Interpolate x value at single y value in dataset.
Afficher commentaires plus anciens
Hi,
I have a CSV file with time and temperature values. In the app designer I kind of figured everything out (import file, plot the data, display some data in a table) however I just can’t find a solution for the following problem:
I would like to determine/interpolate the time at which my temperature reaches my threshold of 7°C. The problem is that my temperature is sampled every 15 min so my sampled values bounces above and below my threshold. Here I have the code so far with a code i integrated from a solution to a similar problem I found in Matlab Help (Sorry in advance I am not a good programmer):
% Get the data from the TemperaturDaten table
temperaturTableData = app.Temperaturdaten.Data;
% Get the selected row indices from UserData for Temperaturdaten
temperaturIndices = app.Temperaturdaten.UserData;
% Get selected data on the new figure for Temperaturdaten
for i = 1:length(temperaturIndices)
% Get the data from the selected row
rowData = temperaturTableData(temperaturIndices(i), :);
% Read the CSV file from the selected entry
filename = rowData{1};
fullFilePath = fullfile(pwd, filename);
% Read the CSV file
csvTable = readtable(fullFilePath);
% Convert the table to a timetable to combine Date and
% time of day to a single column
tt = table2timetable(csvTable, 'RowTimes', datetime(csvTable.Datum + csvTable.Zeit,'Format','dd/MM/yyyy HH:mm:ss'));
%convert timetable back to table TT.Wert being the Temperature value
%and TT.Time being the date and time values
TT = timetable2table(tt);
%Get all values above 7°C threshhold
TT.ubersieben = TT.Wert;
TT.ubersieben(TT.ubersieben<7)=0;
%Get time at 7°C threshhold. "This is my problem I can't solve."
dy=diff([0 TT.Wert]);
dyix=find(dy==0);
TT.Wert(dyix)=TT.Wert(dyix-1)+1E-8;
schnittsieben = interp1(TT.Wert,TT.Time,7,"linear");
disp(schnittsieben);
%Export to app design table.
Tabellenausgabe = table(TT.Time, TT.Wert,TT.ubersieben);
end
% Assign data to the UITable
app.UITable.Data = Tabellenausgabe;
The error I get is: “Dimensions of arrays being concatenated are not consistent.”
I added a plot to potentially clarify the time values I am looking for.
If someone could point me in the right direction, I would be very thankful.
1 commentaire
imshow(imread('Matlab Frage.png'))
.
Réponse acceptée
Star Strider
le 12 Fév 2024
Modifié(e) : Star Strider
le 12 Fév 2024
Having your data definitely helps!
Try this —
% DT = datetime(2024,2,12) + minutes(0:15:13*60).';
% TC = rand(size(DT))*6+3;
T1 = readtable('311FT01Temp.csv')
T1.DateTime = datetime(string(T1.Datum)) + timeofday(datetime(string(T1.Zeit))); % Create & Concatenate Single 'datetime' Array Variable
L = size(T1,1);
TC = T1.Wert;
DT = T1.DateTime;
% return
% L = numel(DT);
TC_Target = 7; % Target Temperature Value
idx = find(diff(TC - TC_Target)); % Indices Of Approximate 'TC_Target' Values
for k = 1:numel(idx)
idxrng = max(1,idx(k)) : min(L,idx(k)+1); % Index Range For Interpolation
[TCu,ixu] = unique(TC(idxrng)); % 'Fudge Factor' To Avoid Non-Unique Temperatures
TE(k) = interp1(TC(idxrng(ixu)), DT(idxrng(ixu)), TC_Target); % Interpolate Temperatures To Get Time
end
TC7_Times = TE(~ismissing(TE)).';
TC7_Times.Format = 'HH:mm:ss.SSSSSS'
figure
plot(DT, TC)
hold on
plot(TE, ones(size(TE))*7, 'sr')
hold off
grid
yline(7, '--g')
xlabel('Time')
ylabel('T (°C)')
title('Temperature Data Showing Intercept Times')
figure
plot(DT, TC)
hold on
plot(TE, ones(size(TE))*7, 'sr')
hold off
grid
yline(7, '--g')
xlabel('Time')
ylabel('T (°C)')
title('Temperature Data Showing Intercept Times - Detail')
xlim([datetime(2024,1,8,06,00,00) datetime(2024,1,8,14,00,00)])
I had to tweak the approach that I usually use for these sorts of problems to work with your data.
