基于多源定位方法的工程机械臂精确姿态感知研究

发布时间：2018-04-12 07:25

本文选题：无线传感器网络 + NLOS　；参考：《湘潭大学》2017年硕士论文

【摘要】：无线传感网络(WSNs)定位问题的规模与非视距误差是影响定位算法精度的主要原因,尤其是在三维的定位问题中,目标函数的非凸性是主要难点,半正定松弛方法是目前较为流行且在视距情况下能很好解决非凸问题的优化方法之一,但随着目标节点数目的增多,其优化模型中的约束条件数目也随之增多,大大增加了算法的计算复杂度,而且约束的构建需要随着环境的不同而改变。当测距环境处在非视距(NLOS)场景中时,松弛方法仍然是目前定位算法中主要的优化方法,但其定位精度下降,且计算复杂度增高。对于具体的应用来说,一般存在可利用的先验信息,比如工程机械臂架的结构特征、运动特征,若能较好运用这些先验信息,性能必然提高。因此,研究在现有方法上,如何较好地利用定位场景或者待测目标的先验信息,对于提高定位算法的性能大有帮助,尤其是具有鲜明结构特征和运动特性的工程机械臂更加具有很好的应用前景。本文在传统主要定位算法与优化方法充分调研的基础上,基于最优欧氏距离阵(EDM)的多源定位算法对工程机械臂姿态感知问题进行建模,建立工程机械臂(EMA)坐标系,并针对其结构设计欧氏距离阵,对其结构特征进行约束设计,构造了一套臂架姿态感知的新方法。其次,将此算法应用到泵车臂架姿态感知问题中来验证此算法,根据泵车固有运动特征,建立了泵车臂架坐标系,并对原算法进一步改进,设计了一套坐标降维方法,使三维模型降到二维,提高算法的定位精度并且降低计算复杂度。根据定位出的浇筑目标点,运用泵车臂架各连杆与水平夹角间的关系,设计了求取构建运动学方程的关键参数连杆间夹角的方法。最后,对本文提出的基于最优欧氏距离阵的定位算法及泵车臂姿态感知方法进行算法理论分析与仿真实验分析。与传统方法和其它相似方案比较,本文算法使欧氏距离阵更接近真实解,并且将目标函数改为求解基于原问题的对偶问题的罚函数,提高了算法的精度并降低了计算法复杂度,从理论上分析了该算法的优越性。再利用结构特征与运动特征进一步提高算法性能。仿真结果中,均方误差降低了45.92%,说明该算法可行且具有一定的实用性。
[Abstract]:The scale and non-line-of-sight error of WSNs localization problem in wireless sensor networks are the main reasons that affect the accuracy of the localization algorithm, especially in the three-dimensional localization problem, the non-convexity of the objective function is the main difficulty.The positive semidefinite relaxation method is one of the most popular optimization methods which can solve the non-convex problem well under the condition of visual distance. However, with the increase of the number of target nodes, the number of constraints in the optimization model also increases.The computational complexity of the algorithm is greatly increased, and the construction of constraints needs to change with different environments.When the ranging environment is in the non-line-of-sight (NLOS) scene, the relaxation method is still the main optimization method in the current localization algorithm, but its localization accuracy is reduced and the computational complexity is increased.For specific applications, there are generally available prior information, such as the structural characteristics and motion characteristics of the boom of construction machinery. If these prior information can be used well, the performance will be improved.Therefore, in the existing methods, how to make good use of the prior information of the location scene or the target to be tested is of great help to improve the performance of the localization algorithm.Especially the engineering manipulator with distinct structure and motion characteristics has a good application prospect.In this paper, based on the research of traditional main positioning algorithms and optimization methods, the multi-source positioning algorithm based on Euclidean distance array (EDM) is used to model the attitude sensing problem of engineering manipulator, and the EMA-coordinate system is established.The Euclidean distance matrix is designed for its structure, and its structural characteristics are constrained, and a new method of attitude sensing is constructed.Secondly, the algorithm is applied to the attitude perception of the boom of the pump car to verify the algorithm. According to the inherent motion characteristics of the pump car, the coordinate system of the boom frame of the pump car is established, and the original algorithm is further improved, and a set of coordinate dimension reduction method is designed.The 3D model is reduced to 2 D, which improves the accuracy of the algorithm and reduces the computational complexity.Based on the positioning of the pouring target point and the relationship between the connecting rod and the horizontal angle of the boom of the pump car, the method of calculating the angle between the key parameters of kinematics equation is designed.Finally, the theoretical analysis and simulation analysis of the positioning algorithm based on the optimal Euclidean distance matrix and the attitude sensing method of the pump arm are carried out in this paper.Compared with the traditional method and other similar schemes, the proposed algorithm makes the Euclidean distance matrix closer to the real solution, and changes the objective function into a penalty function for solving the dual problem based on the original problem, which improves the accuracy of the algorithm and reduces the computational complexity.The superiority of the algorithm is analyzed theoretically.The performance of the algorithm is further improved by using the structural features and motion features.The simulation results show that the mean square error is reduced by 45.92, which shows that the algorithm is feasible and practical.
【学位授予单位】：湘潭大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN929.5;TP212.9;TP241

【相似文献】