【作者】 史春雪；

【导师】 卜英勇；

【作者基本信息】 中南大学 ， 机械设计及理论， 2011， 博士

【摘要】 海底钴结壳是一种生长在水深500-3000m海底壳状物,它富含钴、铂等稀贵金属,还含有锰、镍、铁、铜等金属,是一种重要的矿产资源。随着陆地资源的日趋枯竭,深海钻结壳作为一种极具商业开采价值的海生矿产资源,获得了美国、德国、日本、俄罗斯等西方国家的重视。随着面向钻结壳勘探开采研究工作的日渐深入,部分国家已经步入试开采阶段。与西方发达国家相比,我国对钴结壳的勘探开采研究工作起步较晚。为维护我国的海洋权益,开辟我国新的矿产资源来源,面对国际社会对海洋资源争夺的形势,有必要开展钴结壳的探测采集技术与装备研究。本文在国务院大洋专项的支持下,是“钴结壳采集模型机关键技术及装备研究”项目的一项重要内容。作者在查阅国内外大量相关文献的基础上,对海底采矿车路径规划问题相关的海底矿区环境建模问题、海底大尺度遍历路径规划问题、海底采矿车静态路径规划问题和海底采矿车动态路径规划问题进行了深入并系统的研究。主要的研究成果如下：1)对高度非结构化,含不同类型底质的海底环境,提出了一种针对不同类底质的环境建模方法。利用先验的含底质类属性的DEM数据,提取地形几何特征,得到了海底环境四维混合属性数据；通过模糊推理的方法,获得不同类底质地形的通行性指数；并通过设置综合通行性代价函数,对不同底质地形的通行性进行有效整合,得到了环境可通行性地图。为路径规划研究提供了模型基础,并通过实验验证了算法的有效性。2)提出一种矿区大尺度遍历路径规划方法。首先设置遍历路径规划的性能评价函数,通过对评价函数的计算,确定平坦地形往复式采集的方式；后用Boustrophedon方法对矿区环境进行子区域划分；为可采子域之间建立综合连通距离矩阵,将子域连接问题转化为TSP问题,通过蚁群算法求解,达到最大覆盖率的优化目标；对于非相邻可采子域之间的局部路径搜索问题,将其转化为SPP问题,通过Floyd算法求解,满足了最小重复率的要求；最后提出矿区大尺度遍历路径规划算法。通过仿真,验证了算法的可行性。3)提出基于改进蚁群算法的采矿车静态路径规划方法。首先指出遍历路径中存在连接路径和采集路径两种路径；之后本着实时性要求对基本蚁群算法进行改进,对环境模型膨化后,提出了采矿车静态路径规划的改进蚁群算法,并证明了算法的收敛性；依据两种路径的不同要求,分别设置不同的启发函数和适应度函数,提出两种静态路径规划的改进蚁群算法；通过仿真,验证了两种算法的可行性。4)提出了基于改进滚动窗口法的采矿车动态路径规划方法。首先提出算法；之后按照A*算法的思想,确定子目标选取的方法；证明了算法的收敛性；并且证明了滚动规划的全局次优性,且行走优化系数γ的设置能够提高算法的优化性能；通过仿真验证了算法的可行性。5)对采矿车在线路径规划系统进行实车实验,通过实验过程及结果分析,证明该系统是可行的。更多还原

【Abstract】 Deep-sea cobalt-rich crust is a kind of resource grown on seabed in500-3000m deep, which contain many rare metals such as cobalt, platinum, and also contain manganese, nickel, iron, copper and other metals. It is a kind of important mineral resource. With rapidly reducing of resources on land, as a kind of marine mineral resource with extremely commercial value to exploit, deep-sea cobalt-rich crust is catching attentions of USA, Germany, Japan, Russia and other western countries. Some countries have entered into trial mining stage with gradual deepening research on mining and exploration of cobalt-rich crust. Compared with western developed countries, In China, Research on cobalt-rich crust mining and exploration was carried out later. To safeguard the marine rights of China, and open up new mineral resource in our country, and face the situation of striving marine resources in international community, it is necessary to carry out research on cobalt-rich crust mining, exploration and equipment.This paper is supported by item DY105-03-02(Research on Key Technology and Equipment of Cobalt-rich Crust Mining Model Vehicle). After consulting a large quantity of related research papers and technique reports, the author is carried out thorough and systemic study on deep-sea world modeling problem, deep-sea large-scale traversal path planning problem, deep-sea static path planning problem and deep-sea dynamic path planning problem. Main research results are as follows:1) A kind of world modeling model with different types of bottom materials in highly unstructured seafloor environment was proposed. DEM data with bottom material class attributes detected in prophase was utilized, by extracting terraingeometrical characteristic, four-dimensional mixed attributes data of seafloor environment was obtained. Terrain trafficability index with different class was got by using the fuzzy reasoning method. By setting up comprehensive trafficability cost function, terrain trafficabilities in seafloor environment with different bottom materials were normalized, and then the world trafficability map was obtained. This model could provide a modeling foundation for study on path planning. The efficiency of the method was verified by and simulations experiments.2) A large scale traverse path planning method in mineral domain was put forward. Firstly traverse path planning performance evaluation function was set. Through the calculation of evaluation function, the way of reciprocating mining method was determined. The world model was divided into some subdomains by using Boustrophedon method. Comprehensive connection distance matrix for subdomains was set up, then the problem of connection among subdomains could be turned into TSP. After solved the problem through ACO, the optimal objective of maximum coverage was achieved. Transfered local path searching problem among no-adjacen subdomains into SPP, and then the SPP was sloved by Floyd algorithm, so that the minimum repeat rate requirement was met. Finally the large-scale mining traverse path planning algorithm was put forward. The simulations proved the feasibility of the presented algorithm.3) A improved ACO for mining vehicle static path planning was presented. Firstly, that traverse path is composed of two kinds of paths such as connecting path and mining path was pointed out. After expanding the world model, basic ACO was improved to meet mining real-time demand. The improved ACO was proposed and the convergence of the algorithm was proved. According to different demands of the two paths, different heuristic functions and fitness functions was set up, and then the two improved ACOs for static path planning were proposed. Through simulations, the feasibilities of the two algorithms were verified.4) A improved rolling window method for mining vehicle dynamic path planning was proposed. By using the ideology of A*algorithm, the method of choosing sub-aim points was determined. The convergence of the algorithm was proved. Then the global sub-optimality of rolling planning was proved, and that the optimal performance of the algorithm could be improved by y was also proved. The simulations show the feasibility of the algorithm. 5) Experiments were carried out to test and verify mining vehicle online path planning system, and the system was proved feasible by analysis of the experiments results.更多还原