H.3.9. Problem Solving, Control Methods, and Search
Zahra Jahan; Abbas Dideban; Farzaneh Tatari
Abstract
This paper introduces an adaptive optimal distributed algorithm based on event-triggered control to solve multi-agent discrete-time zero-sum graphical games for unknown nonlinear constrained-input systems with external disturbances. Based on the value iteration heuristic dynamic programming, the proposed ...
Read More
This paper introduces an adaptive optimal distributed algorithm based on event-triggered control to solve multi-agent discrete-time zero-sum graphical games for unknown nonlinear constrained-input systems with external disturbances. Based on the value iteration heuristic dynamic programming, the proposed algorithm solves the event-triggered coupled Hamilton-Jacobi-Isaacs equations assuming unknown dynamics to develop distributed optimal controllers and satisfy leader-follower consensus for agents interacting on a communication graph. The algorithm is implemented using the actor-critic neural network, and unknown system dynamics are approximated using the identifier network. Introducing and solving nonlinear zero-sum discrete-time graphical games in the presence of unknown dynamics, control input constraints and external disturbances, differentiate this paper from the previously published works. Also, the control input, external disturbance, and the neural network's weights are updated aperiodic and only at the triggering instants to simplify the computational process. The closed-loop system stability and convergence to the Nash equilibrium are proven. Finally, simulation results are presented to confirm theoretical findings.
M. Shokohi nia; A. Dideban; F. Yaghmaee
Abstract
Despite success of ontology in knowledge representation, its reasoning is still challenging. The most important challenge in reasoning of ontology-based methods is improving realization in the reasoning process. The time complexity of the realization problem-solving process is equal to that of NEXP Time. ...
Read More
Despite success of ontology in knowledge representation, its reasoning is still challenging. The most important challenge in reasoning of ontology-based methods is improving realization in the reasoning process. The time complexity of the realization problem-solving process is equal to that of NEXP Time. This can be done by solving the subsumption and satisfiability problems. On the other hand, uncertainty and ambiguity in these characteristics are unavoidable. Considering these requirements, use of the Fuzzy theory is necessary. This study proposed a method for overcoming this problem, which offers a new solution with a suitable time position. The purpose of this study is to model and improve reasoning and realization in an ontology using Fuzzy-Colored Petri Nets (FCPNs).To this end, an algorithm is presented for improve the realization problem. Then, unified modelling language (UML) class diagram is used for standard description and representing efficiency characteristics; RDFS representation is converted to UML diagram. Then fuzzy concepts in Fuzzy-colored Petri nets are further introduced. In the next step, an algorithm is presented to convert ontology description based on UML class diagram to an executive model based on FCPNs. Using this approach, a simple method is developed to from an executive model and reasoning based on FCPNs which can be employed to obtain the results of interest by applying different queries. Finally, the efficiency of the proposed method is evaluated with the results indicating the improve the performance of the proposed method from different aspects.
