基于优化神经网络的混合网络流量预测模型仿真与实现

发布时间：2018-04-17 04:09

本文选题：网络流量建模 + 流量预测　；参考：《北京交通大学》2017年硕士论文

【摘要】：现代社会网络在信息化社会中具有越来越重要的作用。通过互联网可以进行高效率的通信与沟通,这些技术的使用与推广提高了人们的生活质量,同时也促使了诸如数学、信息学等科学领域的进一步发展。伴随越来越多基站建设,网络网点的提供,网络规模与日俱增,这同时促使网络环境变得更加复杂。对于网络流量的模型、特征、可靠性等研究有着越来越高的需求,研究结果将有益于网络工程、网络安全、网络服务等应用领域的各类问题。本文针对网络流量特性基于优化人工神经网络的混合模型方法对网络流量进行了分析。主要工作及贡献可以总结如下:a)对网络流量数据集进行了研究,通过分析数据的混沌特性,验证了网络流量的混沌特性。b)对时间序列预测方法进行了调研,研究了传统时间序列分析、混沌时间序列分析的方法,并重点研究了自回归滑动平均模型。该模型在网络流量预测中存在一定缺陷,需要可靠性、精确性更高的预测模型。c)通过对人工神经网络、小波变换理论和量子遗传算法的研究,提出一种基于量子遗传算法高效的全局搜索能力之上的神经网络优化方法。该方法结合了小波变换对数据的处理,保留了人工神经网络良好的鲁棒性和非线性处理能力,基于优化神经网络对混合网络流量进行预测,该预测模型被命名为量子遗传人工神经网络模型。d)运用量子遗传人工神经网络模型对网络流量数据进行了单步、多步预测并对预测结果进行了评估。通过与自回归滑动平均模型预测结果比较,验证了新模型在自适应性和准确率方面的优越性。综上所述,本文所提出的量子遗传人工神经网络模型能够更加准确的对网络流量进行预测,预测结果在网络安全领域可用于监测网络异常,在网络服务领域可用于提高服务质量,在网络工程领域可用于研究网络行为以及更有效的网络优化方案。
[Abstract]:Modern social network plays a more and more important role in the information society.The use and popularization of these technologies not only improves the quality of life, but also promotes the further development of science fields such as mathematics, informatics and so on.With the construction of more and more base stations and the provision of network network, the network scale is increasing day by day, which makes the network environment more complex.The research on network traffic model, characteristics, reliability and so on has more and more high demand. The research results will benefit all kinds of application fields such as network engineering, network security, network service and so on.In this paper, the traffic characteristics of the network are analyzed based on the hybrid model of optimized artificial neural network.The main work and contributions can be summarized as follows: (a) the network traffic data set is studied. By analyzing the chaotic characteristics of the data, the chaotic characteristics of the network traffic is verified. The time series prediction method is investigated.The methods of traditional time series analysis and chaotic time series analysis are studied, and the autoregressive moving average model is studied.This model has some defects in network traffic prediction, and requires a more reliable and accurate prediction model. Based on the research of artificial neural network, wavelet transform theory and quantum genetic algorithm,A neural network optimization method based on quantum genetic algorithm (QGA) is proposed.This method combines wavelet transform to deal with the data, and retains the good robustness and nonlinear processing ability of artificial neural network, and forecasts the traffic of hybrid network based on optimized neural network.The prediction model is named quantum genetic artificial neural network model.Compared with the prediction results of the autoregressive moving average model, the superiority of the new model in self-adaptability and accuracy is verified.To sum up, the quantum genetic artificial neural network model proposed in this paper can predict network traffic more accurately, and the prediction results can be used to monitor network anomalies in the field of network security.In the field of network service, it can be used to improve the quality of service, and in the field of network engineering, it can be used to study network behavior and more effective network optimization scheme.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP183;TP393.06

【相似文献】