毕业设计（论文）_粒子群优化算法及改进的比较研究.docVIP

粒子群优化算法及改进的比较研究 摘要 粒子群优化（Particle Swarm Optimization，PSO）算法是一种优化计算技术，由 Eberhart 博士和Kennedy博士提出，它源于对鸟群和鱼群群体觅食运动行为的模拟。PSO算法是一种基于迭代的优化工具，系统初试化为一组随机解，通过迭代搜寻最优解，粒子在解空间中追随最优的粒子进行搜索。它的主要特点是原理简单、参数少、收敛速度较快、易于实现。目前，粒子群优化算法应用于神经网络的训练、函数优化、多目标优化等领域并取得了较好的效果，有着广阔的应用前景。 但就其本身而言，在理论和实践方面还存在很多不足之处。粒子群优化算法根据全体粒子和自身粒子的搜索经验向着最优解的方向发展，在进化后期收敛速度变慢，同时，算法收敛精度不高，尤其是对于高维多极值的复杂优化问题。 论文的主要工作有： （1）对研究PSO算法相关基础知识进行回顾，主要是优化问题和群体智能。对粒子群优化算法的理论基础和研究现状作了简要介绍，分析了粒子群优化算法的原理和算法流程。 （2）分析粒子群算法的生物模型和进化迭代方程式，粒子速度概念不是必需的，粒子移动速度不合适反而可能造成粒子偏离正确的进化方向，因此提出了只基于“位置”概念的简化粒子群算法。粒子群收敛于局部极值的根本原因在于进化后期没有找到优于全局最优的位置，对个体极值和全局极值进行随机扰动，提出了带极值扰动的粒子群优化算法。两种策略结合，提出了带极值扰动的简化粒子群优化算法。 （3）简要介绍了粒子群优化算法在整定PID参数中的应用。 关键词：粒子群优化算法；粒子速度；极值扰动 Comparative Study on Several Improved Particle Swarm Optimization Algorithms ABSTRACT Particle Swarm Optimization（PSO）originally introduced by Doctor Eberhart and Kennedy is an optimization computing technology which derived from imitating the bird and fish flock’s praying behavior. It is a kind of optimization tool based on iterative computation. System initializes a group of random solution，then it searches the optimal solution through iteration ,and particles follow the optimal particle to run search in the solution space. The main trait of PSO is simple in principle，few in tuning parameters，speedy in convergence and easy in implementation．Now, PSO is used for training of neural networks，optimization of functions and multi-target and it obtains good effect, its applied foreground is very wide． In itself, there are still a lot of defect in theory and practice．PSO develop towards the optimal solution’s direction depending on all the particles and its own particle’s search experience. In the later evolution, its convergence velocity becomes slower. Meanwhile, its convergence precision is not high especially for the complex high dimensional multi-optima optimization problems. The main works of the dissertation can be summarized as follows： (1)Reviewed some basic knowledge that