1: \begin{abstract}

2: When facing time-variant problems in analog computing,

3: the desirable RNN design requires finite-time convergence

4: and robustness with respect to

5: various types of uncertainties,

6: due to the time-variant nature and difficulties in implementation.

7: %

8: It is very worthwhile to explore

9: terminal zeroing neural networks,

10: through examining and applying

11: available attracting laws.

12: %

13: In this paper, from a control-theoretic point of view,

14: an error recurrence system  approach is presented

15: by equipping with uncertainty compensation

16: in the pre-specified error dynamics,

17: capable of enhancing robustness properly.

18: %

19: Novel rectifying actions

20: are designed to make finite-time settling so that the convergence speed and the computing accuracy of time-variant computing can be improved.

21: Double-power and power-exponential rectifying actions are respectively formed to construct specific models,

22: while the particular expressions of settling time function for the former

23: are presented, and for the latter the proximate settling-time estimations are given,

24: with which the fixed-time convergence of the

25: corresponding models is in turn established.

26: %

27: Moreover, the uncertainty compensation

28: by the signum/smoothing-signum techniques are adopted for finite-duration stabilization.

29: %

30: Theoretical results are presented

31: to demonstrate effectiveness (involving

32: fixed-time convergence and

33: robustness)

34: of the proposed computing schemes for the time-variant QP problem solving.

35: \end{abstract}

36: