🚀 go-pugleaf

  Stability problem in NARMA identification
  -----------------------------------------

I'm doing a nonlinear system identification of the follow process:
	Y[n]=f{Y[n-1],Y[n-2],....X[n],X[n-1],...}
In the learning phase of open-loop I can reach avery small error criteria.
From the open loop learning I can see that the system is learning on the
wright direction, when I close the loop I have stability problem.
  I'll be interested to hare any ideas or  solution to this problem (articles
and books).

			thanks
				Pini Tandetnik