【作者】 刘利强；

【导师】 袁赣南；

【作者基本信息】 哈尔滨工程大学 ， 导航、制导与控制， 2008， 博士

【摘要】 潜艇导航规划是实现潜艇自动化、智能化航行的关键技术之一,其研究目的在于如何使潜艇更快更好的完成上层指令规定的任务,按照安全航行、隐蔽航行、快速航行的优化目标自动规划出由当前点到指定目标位置的最优航线。从算法执行过程来看,导航规划主要由环境建模和路径优化搜索两部分组成。因此,导航规划问题可以归结为优化搜索问题。蚁群算法是20世纪90年代初提出的一种群智能优化算法。其优越的问题分布式求解模式在组合优化问题的求解中取得了极大成功,引起了相关领域学者的广泛关注。实践已经证明,蚁群算法能够很好的解决复杂非线性条件下的多约束优化问题。本文主要针对蚁群优化方法及其在潜艇导航规划中的应用进行研究,论文主要研究工作如下:1.概述了蚁群算法的基本原理,对ACS算法的值收敛性和解收敛性进行了详细分析,证明了标准ACS算法能够值收敛,但无法实现解收敛。通过对标准ACS算法进行改进,选择适当的信息素值下限函数,可以同时实现ACS算法的值收敛和解收敛。2.给出了状态转移策略的一般表示形式,提出了蚁群算法中选择函数、选择概率、选择强度的概念,并设计了三种选择函数,就不同选择函数对蚁群算法性能的影响进行了理论和仿真分析。提出了一种基于粒子群优化的蚁群算法参数选取方法,将蚁群算法的参数选取看作一个优化问题,使用粒子群优化算法对其进行迭代寻优。该方法能够方便有效地实现蚁群算法参数的优选,有利于蚁群算法的应用和推广。3.基于提高离散域蚁群算法寻优能力和寻优速度两方面考虑,分别提出并设计了协同多蚁群伪并行优化算法和空间收缩蚁群优化算法。协同多蚁群伪并行优化算法使用采用不同算法实例模型的多个子蚁群并行独立的构建问题的解,通过信息交互的方式综合各子蚁群信息素矩阵上累积的历史经验信息,保证信息素分布的指导性和多样性,提高了算法的寻优能力和稳定性。空间收缩蚁群优化算法在运行过程中通过不断的整合信息素浓度较高的构造块,从而缩小解的组成成分集合和构造块集合的规模,加快了算法运行速度。4.通过对蚂蚁觅食过程中蚁群的分布与食物源的关系进行深入的分析,提出了一种基于蚁群觅食行为的改进连续域优化算法。设计了算法解的表达形式、信息素分布模型、状态转移策略、信息素更新规则和约束条件处理方法,并就算法参数对算法性能的影响进行了定性分析。对一类无约束和一类有约束基准测试函数的求解结果表明,该算法具有收敛速度快和全局寻优能力强的优点。5.针对基于蚁群优化的潜艇三维空间导航规划的实现方法及其关键技术进行了深入研究。分别以空间收缩蚁群优化算法和基于蚁群觅食行为的改进连续域优化算法为基础,设计了潜艇三维空间导航规划算法。获得的两种导航规划算法不仅能够灵活地规划出具有不同特点的优化路径,而且可以方便容易地处理各种约束,具有较强的搜索能力,能够很好的解决潜艇三维空间导航规划问题。更多还原

【Abstract】 Submarine navigation planning is one of the key technologies in Submarine automation and intelligent navigation. The aim of its research is how to make submarine accomplish the mission from upper command fasterly and better; according to the optimization objective of safe, hidden, fast navigation, planning automatically optimum route from recent point to intended target point. In the process of algorithm execution, the navigation planning mainly composese by two parts, environment modeling and path optimization search. Thus, the problem of navigation planning can be regarded as an optimization search.Ant colony optimization algorithm is a swarm intelligence optimization algorithm proposed in the early 1990s. The superiority of distributed solution model of problem in solving combinatorial optimization problem is very great, and this causese scientists’ large attention in concerned fields. The practice has shown that, ant colony optimization algorithm can well solve problems of multi-constraint optimization under complicated nonlinear conditions. This paper is mainly about ant colony optimization algorithm and its application in submarine navigation planning, including:1. The basic principle of the ant colony optimization algorithm is summarized, and the convergence of value and solution for the standard ACS algorithm is analysed in detail. It can not prove the ablitiy of the solution convergence, but the value convergence can. By improving the standard ACS algorithm, choosing appropriate lower function of pheromone value, we can get both of them.2. The general representation of state transition strategy is given. We also propose three concepts of ant colony algorithm: selection function, selection probability and selection intensity, then design three selection functions. To each function, theory and simulation analysis on influence of performance of ant colony algorithm have been done. A parameter selection method of ant colony algorithm is proposed. Based on particle swarm optimization, it regards parameter selection as an optimization problem, then iterates particle swarm optimization algorithm for optimization. This method can realize the optimization of ant colony algorithm and is good for its application and popularization.3. In order to inproving the optimization ability and speed in ant colony algorithm in discrete domain, we design two algorithms: virtual parallel ant colonies optimization algorithm based on cooperative multiple ant colonies and ant colonies optimization algorithm based on space contraction. In the former one, many children ant colonies adopted different case algorithm model consist the problem solution in concurrent and independence. It can ensure the instructive and diversity of pheromone distribution by synthetizing experience information from each pheromone matrix of children ant colonies through information interaction, and it can improve the optimization ability and stability. In the latter one, it can reduce the solution compositions and the scale of structure block set by continually integrating the structure block with thick pheromones in execute process, and it can work faster.4. Based on deep analysis of relationship between ant colony distribution and food source in ant foraging process, an improved ant colony algorithm in continuous domain based on ant foraging behavior is proposed. We design the representation of algorithm solution, pheromone distribution model, state transition strategy, pheromone update rules and processing method of constraint conditions. Then, influence of algorithm performance to parameters is qualitatively analyzed. The results of one kind of benchmark test function with or without constraints show that, this algorithm is converged faster and has strong global optimization ability.5. According to the deep research of way and key technologies of submarine navigation planning in three dimensional spaces based on ant colony optimization, we design submarine navigation planning algorithms in three dimensional space, which is respectively based on space contraction ant colonies optimization algorithm and ant colony algorithm in continuous domain. These two algorithms for navigation planning not only can plan neatly optimization path with different characters, but also can deal with different constraints. It has srong searching ability and can work out the problem of submarine navigation planning in three dimensional spaces perfectly.更多还原