I am pleased with these results.
EDIT — (12 Feb 2024 at 14:37)
Re-ran my previous code with the data when it posted. Required a minor tweak to my earlier, original code.
.
5 commentaires
Alex
le 12 Fév 2024
As always, my pleasure!
Integrating everything above ‘TC_Target’ would be straightforward except for the independent variable being time here. That required a bit of experimentation because the trapz function (used to do the integration) wants only numerical data, not datetime values. An additional consideration is where the temparatures exceed the ‘TC_Target’ threshold, and here that begins with the second intersection, so the integration loop begins with the second intersection and ends with the last intersection given that there must be an even number of them. I added a pair of if blocks that check for those in order to make this as autonomous as possible. I believe all that now works correctly. This now does everything including the integrations. I also added detail plots at the beginning and end of the data. It took a bit of time to get everything working the way I want it to.
The code —
T1 = readtable('311FT01Temp.csv')
T1.DateTime = datetime(string(T1.Datum)) + timeofday(datetime(string(T1.Zeit))); % Create & Concatenate Single 'datetime' Array Variable
L = size(T1,1);
TC = T1.Wert;
DT = T1.DateTime;
TC_Target = 7; % Target Temperature Value
idx = find(diff(TC - TC_Target)); % Indices Of Approximate 'TC_Target' Values
for k = 1:numel(idx)
idxrng = max(1,idx(k)) : min(L,idx(k)+1); % Index Range For Interpolation
[TCu,ixu] = unique(TC(idxrng)); % Avoid Non-Unique Temperatures
TE(k) = interp1(TC(idxrng(ixu)), DT(idxrng(ixu)), TC_Target); % Interpolate Temperatures To Get Time
slope(k,:) = TC(idxrng(end))-TC(idxrng(1));
end
Lv = ~ismissing(TE);
slope = slope(Lv);
TC7_Times = TE(~ismissing(TE)).';
TC7_Times.Format = 'HH:mm:ss.SSSSSS'
start_idx = 1;
if slope(1) < 0
start_idx = 2;
end
if rem(numel(TC7_Times)-start_idx+1, 2) == 0
end_idx = numel(TC7_Times)-1;
end
for k = start_idx : end_idx % Start At Second Intersection
k = k-1; % Correction For Beginning With Second Entry
TI(k,:) = [TC7_Times(k) TC7_Times(k+1)]; % Interval Times
Timev = TI(k,1) : minutes(1) : TI(k,2); % Minutes In Interval
Tempv = interp1(DT, TC, Timev); % Interpolate Temperature In Interval
IntervalDuration(k,:) = numel(Timev); % Interval Curation (Minutes)
T_int(k,:) = trapz((1:numel(Timev)), Tempv); % Units: °C x Minute
end
Time_Temp_Table = table(TI, IntervalDuration, T_int, 'VariableNames',{'Time Interval','Interval Duration (Minutes)','Integrated Temperature (°C x Minute)'})
figure
plot(DT, TC)
hold on
plot(TE, ones(size(TE))*7, 'sr')
hold off
grid
yline(7, '--g')
xlabel('Time')
ylabel('T (°C)')
title('Temperature Data Showing Intercept Times')
figure
plot(DT, TC)
hold on
plot(TE, ones(size(TE))*7, 'sr')
hold off
grid
yline(7, '--g')
xlabel('Time')
ylabel('T (°C)')
title('Temperature Data Showing Intercept Times - Detail')
xlim([datetime(2024,1,8,06,00,00) datetime(2024,1,8,14,00,00)])
Ax = gca;
Ax.XMinorTick = 'on';
figure
plot(DT, TC)
hold on
plot(TE, ones(size(TE))*7, 'sr')
hold off
grid
yline(7, '--g')
xlabel('Time')
ylabel('T (°C)')
title('Temperature Data Showing Intercept Times - Start Detail')
xlim([DT(1) DT(1)+hours(4)])
Ax = gca;
Ax.XMinorTick = 'on';
figure
plot(DT, TC)
hold on
plot(TE, ones(size(TE))*7, 'sr')
hold off
grid
yline(7, '--g')
xlabel('Time')
ylabel('T (°C)')
title('Temperature Data Showing Intercept Times - End Detail')
xlim([DT(end)-hours(5) DT(end)])
Ax = gca;
Ax.XMinorTick = 'on';
.
