Archived OpenModelica forums. Posting is disabled.

Alternative forums include GitHub discussions or StackOverflow (make sure to read the Stack Overflow rules; you need to have well-formed questions)


Forgot password? | Forgot username? | Register
Page Start Prev 1 Next End

Long executable execution time

#10103  Permalink Top

Long executable execution time

Jun-12-20 15:10:03

Hi,
I have a problem with execution of optimization problem in OMNotebook.

To begin with I'm working with a model in which I'm using ipopt in order to optimise two variables over time. It's a typical control problem with charging and discharging of energy storage.

I've got the model working but a strange thing happens after it's built with higher number of intervals.

In this situation the program executes very fast (5 s or less ) when it is done by optimize(model,....).

Code:

optimize(Mors_opt_src, stopTime=t_f, tolerance=1e-8, numberOfIntervals=500, simflags="-optimizerNP 1 -ipopt_max_iter 500 -s optimization");

After that when I call model.exe it takes ages to compute (5 min or so).

Code:

system("Mors_opt_src.exe");

In order to make the model.exe work at all i've had to add "C:\OpenModelica1.13.0-32bit\lib\omc" to path.

The model reads two variables from txt files by:

Code:

Modelica.Blocks.Sources.CombiTimeTable Zapotrzebowanie(columns = {2}, extrapolation = Modelica.Blocks.Types.Extrapolation.HoldLastPoint, fileName = "F:\xxx\xxx\xxx\xxx\xxx.txt", smoothness = Modelica.Blocks.Types.Smoothness.LinearSegments, tableName = "tab", tableOnFile = true);

Can anyone tell me what may cause the problem? I wanted to use the exe file in optimal control system In this process execution time is crucial and I would rather not have to build the model each time.

I'm using Win10 and I've tried OM 1.13 and 1.14 on both 32 and 64-bit versions. Also running the whole thing on Linux is not a bang for a buck for me. (At least not at this moment.)

The problem doesn't occur with small number of intervals. With

Code:

numberOfIntervals=50

but the thing is that i will be dealing with large number of those.

Also the process uses less than 13% of cpu time,

Thanks for any help



Edited by: adamwoj01 - Jun-12-20 15:10:54
#10203  Permalink Top

Re: Long executable execution time

Aug-27-20 15:29:24

So if anyone finds this topic one day. I've worked around the problem in maybe not the most elegant, but working way.

Instead of calling the model.exe file with modified Model_init.xml file I've written the optimization.mos script which loads models and runs optimization with overriden parameters.

Code:


setCommandLineOptions("+g=Modelica");

loadModel(Modelica,{"3.1"}); //Without older modelica library I get some errors



loadFile("model1_path");

loadFile("model2_path");



setCommandLineOptions("+gDynOpt");

t_f := 604800;

optimize(model_name, stopTime=t_f, tolerance=1e-9,numberOfIntervals=504, outputFormat="csv", simflags="-optimizerNP 1 -ipopt_max_iter 2000 -s optimization -lv=LOG_SOLVER -overrideFile=C:\path_path_path\parameters_mod.txt");

getErrorString();



//optimization -lv=LOG_SOLVER is for analysing the process

//override file has model parameters and input file paths in it

Than I've written a batch script caller.bat in which I change the directory to a path for the built/compiled files and output and from there I call the OM complier with the script as an argument.

Code:


cd  C://output_path_output_path

C://OpenModelica1.13.0-64bit//bin//omc.exe C://pathpathpath//optimization.mos

pause  // for taking a look at the console response

That way I can call multiple instances of the optimization with different input simultaneously in multiple threads from python or another batch file.

I still have to deal with compilation time but, I don't get the problem mentioned above.

Page Start Prev 1 Next End
Previous topic
Next topic
There are 0 guests and 0 other users also viewing this topic
You are here: