基于典型事例推理的路径规划方法研究

Case-based reasoning is an AI technique in which the previous solutions are stored for future use. People are used to guiding themselves according to those routes that are stored in their memories and have been used by them before. It is just based on people’s preference to familiar routes, which are gained through the study of the cognitive activities. We propose to apply the intelligent method based on the case reasoning to path planning. It is impossible for a case base to store all the solutions to all the shortest paths; therefore, part of them should be stored. However, which routes should be stored and which should not be? How do we adapt the cases that have already been stored and how do we acquire the shortest route based on them? All these issues need to be explained by integrating knowledge of the network on account of case-based reasoning techniques. This paper suggests the case-based reasoning in another point. This means finding some irreplaceable links on the basis of the complete analysis of the problems space, which are called the must_be_passed link between the source and destination. Merely compute the shortest path case from those best exit/entry nodes of the grids to the irreplaceable links, and then add them into the case base storing for future use. This method is based on case-based reasoning technique and completely considers the properties of the problem space. In addition to the use of knowledge of the natural grid in the route network, this method is more efficient than existing algorithms on computing efficiency.     

空间方向关系粗糙推理

针对现有的空间方向关系表达模型不能捕获其所固有的不精确性和模糊性这一缺点，把粗糙集理论和方法引入方向关系的表达，提出了方向关系粗糙表达方法、变精度粗糙表达方法以及方向关系的粗糙推理方法，以加强对不精确性和模糊性的处理和分析能力，并能够把模糊对象和精确对象之间的方向关系统一到一个框架中来描述。     

Computational complexity of spatial reasoning with directional relationship

The property of NP-completeness of topologic spatial reasoning problem has been proved. According to the similarity of uncertainty with topologic spatial reasoning, the problem of directional spatial reasoning should be also an NP-complete problem. The proof for the property of NP-completeness in directional spatial reasoning problem is based on two important transformations. After these transformations, a spatial configuration has been constructed based on directional constraints, and the property of NP-completeness in directional spatial reasoning has been proved with the help of the consistency of the constraints in the configuration.     

An optimum vehicular path algorithm for traffic network based on hierarchical spatial reasoning

Human beings’ intellection is the characteristic of a distinct hierarchy and can be taken to construct a heuristic in the shortest path algorithms. It is detailed in this paper how to utilize the hierarchical reasoning on the basis of greedy and directional strategy to establish a spatial heuristic, so as to improve running efficiency and suitability of shortest path algorithm for traffic network. The authors divide urban traffic network into three hierarchies and set forward a new node hierarchy division rule to avoid the unreliable solution of shortest path. It is argued that the shortest path, no matter distance shortest or time shortest, is usually not the favorite of drivers in practice. Some factors difficult to expect or quantify influence the drivers’ choice greatly. It makes the drivers prefer choosing a less shortest, but more reliable or flexible path to travel on. The presented optimum path algorithm, in addition to the improvement of the running efficiency of shortest path algorithms up to several times, reduces the emergence of those factors, conforms to the intellection characteristic of human beings, and is more easily accepted by drivers. Moreover, it does not require the completeness of networks in the lowest hierachy and the applicability and fault tolerance of the algorithm have improved. The experiment result shows the advantages of the presented algorithm. The authors argued that the algorithm has great potential application for navigation systems of large-scale traffic networks.     

不确定拓扑关系模糊推理

针对定性空间推理难以处理不确定空间拓扑关系推理的缺憾,将确定拓扑关系复合表推理方法扩展到模糊拓扑空间,建立了基于拓扑关系复合表的不确定拓扑关系模糊推理模型,设计了基于RCC5的拓扑关系复合表的模糊推理规则库和推理的方法流程,并结合具体计算实例对模糊推理模型进行了实验研究,推理结果验证了模型的认知合理性和可行性。     

邻接凸多边形方向关系计算及其推理

在对拓扑邻接凸多边形方向关系计算特殊性分析的基础上,提出了一个邻接凸多边形方向关系计算算法和二维空间方向关系推理方法。与Voronoi图所表达的邻近拓扑关系相结合,该算法可以用于计算离散二维空间目标之间的方向关系     

地理空间可视分析及其研究方向综述

介绍了可视分析的概念和内涵,分析了可视分析与可视化、地理空间可视分析等概念的关系。研究了地理空间可视分析的概念和内涵,在此基础上提出了地理空间可视分析的5个重要研究方向:面向可视分析的地理空间数据转换;多维地理空间数据的深层可视化表达和分析;地理空间可视分析推理;地理空间数据的协同可视分析和决策;地理空间数据可视分析工具的设计和评估。     

地震危险性与核电厂密闭可靠性的评价方法

本文就考虑地震危险性时，核电厂安全壳的防泄露可靠性的评价方法进行了理论研究，探讨了地震模型问题中的参数确定的一些方面，给出了可靠性评价的总体思路、分析方法及具体步骤，确定了要考虑的不定参数及其描述，并进而利用概率断裂力学推导出可靠性计算的一般公式。     

不确定海洋环境中基于贝叶斯理论的多声源定位算法

环境参数失配导致定位性能大幅度下降是匹配场定位所面临的难题之一。应用贝叶斯理论对环境聚焦,是当前解决该难题的研究热点。环境聚焦方法的实质是将未知环境参数和声源位置联合优化估计,当出现多个目标时,估计的参数会随着声源个数成倍增加,因此不得不利用有限的观测信息来实现众多参数的估计。本文采用最大似然比方法,获得信号源谱和误差项的最大似然估计,实现这些敏感性较弱参数的间接反演,有效降低了反演参数维数和定位算法复杂度。针对遗传算法的早熟和稳定性差的问题,改进了似然函数的经验表达式。将多维后验概率密度在参数起伏变化范围内积分,得到反演参数的一维边缘概率分布,求解最优值的同时进行反演结果的不确定性分析。本文仿真了位于相同距离、不同深度的两个声源,使用仿真实验验证了提出算法的有效性。     

基于T-S模型不确定时滞系统的保成本控制

针对一类基于T-S 模糊模型描述的具有输入与状态时滞的不确定系统,研究了保成本控制问题,得到了保成本控制律存在的充分条件,并以线性矩阵不等式的形式给出了控制器的设计方法.最后仿真实例验证了该方法的可行性.