改进的粒子群算法及其在聚类算法中的应用

发布时间：2018-05-23 13:01

本文选题：粒子群算法 + 数据分析　；参考：《广东工业大学》2017年硕士论文

【摘要】：最优化方法是研究给定约束条件下如何使某一(或某些)指标达到最优的一门学科,而优化算法研究一直是该领域研究的关键问题.粒子群算法是优化算法中一个参数简单且效果出众的算法,它结合个体学习经验和社会经验调整粒子的进化方向,从而获得最优解.在互联网快速发展的今天,每天产生的数据量急速增加,数据规模从TB跃升到PB甚至EB;数据类型多且数据结构复杂,处理难度增加.目前大数据的处理和分析技术越来越受到政府和企业的关注.而大多数数据挖掘算法的本质基本上都是建立优化模型,并用最优化方法对目标函数(或损失函数)进行优化,以确定最优解.本文对优化算法进行研究,针对粒子群算法容易早熟收敛和陷入局部最优解的问题,提出一种改进的粒子群算法.并将改进后的粒子群算法应用到K-means聚类算法与大数据处理平台应用中.本文的主要工作如下:首先针对粒子群算法容易早熟收敛和陷入局部最优解的缺点,利用远离个体最差经验和最差群体经验,提出一种远离最差解的粒子群算法,并进行了仿真实验,验证算法具有良好的全局收敛性.其次将改进后的粒子群算法并行化在Spark集群上编程实现.Spark平台是目前应用最广的大数据分析平台,支持Java、Scala、Python和R等多种语言,能够无缝结合Hadoop平台等.最后将改进后的粒子群算法应用到K-means聚类算法中,对Iris和Wine数据集进行了仿真实验,实验结果较好,并将其应用到电信定位楼群中,对所得到的所属楼群用户MR信息进行聚类,聚类后提取簇间无线基站接入特征作为学习特征,以期后来无线接入特征相同或相似的MR定位到所属楼宇.
[Abstract]:Optimization method is a discipline to study how to achieve the optimal performance of one or some indexes under a given constraint condition, and the research of optimization algorithm has always been the key problem in this field. Particle swarm optimization (PSO) is an algorithm with simple parameters and excellent effect. It adjusts the evolutionary direction of particles by combining individual learning experience and social experience to obtain the optimal solution. With the rapid development of Internet, the amount of data generated every day increases rapidly, the scale of data leaps from TB to PB or even EB.There are many types of data and complicated data structure, which makes processing more difficult. At present, the processing and analysis technology of big data is paid more and more attention by the government and enterprises. The essence of most data mining algorithms is to establish the optimization model and optimize the objective function (or loss function) with the optimization method to determine the optimal solution. In this paper, the optimization algorithm is studied, and an improved particle swarm optimization algorithm is proposed to solve the problem that particle swarm optimization is easy to converge prematurely and fall into local optimal solution. The improved particle swarm optimization algorithm is applied to K-means clustering algorithm and big data processing platform. The main work of this paper is as follows: firstly, aiming at the shortcomings of particle swarm optimization (PSO), which is easy to converge prematurely and fall into local optimal solution, a particle swarm optimization algorithm is proposed, which is far from the worst individual experience and the worst group experience. The simulation results show that the algorithm has good global convergence. Secondly, the improved particle swarm optimization algorithm is parallelized on Spark cluster to realize the Spark platform, which is the most widely used big data analysis platform at present. It supports Java Scala Python, R and other languages, and can seamlessly combine Hadoop platform and so on. Finally, the improved particle swarm optimization algorithm is applied to the K-means clustering algorithm, and the simulation experiments on the Iris and Wine data sets are carried out. The experimental results are good, and the improved PSO algorithm is applied to the telecom location-oriented buildings. After clustering, the access feature of wireless base station between clusters is extracted as the learning feature, so that the Mr with the same or similar wireless access features can be located to the building later.
【学位授予单位】：广东工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18;TP311.13

【参考文献】