S. L. Chung, S. Lafortune, and F. Lin. Supervisory control using variable lookahead policies. Discrete Event Dynamic Systems: Theory and Applications, 4(3):237--268, July 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....sequences conservative and optimistic. In [6] the authors have further studied how to use previous computations to help in the next computation of the control action. Effect of taking an undecided attitude to help improve the computational complexity has been investigated by the authors in [7], on line computational technique in which no constraint is imposed on the depth of the lookahead window has been investigated in [9] assuming that the additional state information is also available. The application to the case of partial observation has been considered in [11, 10] 2 Lookahead ....

....sublanguage is of the same order as that of computing a supremal controllable sublanguage. In the prefix closed case, the computational complexity is also affected by the complexity of K legal . Finally, there are many possible ways that this work presented here could be extended: 1. As in [6, 7, 9], it is important to develop algorithm to perform recursive computation of supRC( Delta; Delta) from one N level tree to another as the limited lookahead windows roll through the execution of each event. 2. For some application areas, the representations of K legal and L complete may be too ....

S. L. Chung, S. Lafortune, and F. Lin. Supervisory control using variable lookahead policies. Discrete Event Dynamic Systems: Theory and Applications, 4(3):237--268, July 1994.

....non probabilistic languages is a complete upper semi lattice [22] and the set of probabilistic languages is a complete partial order [7] non unique maximally permissive supervisors do exist. We present an e ective algorithm for on line computation of a maximally permissive supervisor (refer [9, 2, 8] for other on line supervisory computation algorithms) The computational complexity of each step is again O(n 3 ) where n is the product of the number of states in the automata representations of the plant and the lower bound speci cation. We also show that a unique minimally permissive ....

S. L. Chung, S. Lafortune, and F. Lin. Supervisory control using variable lookahead policies. Discrete Event Dynamic Systems: Theory and Applications, 4(3):237-268, July 1994.

....of probabilistic languages is a complete partial order [7] non unique maximally permissive supervisors do exist. An off line computation of a maximally permissive supervisor need not be feasible. We present an effective algorithm for on line computation of a maximally permissive supervisor (refer [9, 2, 8] for other on line supervisory computation algorithms) The computational complexity of each step is again O(n 3 ) where n is the product of the number of states in the automata representations of the plant and the lower bound specification. We also show that a unique minimally permissive ....

S. L. Chung, S. Lafortune, and F. Lin. Supervisory control using variable lookahead policies. Discrete Event Dynamic Systems: Theory and Applications, 4(3):237-- 268, July 1994.

....state set of the generator for the legal language, Sigma is the plant s event set, and Sigma c is the subset of events that are controllable. We present an online supervisor here that is of complexity O(jXjj Sigmaj) The VLP PO algorithm can also employ a variable lookahead technique proposed in [17] that reduces the computation time at the cost of possibly generating a non maximal language. Furthermore, in [14] a scheme of event ordering is given that ensures that the generated maximal language includes supCCN(L) Similarly, as remarked in [14] 3 , another event ordering scheme exists to ....

S.-L. Chung and S. Lafortune, "Supervisory control using variable lookahead policies," in Proceedings of the American Control Conference, (San Francisco, CA), pp. 1203--1208, June 1993.

....and is modelled by a finitely recursive process) the loop cannot be identified and therefore the back propagation is not efficient. Only if the existence of the loop is known a priori from the model, a computable result can be derived. However, this is consistent with the same assumption given in [6]. It is worth mentioning here that the complexity of the back propagated algorithm depends on the number of the events of the model and on the complexity of the associated conditions. 6 Combination of results Now suppose the alphabet of P contains both controllable uncontrollable events and ....

S. Chung, S. Lafortune, and F. Lin, "Supervisory control using variable lookahead policies", Discrete Event Dynamic Systems: Theory and Applications, pp. 237--268, 1994.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC