基于SVM和K-均值聚类的RBF神经网络短时交通流预测-控制科学与工程专业论文.docxVIP

摘 要 交通出行在日益繁忙的工作生活中愈发显得重要，如何解决道路拥挤，减少交 通事故发生率，已经成为世界各国亟待解决的复杂难题。短时交通流的预测作为应 对 上 述 问 题 的 方 案 ， 因 为 其 具 有 实 时 准 确 等 特 点 ， 已 经 被 当 作 智 能 交 通 系 统 （Intelligent Transport Systems，ITS）的主要组成部分。 本文首先介绍了 K-均值聚类算法的基本理论，它的特点是简单快速，同时应用 广泛。在处理密集样本的时候，同时样本中的类彼此线性可分，那么运用 K-均值聚 类算法所得到的聚类效果就会很出色。在 RBF 神经网络的支持下，对短时交通流进 行预测，得到了比较理想的数据。 支持向量机（Support Vector Machine，SVM）作为机器学习方法中的佼佼者， 它能够出色的解决许多算法中的实际问题，比如 K-均值聚类算法很容易陷入局部极 小值。通过引入支持向量机，我们能够在短时交通流的预测上，得到更加精确的结 果。实验结果说明，利用基于支持向量机的 K-均值聚类算法，RBF 神经网络能够有 效的预测短时交通流，证明该模型是有效的。 关键词：短时交通流预测；K-均值聚类算法；支持向量机；RBF 神经网络 Abstract Traffic travel in an increasingly busy work life all the more important, how to solve road congestion, reduce accident rates, the world has become a complex problem to be solved. Short-term traffic flow forecasting as a response to the above problem solution because it has the characteristics of real-time and accurate, has been used as a major component of the intelligent transportation system. This paper introduces the basic theory of K- means clustering algorithm, which is characterized by simple and fast, while widely used. Time-intensive sample handling, and sample each other linearly separable classes, then use the clustering effect K- means clustering algorithm obtained will be very good. In support of RBF neural network, the forecasting of traffic flow, has been ideal data. Support vector machine as the machine learning methods in the crowd, it can solve many excellent algorithms practical problems, such as K- means clustering algorithm is very easy to fall into local minima. By introducing support vector machines, we are able to predict traffic flow on, to get more accurate results. Experimental results demonstrate that the use of SVM K- means clustering algorithm, RBF neural network can effectively predict the traffic flow, show that the model is valid. Key words: short-term traffic flow forecasting; K- means clustering algorithm; Support Vector Machines; RBF neural network 目 录 第一章 绪论 ........................