you the solution.

[Hiril Patel]

Hello, sorry to be bother again but I would like to know if I can plot the value of the angle theta only concerning with acutal strain not the noise so currently I am using solution provided by Steven Lord but the problem arises when I zero down the offset of the signal. Here is the image. I though of plotting only section of angle value once strain is acted up it but it would be nice if I could remove noise somehow. Since the eqution considers everything as strain value and plot angles for it as well