【作者】 奚茂龙；

【导师】 须文波；

【作者基本信息】 江南大学 ， 轻工信息技术与工程， 2008， 博士

【摘要】 群体智能是指无智能的主体通过合作表现出智能行为特性的系统,群体智能在没有集中控制且不提供全局模型的前提下,为寻找复杂问题的解决方案提供了基础。群体智能算法是基于群体行为对给定的目标进行寻优的启发式搜索算法,其寻优过程体现了随机、并行和分布式等特点。群体智能算法的典型代表是1995年Kennedy和Eberhart提出的粒子群优化(Particle Swarm Optimization,PSO)算法。PSO算法自提出以来,由于其计算简单、易于实现、控制参数少等特点,引起了国内外相关领域众多学者的关注和研究。J.Sun等在深入研究PSO算法单个粒子收敛行为的基础上,受量子物理学的启发提出了具有量子行为的粒子群优化(Quantum-behaved Particle SwarmOptimization,QPSO)算法,QPSO算法具有控制参数更少,收敛速度快,全局搜索能力强等特点。本文以QPSO算法为基础,提出了几种不同搜索策略的改进QPSO算法;针对离散空间优化问题,提出了二进制编码的具有量子行为的粒子群优化(Binary QPSO,BQPSO)算法,对其进行了深入的分析和研究;并研究了群体智能算法在移动机器人路径规划和轨迹跟踪控制中的应用方法,具体内容如下:(1)阐述了群体智能优化算法及移动机器人路径规划和轨迹跟踪控制的研究背景;介绍了两种典型的群体智能算法的研究现状,即蚁群优化(ACO)算法和PSO算法;对移动机器人路径规划和轨迹跟踪控制的设计方法作了详细的介绍;提出了本课题的研究思路和方法。(2)为了进一步提高QPSO算法在解决多峰优化问题中全局搜索能力,增加算法后期群体的多样性,提高算法在大范围内进行搜索的能力,提出了在群体中采用随机选择最优个体的操作策略以避免群体的多样性过小,从而提高算法的全局搜索能力,通过对标准测试函数的求解表明改进算法的全局求解能力得到了提升。(3)详细分析了智能群体的决策机制,发现在智能群体决策过程中,个体粒子参与决策的权利根据个体的优劣程度是不相同的,提出了在QPSO算法中引入线性权重算子和精英选择算子进一步提高QPSO算法的搜索效率及优化性能。两种改进算法通过在标准测试函数中的求解显示了较好的性能,其中引入线性权重算子的改进方法取得了较为明显的优势。(4)针对离散空间优化问题,给出了二进制编码的QPSO算法的设计思路,提出了BQPSO算法的进化方程。通过泛函分析的方法和随机过程方法分析了BQPSO算法的收敛性,得出全局收敛的结论。对多个测试函数的求解结果显示了BQPSO算法相对于二进制编码的PSO(BPSO)算法的优越性。(5)算法参数是影响算法性能和效率的关键,文中对BQPSO算法的控制参数的取值方式作了系统的研究,提出了该参数的三种控制策略,即固定取值策略、线性取值策略和自适应取值策略,通过对测试函数的求解分别研究了这三种控制策略,得出了具有指导意义的结论。分析了BQPSO算法的多样性控制方法,基于群体多样性的度量方式,提出了采用重新初始化最优平均值和变异最优个体的方式对群体操作避免群体的多样性过小,从而提高算法的性能。(6)分析了移动机器人路径规划的不同技术方法,提出了BQPSO算法和栅格技术相结合的机器人路径规划方法,并利用多个设计实例来验证该方法的设计效果,仿真结果显示了该方法能够在工作空间中规划出适合机器人运动的无碰撞路径。(7)分析了移动机器人轨迹跟踪控制的两种控制器设计方法。根据反演设计方法设计了轨迹跟踪控制器;分析了滑模变结构控制的基本原理,采用指数趋近律和幂次趋近律相结合的方法,设计了新的滑模跟踪控制律,并使用PSO算法、QPSO算法和改进QPSO算法分别优化了反演跟踪控制器和滑模跟踪控制器中的参数,使用多个实例验证了优化后的跟踪控制器的设计效果;通过对设计效果的分析和比较,表明了两种跟踪控制器都能够控制机器人实现对既定轨迹的跟踪,仿真结果显示QPSO算法及改进算法能够在轨迹跟踪控制器的参数优化中取得更好的优化效果。(8)使用一个综合实例验证了BQPSO算法结合栅格法应用于移动机器人路经规划,改进QPSO算法应用于移动机器人轨迹跟踪控制器设计中的有效性和可行性,为移动机器人路径规划和轨迹跟踪整体设计提供了一种新的方法。论文最后对所做工作与主要研究成果进行了总结,并提出了进一步的研究方向。更多还原

