Modeling,Simulation And Engineering Application Of Multi-agent System

In engineering projects, there usually a problem to search optimal or near-optimal solution within the solution space of the specific problem according some constraints and objectives. Compared with traditional optimization problems, searching algorithms for engineering application should has low computing complexity, adaptability to complex environment, and a good maintainability. This paper introduced a novel agent solution searching framework based on complex adaptive system (CAS). This agent senses information of local environment, and then chooses apt option rule to realize its intended goals. During the decision making updating, CAS agent changes fitness of the option rule according to a performance evaluation. In this way, the agent is adaptive to different environment. In addition, CAS agent is simple in structure and easy to expand. This paper consists the following content:A Multi-agent situation model using CAS agent. In this model, there are bottom-top feature extraction and situation pattern matching for information fusion. Situation representation and recognition in complex environment are realized using a distributive strategy. During the recognition, CAS agent compares reasoning results and environment information. In this way, fitness of its internal rule could be adjusted and complex partial input data could be handled.CAS agent searching is used for unmanned aerial vehicle (UAV) path planning in complex environment with a global Probabilistic A*guidance. The whole algorithm uses a global-local combination strategy to keep the agent from lossing in local optimization and promote safety and smoothness performance of the path. Under the global guidance, CAS agent could adjust its optional rule fitness and in this way adaptive to different environment situation.Besides, algorithm simulation is important to engineering application. This paper introduce a3-D dynamic simulation platform using MFC and OpenGL. Relevant algorithms in situation assessment and path planning were simulated and tested.This research is supported by the Basic Research Fund of Xidian University (No. JY10000902033).