what to do when lsqnonlin gets stuck

I am running a monte carlo parameter optimization routine that calls lsqnonlin using random starting points selected with bounded parameter values. In addition, lsqnonlin is set to keep the search parameter values within the same bounds used to select the initial starting point. I am fitting the model to real data and in some conditions, the process runs well. Using other data sets, however, the process may get stuck after a few tens or even hundreds of trails and I am wondering if there is a way to detect when the search is going nowhere so I could abandon that trial so that a new, perhaps better starting point might be tried to allow the monte carlo process to continue. Thanks for any advice.