【Abstract】 Swarm Intelligence(SI) is the property of a system whereby the collective behaviors of agents interacting locally with their environment.SI provides a basis with which it is possible to explore collective problem solving without centralized control or the provision of a global model.SI algorithm is a kind of heuristic search method that can solve the specified problems based on collective behaviors.The characteristic of SI algorithm is stochastic,parallel and distributed.Particle Swarm Optimization(PSO) algorithm is one of the typical SI algorithms which is developed by Kennedy and Eberhart in 1995.Since PSO algorithm was developed,it has attracted many researchers in the fields concerned as its characteristics of simple computation,easy realization and few parameters.J.Sun proposed Quantum-behaved Particle Swarm Optimization(QPSO) algorithm based on the deep study of PSO algorithm and inspired by quantum physics.QPSO algorithm has much fewer parameters and much stronger global search ability than the PSO algorithm.Several improved QPSO algorithms with different search strategies are proposed base on QPSO algorithm.QPSO algorithm by binary encoding(BQPSO) is proposed for solving discrete problems,and is given deep analysis and study on it.The applications of SI in the path planning and tracking control of mobile robot are also studied in the work.The main contents of this dissertation are as follows:(1) Research background of SI algorithms and path planning,tracking control of mobile robot are expatiated.The current research situations of two typical SI algorithms are detailed introduced,which are Ant Colony Optimization(ACO) algorithm and PSO algorithm.An introduction of design of path planning and tracking control of mobile robot is presented.Research methods and ideas in the work are proposed.(2) In order to maintain the diversity of the population and enhance global search ability of the quantum-behaved particle swarm optimization,an improved QPSO algorithm with the best individual of random selection is proposed to avoid the swarm diversity getting into a low level.The improved algorithm shows preferable ability in solving the multi-modal problems.(3) Decision-making rights are different according to the individuals’ fitness values by the analysis of SI decision-making mechanism.Improvements of QPSO algorithm are proposed to improve searching efficiency and optimal performance by introducing linear weighted operator and elitist selection operator into algorithm.Two improved QPSO algorithms show good performance by tested them on the benchmark functions and the QPSO algorithm with linear weighted operator is better.(4) The thought of BQPSO is discussed and evolution equations are given which are completely different from the QPSO algorithm.The convergence of BQPSO algorithm is analyzed through functional analysis and stochastic process method,and conclusion of global convergence is drawn.BQPSO algorithm shows better performance than BPSO algorithm in solving test functions.(5) The parameter of an algorithm is the key issue that affects the algorithm’s performance and efficiency.The methods for taking value of control coefficient in BQPSO algorithm are analyzed systemically.Three strategies are proposed,including:setting the parameter as a fixed value,making the parameter take value linearly of nonlinearly according to the iteration,and letting the value adaptively according to the evolution results.Some guiding conclusions are summarized.An improvement on BQPSO algorithm is proposed to avoid the swarm diversity getting a low level based on diversity measure,the improvement is realized by resetting average best value of all particles and mutating the swarm’s global best particle.(6) Path planning method of robot that is combined BQPSO algorithm and grids technologies is proposed by base on the analysis of different path planning technologies.The design result of this method is validated by a number of design examples.Simulation results show that collision-free paths can be planned by this method in robot working space.(7) Two controller design methods of robots trajectory tracking are analyzed.Trajectory tracking controller is designed according to back-stepping approach.The basic tenet of sliding mode is analyzed and new sliding mode control law is designed by combining index reaching law and power reaching law.The parameters in back stepping tracking controller and sliding mode tracking controller are optimized through PSO algorithm,QPSO algorithm and improved QPSO algorithm.The design result of optimal tracking controller is validated by a number of design examples.It shows that two tracking controllers can control robot to track planned trajectory.Simulation results show that QPSO algorithm and its improvement can get better effect in optimizing parameters of controllers.(8) An integrated example shows that the BQPSO algorithm applied in path planning and improved QPSO algorithm applied in design of tracking controller is effective and feasible.It proposes a new method for the design of integrated system with path planner and tracking controller.The main contributions in this work are summarized at last and further research considerations and put forward.更多还原