Research On Object-Oriented Algebra Petri Net Model

With the expanding of the computer system scale, the system structure are getting more and more complicated. And concurrency and distribution are considered the basic characteristics of the computer software and hardware systems. Many systems, such as Web-based application systems, electrical systems, embedded systems, and real-time systems, all have the characteristics of concurrency and distribution, because their functions are embodied by the subsystems'reactions between each other and their environments. Therefore, it is very difficult to deign such a system using the traditional analysis and design methods,How to analyze, model and design such systems has become the priority and difficulty in the study of software engineering. Agent is a software entity which can achieve its function in a certain environment continuously and spontaneously by associating to the relative agents and processes. As a new complex software system development technology, software agent is greatly concerned by the academics and businesses, and an increasing number of application areas began to construct software systems using software agent technology. As a result, software agent technology become gradually a key and mainstream technology in the areas of computer and automation, and is considered an important computing paradigm.Two description methods are mainly used in the design and development of the multi-agent systems, that is, non-formal and formal methods. Although the non-formal method is simple, accurate and clear, it can not descript the multi-agent system and its key features accurately, and can not analyze the systems as well, for the reason of its lack of precise semantics. Conversely, formal method, which can descript the multi-agent systems accurately, can help set up a bridge between theoretical study and engineering practice, and reduce the potential errors in the early system development. Hence, multi-agent technology can lower the development costs and improve software quality. Researches have indicated that the lack of formal specification is the major reason impeding large-scale multi-agent applications, and the absence of the prototyping developing methods and tools based on multi-agent software engineering methods is the root caused that the Agent technology is not fully applied in practice. However, the Formal modeling methods, which mainly concern the functions and data of a specific system, are difficult in describing the aspects of timing, control and behavior of a certain problem. Furthermore, this method can not be easily used for the reason that it uses mathematical symbols to describe the problems.For the reasonable and efficient collaboration between the Agents, the Agent structure and the coordination mechanisms between the Agents are also the priorities and difficulties in the development of the formal software based on multi-agent system.In order to analyze and model complex systems more efficiency to put the multi-agent technology into practice, this paper carried out an in-depth study of the current software engineering methods, and then proposed a model of Object-Oriented Algebra Petri Nets (OOAPN) by integrating the popular Object-Oriented technology, the simple and easy-to-use structured method, Petri nets technology and multi-Sorted algebraic theory together. And the following three issues are thoroughly discussed in this paper: the micro and macro levels of agent structures, the coordination mechanism among the agents and the formal development methods for multi-agent systems. This paper includes the following three aspects:1. We proposed a model of Object-Oriented Algebra Petri Net (OOAPN) and carried out an in-depth studied on it. The main contributions are: (a) the integration of multi-type algebra theory, Petri net model, structural theory and Object-Oriented technology lays the foundation for the agent structure construction and the formal modeling of multi-agent system; (b) the behavior semantic of OOAPN model is defined; (c) the formal Object-Oriented performance capability of OOAPN model is described; (4) the modeling methods for complex systems based on OOAPN model are given; (5) the formal analysis method of OOAPN model is perfected.2. An intelligent scheduling model for multi-agent system coordination and its relative algorithm are proposed to solve the coordinating problem among the agents in the multi-agent system. The main contributions are: (a) a multi-task dynamic scheduling architecture, which uses event-driven program as its framework and Petri nets as its event processor, is established. And an event-driven multi-agent system coordinated scheduling model is also given; (b) an event-driven multi-agent coordinated intelligent scheduling algorithm is worked out, and then demonstrated the effectiveness of this algorithm through examples. 3. To facilitate the engineering practice, the agent structure based on Object-Oriented Algebra Petri Nets is given. Then, the multi-agent system modeling steps is also illustrated based on the discussion of the present software system modeling methods. The key features are: (a) studied the OOAPN based multi-agent systems at to levels, that is, macro and micro levels. At the micro level, the behavior and structure of the single agent used in multi-agent system is defined, and at the micro level, the formal framework for OOAPN based multi-agent system is proposed; (b) defined the concept of communication event, and discussed the interaction among the agents, as well as the dynamic structure and behavior of the OOAPN based multi-agent system.; (c) proposed the analysis and modeling methods for multi-agent systems.