基于混合粒子群算法的信息物理融合系统多目标跟踪优化
发布时间：2019-03-09 17:38
【摘要】：信息物理融合系统(CPS)是融合计算、通信与控制于一体的复杂信息系统。多目标跟踪是CPS的重要应用领域,涉及多目标的信息获取、实时定位和跟踪预测等。为进一步提高多目标跟踪在定位的精度和路径预测方面的准确性,本学位论文在CPS多目标跟踪中引入混合粒子群算法,设计与实现多目标定位与路径预测的优化方案。本文首先设计CPS多目标定位模型,再提出基于混合粒子群算法的CPS多目标定位方法,接着设计CPS多目标跟踪模型,最后提出基于混合粒子群算法的CPS多目标路径预测方法。针对上述CPS多目标定位和路径预测的优化方案,本文还分别进行仿真实验与性能分析。本文的工作创新主要体现在以下两个方面:(1)针对CPS当前目标定位算法在处理约束优化问题时存在收敛速度慢和易陷入局部最优等缺点,在应用粒子群算法时,引入了交叉和变异策略,避免CPS目标在迭代过程中陷入局部最优。此外,将空间距离约束与几何拓扑约束作为CPS多目标定位模型中的目标函数,采用混合粒子群算法求得CPS多目标跟踪中的最优解,与标准粒子群定位算法相比较,上述方案减少了CPS多目标定位误差,并缩短了目标到达极值点的时间。(2)针对CPS当前多目标跟踪方法在对目标进行路径预测方面精度不足等缺点,在应用粒子群算法时,引入采样和重采样策略,将预测更新过后的粒子权值进行归一化处理,挑选和复制权值较大的种群粒子再次进行迭代,避免了粒子的退化现象。运用优化后的公式获取种群中的粒子在不同时刻的观察值,使得CPS中的目标粒子不停地朝实际状态慢慢逼近,解决了粒子在不断迭代时因粒子稀少而降低预估精度等问题。
[Abstract]:Information physical fusion system (CPS) is a complex information system which integrates computing, communication and control. Multi-target tracking is an important application field of CPS, which involves multi-target information acquisition, real-time location and tracking prediction. In order to further improve the accuracy of multi-target tracking in location and path prediction, this thesis introduces hybrid particle swarm optimization (PSO) algorithm into CPS multi-target tracking to design and implement the optimization scheme of multi-target location and path prediction. In this paper, we first design the CPS multi-target location model, then propose the CPS multi-target location method based on hybrid particle swarm optimization (HPSO), then design the CPS multi-target tracking model, and finally propose the CPS multi-target path prediction method based on the hybrid particle swarm optimization (HPSO) algorithm. Aiming at the optimization scheme of CPS multi-objective location and path prediction, the simulation experiment and performance analysis are also carried out in this paper. The innovation of this paper is mainly reflected in the following two aspects: (1) in view of the shortcomings of CPS's current target location algorithm in dealing with constrained optimization problems, such as slow convergence rate and easy to fall into local optimization, particle swarm optimization (PSO) is applied. The crossover and mutation strategies are introduced to avoid the local optimization of the CPS target in the iterative process. In addition, the spatial distance constraints and geometric topological constraints are regarded as the objective functions of CPS multi-target location model. The hybrid particle swarm optimization algorithm is used to obtain the optimal solution in CPS multi-target tracking, which is compared with the standard particle swarm localization algorithm. The above scheme reduces the error of CPS multi-target location and shortens the time of target reaching the extreme point. (2) in view of the shortcomings of the current CPS multi-target tracking method in path prediction, the particle swarm optimization algorithm is applied. The strategy of sampling and resampling is introduced to normalize the predicted and updated particle weights, and the population particles with larger weights are selected and replicated again, thus avoiding the degradation of the particles. The optimized formula is used to obtain the observed values of the particles in the population at different times, which makes the target particles in the CPS approach to the actual state ceaselessly, and solves the problem that the prediction accuracy of the particles in the iterative process is reduced due to the scarcity of the particles.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP29
本文编号：2437706
[Abstract]:Information physical fusion system (CPS) is a complex information system which integrates computing, communication and control. Multi-target tracking is an important application field of CPS, which involves multi-target information acquisition, real-time location and tracking prediction. In order to further improve the accuracy of multi-target tracking in location and path prediction, this thesis introduces hybrid particle swarm optimization (PSO) algorithm into CPS multi-target tracking to design and implement the optimization scheme of multi-target location and path prediction. In this paper, we first design the CPS multi-target location model, then propose the CPS multi-target location method based on hybrid particle swarm optimization (HPSO), then design the CPS multi-target tracking model, and finally propose the CPS multi-target path prediction method based on the hybrid particle swarm optimization (HPSO) algorithm. Aiming at the optimization scheme of CPS multi-objective location and path prediction, the simulation experiment and performance analysis are also carried out in this paper. The innovation of this paper is mainly reflected in the following two aspects: (1) in view of the shortcomings of CPS's current target location algorithm in dealing with constrained optimization problems, such as slow convergence rate and easy to fall into local optimization, particle swarm optimization (PSO) is applied. The crossover and mutation strategies are introduced to avoid the local optimization of the CPS target in the iterative process. In addition, the spatial distance constraints and geometric topological constraints are regarded as the objective functions of CPS multi-target location model. The hybrid particle swarm optimization algorithm is used to obtain the optimal solution in CPS multi-target tracking, which is compared with the standard particle swarm localization algorithm. The above scheme reduces the error of CPS multi-target location and shortens the time of target reaching the extreme point. (2) in view of the shortcomings of the current CPS multi-target tracking method in path prediction, the particle swarm optimization algorithm is applied. The strategy of sampling and resampling is introduced to normalize the predicted and updated particle weights, and the population particles with larger weights are selected and replicated again, thus avoiding the degradation of the particles. The optimized formula is used to obtain the observed values of the particles in the population at different times, which makes the target particles in the CPS approach to the actual state ceaselessly, and solves the problem that the prediction accuracy of the particles in the iterative process is reduced due to the scarcity of the particles.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP29
【参考文献】
相关期刊论文 前7条
1 冯刚;夏晨晖;孙磊;傅建中;;基于非线性预测的机床主轴温升特性快速辨识[J];农业机械学报;2015年06期
2 李周芳;肖乐;;基于混合粒子群算法的粮食应急调度系统中的路径规划[J];河南工业大学学报(自然科学版);2013年05期
3 肖迪;葛启承;林锦国;程明;;一种双种群遗传粒子群算法及在SMB优化中的应用[J];南京理工大学学报;2012年01期
4 李世威;王建强;曾俊伟;;一种模糊偏好排序的多目标粒子群算法[J];计算机应用研究;2011年02期
5 刘衍民;赵庆祯;牛奔;邵增珍;;基于ε占优的自适应多目标粒子群算法[J];控制与决策;2011年01期
6 尹安东;赵韩;;基于混合系统理论的混合动力城市客车控制策略研究[J];汽车工程;2010年02期
7 高鹰,谢胜利;基于模拟退火的粒子群优化算法[J];计算机工程与应用;2004年01期
，本文编号：2437706
本文链接：https://www.wllwen.com/kejilunwen/zidonghuakongzhilunwen/2437706.html
