无人机路径规划方法研究及在油田巡井中的应用

发布时间：2018-03-12 13:43

  本文选题：无人机路径规划　切入点：粒子群算法　出处：《东北石油大学》2017年硕士论文　论文类型：学位论文

【摘要】：无人机巡检技术作为一门新兴技术,近年来在环保、通信、电力、气象等多个领域得到了广泛的推广和应用。无人机路径规划是无人机巡检的关键问题。随着油田产量和规模的不断提升,传统的人工巡井方式已经不足以满足油田的管理需要,因此结合油田行业的特点,研究油田巡井无人机路径规划方法具有重要意义。无人机路径规划是根据任务需要规划从出发点到目标点满足约束条件的飞行路径,是无人机研究的重要内容之一。本文在油田巡井的实际背景下,围绕无人机路径规划方法进行了研究,主要内容如下:首先,研究无人机路径规划问题及环境建模。结合无人机油井巡检过程中的飞行环境和任务要求,确定无人机路径表达形式和约束条件。根据无人机定高飞行,将三维飞行环境抽象为虚拟平面截成的二维空间,采用几何描述法将油井表达为平面坐标特征点,将障碍物表达为平面多边形,利用高斯-克吕格投影法对油井经纬坐标进行无角变处理,建立环境模型。其次,确定无人机路径规划算法,并对其进行改进。通过对遗传算法、蚁群算法和粒子群算法的比较分析,选择粒子群算法作为无人机路径规划算法。针对粒子群算法的不足,提出一种混合蛙跳粒子群算法。在算法搜索前期,利用混合蛙跳算法的分组策略分割初始种群,以局部深搜索思想优化次优个体,提取各层次个体作为新种群,提高算法搜索效率;在算法搜索后期,对最优个体进行三重交叉操作,同时引入基于疏密性的引导变异操作,对稀疏点以较大概率变异,提高粒子多样性。通过TSP标准测试库数据,将改进算法与其他算法进行规划性能比较。然后,研究无人机避障问题。针对已知障碍,在障碍物模型的基础上设计基于计算几何的障碍检测方法,并根据障碍检测结果将障碍威胁系数引入到路径规划算法的评价函数中,使生成的路径满足避障要求;针对突发障碍,利用人工势场法规划生成局部避障路径,躲避突发障碍物。最后,以大庆某油田巡井问题为实例,规划无人机巡井路径。针对巡井环境模型,采用混合蛙跳粒子群算法和基于计算几何的避障方法规划初始最优参考路径,并进行路径平滑,使生成的路径同时满足巡井遍历、路径最短和避障要求;采用人工势场法生成局部动态路径,保证局部路径能够规避突发障碍威胁。给出无人机巡井路径规划的基本步骤,并将本文方法的路径规划结果与其他算法的路径规划结果进行比较。
[Abstract]:As a new technology, UAV patrol and inspection technology has been used in environmental protection, communication, electric power in recent years. Many fields such as meteorology have been widely popularized and applied. UAV path planning is the key problem of UAV inspection. With the increasing production and scale of oil field, the traditional manual well inspection method is not enough to meet the needs of oil field management. Therefore, considering the characteristics of oilfield industry, it is of great significance to study the path planning method of oil well patrol UAV. UAV path planning is to plan the flight path from starting point to target point to meet the constraint conditions according to mission needs. It is one of the important contents of UAV research. In this paper, the path planning method of UAV is studied under the actual background of oil field survey. The main contents are as follows: first of all, The problem of UAV path planning and environment modeling are studied. According to the flight environment and mission requirements of UAV oil well inspection, the UAV path expression form and constraint conditions are determined, and the UAV altitude flight is determined according to the UAV. The 3D flight environment is abstracted into a two-dimensional space cut by a virtual plane, and the oil well is expressed as a plane coordinate feature point by the geometric description method, and an obstacle is expressed as a plane polygon. Gauss-Kruger projection method is used to deal with the warp and weft coordinates of oil wells, and an environment model is established. Secondly, the path planning algorithm of UAV is determined and improved. Comparing and analyzing the ant colony algorithm and particle swarm optimization algorithm, the particle swarm optimization algorithm is selected as the path planning algorithm of UAV. In view of the deficiency of particle swarm optimization algorithm, a hybrid leapfrog particle swarm optimization algorithm is proposed. The initial population is segmented by the grouping strategy of hybrid leapfrog algorithm, the sub-optimal individuals are optimized by the idea of local deep search, and the individuals at all levels are extracted as new populations to improve the search efficiency of the algorithm. Triple crossover operation is carried out on the optimal individual, at the same time, the guided mutation operation based on density is introduced, and the sparsity point is mutated with a large probability to improve the particle diversity. Through the TSP standard test library data, the particle diversity is improved. The improved algorithm is compared with other algorithms in planning performance. Then, the obstacle avoidance problem of UAV is studied. Aiming at the known obstacles, a new obstacle detection method based on computational geometry is designed on the basis of obstacle model. According to the result of obstacle detection, the obstacle threat coefficient is introduced into the evaluation function of the path planning algorithm to make the generated path meet the requirements of obstacle avoidance, and for the sudden obstacle, the artificial potential field method is used to plan and generate the local obstacle avoidance path. Finally, taking a well inspection problem in Daqing oilfield as an example, planning the patrol path of UAV, aiming at the survey environment model, The hybrid leapfrog particle swarm optimization algorithm and the obstacle avoidance method based on computational geometry are used to plan the initial optimal reference path, and the path is smoothed so that the generated path can meet the requirements of well patrol, shortest path and obstacle avoidance simultaneously. The artificial potential field method is used to generate the local dynamic path to ensure that the local path can avoid the threat of sudden obstacles. The results of path planning of this method are compared with those of other algorithms.
【学位授予单位】：东北石油大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：V249;TE938.2
，

本文编号：1601815