Research On Dynamic Cooperative Target Tracking Based On Multi-Agent System

Target tracking is a key link in the process of regional monitoring,which is widely used in civil and military fields,such as military reconnaissance,regional security,threat assessment,and enemy situation analysis.It can provide effective reference for real-time strategic decision-making.Therefore,the study on target tracking is of great significance.In most cases,there may be multiple targets in the monitoring area,and there are multiple monitoring platforms.This may lead to two or more targets competing for a same tracking platform.Based on the target tracking,this thesis mainly studies the tracking platform scheduling.Tracking platform scheduling is a combinatorial optimization problem,that is,studying how to allocate the tracking platform between multiple targets in competition to improve the efficiency of system allocation or resource utilization.A good system model can not only veritably describe the elements of the system,while greatly facilitate the modeling and simulation.Based on the multi-Agent system modeling,the target prioritizing is studied.On this basis,the tracking platform scheduling algorithm is studied.The detailed content and innovations are outlines as follows:1.Research from the aspect of the system model,the MAS model is built for target and tracking platform based on the hypothetical scene.The MAS model can well characterizing the units' attributes and the relationship among them.The rationality is proved through the contrast between MAS model and tracking system on attributes,2.Research from the aspect of multi-target prioritizing,the clustering coefficient based tactical significance algorithm is proposed for multi-target prioritizing based on the clustering coefficient.The algorithm based on the actual scene takes not only the target's own attributes,the target-area relative situation but also the relative situation among respective targets into consideration.In this algorithm,the three kinds of influencing factors are constructed functions and weighted respectively,and the AHP algorithm is used to calculate the three parts weight,which improves the practicality and rationality of the algorithm.CCTSM reflects the nonlinear relationship between the target priority and the influencing factors.Compared with the classical TSM algorithm,it adds the target's own attribute factors and the relative situational factors between targets,so it considers the factors more comprehensive.CCTSM uses the method that weights and adds up the sub-functions of the influencing factors,which not only reduces the parameters to be selected,but also avoids the phenomenon that one factor determines the overall priority.The above advantages of CCTSM can effectively improve the rationality in target prioritizing under battlefield environment.3.Research from the aspect of improves the scheduling efficiency and ensuring the tracking accuracy for the target with high speed and high timeliness.A contract net combined with binary particle swarm optimization(CNP-BPSO)tracking platform scheduling algorithm is proposed based on multi-agent model.The fitness function is established to minimizing the scheduling time and the tracking error based on Agent model.By combining the contract net and the binary particle swarm optimization,the tracking accuracy is ensured while filter out the improper candidate agents.Thus,the search space is narrowed,the searching speed is improved and the system resource is saved.Compared with the existing several representative algorithms,the multi-target tracking platform scheduling algorithm CNP-BPSO effectively improves the scheduling efficiency and ensures the tracking accuracy,which is suitable for the target tracking with high real-time requirements.