Multi-Agent Scheduling System Modeling And Simulation Based On CPN

In this information age when market environment changes rapidly and dynamically, the key for operational success is whether the corporation could make prompt response to these changes and take corresponding actions at the First Opportunity. Consequently, it has become the most urgent and crucial task for many companies to establish Internet-oriented Enterprise Environment Scanning System to monitor and collect relevant web data.The major functions of Enterprise Internet Environment Scanning System can be summarized as: visit customer-assigned web within the scope of physical resources available in the system, study and monitor website update rate, repeatedly crawl the website at the frequency approaching but lower than the update rate, extract the updated information, thus enable the enterprise to have a good command of latest business situation, and to dig out the hidden knowledge behind the phenomena through analyzing the continuously varying data.Nevertheless, the rapid augment of data on the Internet causes that the extent, depth and speed of system information extraction cannot satisfy the requirements of practical applications. With regards to this problem, this paper proposes the multi-level, distributed system architecture, which enhances system performance through strengthening system's parallel capacity. The paper explains the priority-based Task Scheduling strategy for central control Agent in the cross layer and scheduling Agent in the internal layer. Meanwhile, in order to fulfill the special demand of repeatedly gathering updated information, the classical Min-min Heuristic Scheduling Algorithm is assimilated to design the Recycling Task Earliest-Finished Scheduling Algorithm, which empowers the scheduling Agent to appropriately assign sub-task to the Agent who is able to finish the task at the first time.To validate the effectiveness of system structure and scheduling algorithm before building the real system, the paper adopts Petri Nets, a graphical mathematical tool, to build the system model, test its verification and validity, and analyze its performance. The paper primarily accomplishes how to utilize the Colored Petri Nets (CPN) modeling application, CPN Tools, to establish the scheduling system model based on Hierarchical Timed Colored Petri Nets (HTCP-net): using CPN Tools to build triple-layer system architecture model with its support function of hierarchical network; disassembling the gathering task through the definition of color set and the declaration of CPN ML functions, and achieving the priority-based scheduling strategy and the Recycling Task Earliest-Finished Heuristic Scheduling Algorithm. The comparison with the control experiment verifies the validity of the conclusions from theoretic research in the paper and the feasibility of proposed model, and also provides important reference for the following research.