Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Mathematical optimization wikipedia , lookup
Theoretical computer science wikipedia , lookup
Natural computing wikipedia , lookup
Computer simulation wikipedia , lookup
Computational fluid dynamics wikipedia , lookup
Agent-based model wikipedia , lookup
Perceptual control theory wikipedia , lookup
Computational electromagnetics wikipedia , lookup
Control theory wikipedia , lookup
Dzung Du M.S., Computerized Control System, Kiev Institute of Civil Aviation Ph.D. candidate, Aeronautics and Astronautics Engineering, Purdue University Fields of Interests: Numerical Methods for Optimal Control Problems Hybrid Systems Multi-Agent Simulation of the National Airspace System Approximation of Special Mathematical Functions Contacts: Room 337B Potter Building, Purdue University West Lafayette, 47906 Email: [email protected] Office phone: (765)-496-6633 Numerical Methods for Optimal Control Problems: We are developing a computational approach which is i) computationally efficient ii) can handle a broad class of optimal control problems. Switched systems are a very specific class of hybrid systems that consist of several modes (described by differential equations) and switching laws governing the active mode. So far, we have successfully applied this method for switched systems (with both internally and externally forced switching). It is our intent to use the technique to handle more general classes of optimal control problems, like optimal control with maximax or minimax cost functionals, optimal control processes governed by integral equations, etc. Also, the multidimensional version of the algorithm could be applied for solving these PDE like the Hamilton-Jacobi-Bellman PDE (in optimal feedback control), the Hamilton-Jacobi-Isaacs PDE (in dynamic games and reachability analysis for hybrid control systems). Multi-Agent Simulation of the Air Traffic Control System: We are developing a agent-based testbed that called SimUAV. In agent-based simulation, each of the entities in the National Airspace System, like aircraft, air traffic controller, weather factors, navigation and surveillance systems, communication/datalinks, etc. can be modeled as an agent. The performance of the whole system is created by the captured interaction among the agents. Our main aim in building the testbed is: the model should be flexible enough for the purpose of verification and testing of new automation tools and system concepts, such as conflict detection/resolution algorithms, free flight concepts,…etc.