| Popis: |
®is a well-known simulation environment for modeling and simulation of dynamic systems. In several situations such as battlefield scenarios, it is possible to represent such systems from an agent-centric point of view. This raises several modeling and simulation issues as to how to best represent such them inside Simulink. To manage the complexity in the system, we model the system from multiple viewpoints. The highest level is the process view gives information about the operational specifics such as flow metrics of the various agents in the system. The mid-level view is the network view where the various agents are coupled due to imposed constraints on them. For example, in a traffic simulation, the roadway system serves as the coupling network for the various vehicles. The lowest level view gives agent-based view consisting of behavioral models for different agent types. The challenge in realizing such systems is in associating with each agent-event driven, state transition and continuous time dynamics information. The scheme that we use for realizing this synchronization is a centralized database with appropriate filtering for the various agents. Using several Simulink modeling constructs, we introduce a design pattern for associating discrete entities with corresponding discrete-event, continuous-time and state transition model. We enable the simulation of the above system with customizable interactions through hybrid simulation using both event and time-driven execution semantics. As scalability and extensibility are chief concerns for modeling such systems, we also provide best practices on how to scale the type and number of agents. At the network view level, we show how arbitrary network topologies of multiagent system can be implemented using fundamental blocks. Since the comprehension of the dynamics of these agents in a geometric space can be difficult, we also outline a MATLAB ® scripting framework whereby it is possible visualize such systems realistically in 3D. The case study is in this paper is a simplified airport runway management system that embodies the above concepts. Aircraft runway traffic is modeled as entities in SimEvents ® , a discrete event tool available in Simulink. The aircraft dynamics with pilot behavior is modeled in Simulink while the collision detection system of each aircraft is modeled using statecharts in Stateflow ® . The scheduler for routing aircraft in taxiway network based on assigned priorities on the various aircraft. To create the taxiway topologies, we introduce a novel modeling technique for defining these network paths using fundamental building blocks in SimEvents. The dynamic creation and deletion of objects are managed using a MATLAB S-function in conjunction with Simulink 3D Animation™. |
