May be the following links will help you.

Terminate and retrieve solution in ILOG after a certain time limit

https://stackoverflow.com/questions/24449622/terminate-and-retrieve-solution-in-ilog-after-a-certain-time-limit

DropSolve Cplex terminate after T time:

https://developer.ibm.com/answers/questions/385233/dropsolve-cplex-terminate-after-t-time/

How to stop CPLEX after the first feasible solution:

http://www-01.ibm.com/support/docview.wss?uid=swg21400089

As previously mentioned, there is a time limit parameter. However, if you use a small time limit Cplex might terminate the search without finding a solution.

If you only want to quickly find a feasible solution you can consider using the parameter "MIP integer solution limit". By setting this parameter to 1, Cplex will stop as soon as it finds a feasible solution. By setting it to 2, Cplex stops after finding two feasible solutions, and so on.

If you want some guarantees on the solutions quality, I would consider using the "MIP gap tolerance" parameter. Setting the mip gap tolerance between 1-10% will usually reduce the solution time significantly.

Hi,

have you considered to play with other CPLEX parameters? Setting values adapted for your problem might significalntly reduce solution time, but experimenting of course takes a lot of time. If you have to solve the same type of problem often it might be worth it.

- MIP emphasis - default 0 is good for many problems, you might try 4 to see if it helps for your problem

- heuristic frequency - try e.g. 5

- RINS heuristic frequency - time-expensive heuristics, but might help a lot by finding new incumbents and thus reducing B&B tree size by more cutoffs, try e.g. 20

Of course every test with a reasonable small time limit.

It might also help to have CPLEX start the run with one set of settings (like emphasis 0 with small heurfreq and rins freq) with a rather big mipgap (like 1e-3), then switch to different settings (like automatic heurstics frequencies and emphasis 4), reduce the mipgap e.g. to 1e-8 and continue the run using the state of the B&B tree.

The tune tool tries a few other settings, but for my difficult problems it didn't work out really well - and it took a very long time.

Thank you very much for the help you have given

The following website may be help you.

https://www.ibm.com/support/knowledgecenter/SSSA5P_12.6.2/ilog.odms.cplex.help/CPLEX/Parameters/topics/TiLim.html

I use cplex.timelimit in Matlab API. Others could be found in the table for details.

for matlab2019 & cplex12.8 , the following works:

options = cplexoptimset('cplex');

options.timelimit= 1000 ; % second,

options.mip.tolerances.mipgap= 0.005 ; % 0-1 default 0.0001

[x, fval, exitflag, output] = cplexmilp ( )