Alex
le 13 Fév 2024
As always, my pleasure!
The integral function is used to integrate functions, and trapz is used to integrate vectors. Your data are vectors.
This problem requires extracting the necessary data into new table arrays (for efficiency) and then converting those to timetable arrays. The timetable arrays are then combined using the synchronize function to create a single timetable with all the necessary information. After that, it is only necessary to use a bit of the same sort of logic I used previously here to find the next peak following the time the heat is removed. (I assume that the 0 values in the ‘Zust.’ variable mean that the heat is turned off.)
I believe this does what you want —
T1 = readtable('311FT01Temp.csv');
T1.DateTime = datetime(string(T1.Datum)) + timeofday(datetime(string(T1.Zeit))) % Create & Concatenate Single 'datetime' Array Variable
L = size(T1,1);
TC = T1.Wert;
DT = T1.DateTime;
TC_Target = 7; % Target Temperature Value
idx = find(diff(TC - TC_Target)); % Indices Of Approximate 'TC_Target' Values
for k = 1:numel(idx)
idxrng = max(1,idx(k)) : min(L,idx(k)+1); % Index Range For Interpolation
[TCu,ixu] = unique(TC(idxrng)); % Avoid Non-Unique Temperatures
TE(k) = interp1(TC(idxrng(ixu)), DT(idxrng(ixu)), TC_Target); % Interpolate Temperatures To Get Time
slope(k,:) = TC(idxrng(end))-TC(idxrng(1));
end
Lv = ~ismissing(TE);
slope = slope(Lv);
TC7_Times = TE(~ismissing(TE)).';
TC7_Times.Format = 'HH:mm:ss.SSSSSS'
start_idx = 1;
if slope(1) < 0
start_idx = 2;
end
if rem(numel(TC7_Times)-start_idx+1, 2) == 0
end_idx = numel(TC7_Times)-1;
end
for k = start_idx : end_idx % Start At Second Intersection
k = k-1; % Correction For Beginning With Second Entry
TI(k,:) = [TC7_Times(k) TC7_Times(k+1)]; % Interval Times
Timev = TI(k,1) : minutes(1) : TI(k,2); % Minutes In Interval
Tempv = interp1(DT, TC, Timev); % Interpolate Temperature In Interval
IntervalDuration(k,:) = numel(Timev); % Interval Curation (Minutes)
T_int(k,:) = trapz((1:numel(Timev)), Tempv); % Units: °C x Minute
end
Time_Temp_Table = table(TI, IntervalDuration, T_int, 'VariableNames',{'Time Interval','Interval Duration (Minutes)','Integrated Temperature (°C x Minute)'})
% figure
% plot(DT, TC)
% hold on
% plot(TE, ones(size(TE))*7, 'sr')
% hold off
% grid
% yline(7, '--g')
% xlabel('Time')
% ylabel('T (°C)')
% title('Temperature Data Showing Intercept Times')
% figure
% plot(DT, TC)
% hold on
% plot(TE, ones(size(TE))*7, 'sr')
% hold off
% grid
% yline(7, '--g')
% xlabel('Time')
% ylabel('T (°C)')
% title('Temperature Data Showing Intercept Times - Detail')
% xlim([datetime(2024,1,8,06,00,00) datetime(2024,1,8,14,00,00)])
% Ax = gca;
% Ax.XMinorTick = 'on';
% figure
