复杂装备虚拟维修训练系统中数据场景图研究

针对传统场景图结构在开发虚拟维修训练系统时的局限性，提出了一种新的场景图数据结构。该场景图具有对象-表面-面片3层结构，并用四元组的数据结构封装了虚拟对象的属性信息、内部状态集、事件列表以及与事件相对应的行为。利用该场景图结构实现了表面间碰撞检测，并对发生碰撞的表面进行约束识别和约束求解。利用Petri网描述虚拟对象的状态及状态之间的转换，并将其转化为虚拟对象的规则集，从而确定其“事件-状态”之间的因果关系。该场景图结构弥补了商用软件开发虚拟维修训练系统的不足，实现了虚拟维修中表面间的碰撞检测，并利用虚拟对象的行为信息实现了复杂的维修操作。

Research on Scene Graph of Virtual Maintenance Training System of Armament

To remove the limitations of traditional scene graph structure in development of virtual maintenance training system (VMTS), a new scene graph structure is presented. The new scene graph structure is made up of objects, surfaces and polygons. The behavioral characteristics of virtual object, such as attribute information (A ), state (S), event list (E) and behavior (B), are repre?sented by 4-element structure く A, S，E9 B > . The collision between surfaces is implemented by us?ing of the presented scene graph, and constraints involved in collision detection could be recognized and solved. Transition between states of virtual object is described by PetrThet and it is translated into rules, thus causal relation between event and state is defined. Limitations of development of virtual maintenance training system (VMTS) with commercial software could be reduced by the new scene graph structure. Furthermore, collision detection between surfaces in VMTS is realized, and compli?cated maintenance operation could be implemented with characteristics of virtual object.