% plot(DT, TC)
% hold on
% plot(TE, ones(size(TE))*7, 'sr')
% hold off
% grid
% yline(7, '--g')
% xlabel('Time')
% ylabel('T (°C)')
% title('Temperature Data Showing Intercept Times - Start Detail')
% xlim([DT(1) DT(1)+hours(4)])
% Ax = gca;
% Ax.XMinorTick = 'on';
% figure
% plot(DT, TC)
% hold on
% plot(TE, ones(size(TE))*7, 'sr')
% hold off
% grid
% yline(7, '--g')
% xlabel('Time')
% ylabel('T (°C)')
% title('Temperature Data Showing Intercept Times - End Detail')
% xlim([DT(end)-hours(5) DT(end)])
% Ax = gca;
% Ax.XMinorTick = 'on';
% f = fileread('311FT01Heat.csv')
T2 = readtable('311FT01Heat.csv', 'VariableNamingRule','preserve');
T2.DateTime = datetime(string(T2.Datum)) + timeofday(datetime(string(T2.Zeit)))
T1r = T1(:, [7 5]) % Create New, Edited 'table' Arrays For This Part
T2r = T2(:, [8 5]) % Create New, Edited 'table' Arrays For This Part
TT1r = table2timetable(T1r) % Convert Them To 'timetable' Arrays
TT2r = table2timetable(T2r) % Convert Them To 'timetable' Arrays
TT3r = synchronize(TT1r, TT2r) % Use 'synchronize' To Combine The 'timetable' Arrays With A Common Time Reference
figure
hp = plot(TT3r.DateTime, TT3r{:,1}, 'DisplayName','Temperature (°C)');
grid
Lv_Zust = (TT3r.('Zust.') == 0); % Logical Index Vector
hx = xline(TT3r.DateTime(Lv_Zust), '-r', 'DisplayName','activer Zustand aus'); % 'on' Times
xlim([TT3r.DateTime(1) datetime(2024,01,08,05,00,00)])
xlabel('Date & Time')
ylabel('°C')
legend([hp hx(1)], 'Location','northoutside','Orientation','horizontal')
idx_Zust = find(Lv_Zust); % Numerical Index Vector
for k = 1:numel(idx_Zust)-1
idxrng = idx_Zust(k) : idx_Zust(k+1); % Search Region: Heat Off Index(k) To Heat Off Index(k+1)
[Tmax(k,:),idx] = max(TT3r.Wert(idxrng)); % Maximum Temperature In Index Range
Tmaxidx(k,:) = idx_Zust(k) + idx; % Index (Used To Find Time)
DT_Temp(k,:) = TT3r.DateTime(Tmaxidx(k)); % Maximum Temperature Time
DT_Zust(k,:) = TT3r.DateTime(idx_Zust(k)); % Heat Off Time
end
elapsed_time = DT_Temp - DT_Zust;
Results = table(DT_Zust, DT_Temp, elapsed_time, Tmax, 'VAriableNames',{'Heat Off Time','Next Peak Temparature Time','Elapsed Time','Next Peak Temperature (°C)'})
I commented-out the (currently irrelevant) plots from my previous code, however I imported most of that code because that information is necessary here. The new file is a bit easier to work with. (If you want to extract the hours, minutes, and seconds from duration values, you will need to use the hms function. Regular datetime functions do not work with duration arrays.)
.
Plus de réponses (0)
Catégories
En savoir plus sur Operators and Elementary Operations dans Centre d'aide et File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
Translated by 
