多维加权密集连接网络的流量混沌预测仿真  

Traffic Chaotic Prediction Simulation of Multidimensional Weighted Dense Connected Networks

作　　者：杨方捷[1] YANG Fang-jie(Pass College of Chongqing Technology and Business University,Chongqing 401520,China)

机构地区：[1]重庆工商大学派斯学院

出　　处：《计算机仿真》2019年第9期460-464,共5页Computer Simulation

摘　　要：针对现有多维加权密集连接网络的流量预测方法没有充分考虑网络流量的时变性、混沌性、噪声污染性等特点,使得最终预测结果与真实值之间误差较大,预测耗时较长,提出了基于蚁群优化BP神经网络算法的多维加权密集连接网络流量混沌预测方法,采用小波分析对多维加权密集连接网络信号进行消噪处理,并依据混沌理论对消噪处理后的网络信号进行相空间重构,提取多维加权密集连接网络流量变化规律;在此基础上,建立了一个BP神经网络模型,利用梯度下降法调整BP神经网络参数;将调整后的BP神经网络参数作为蚂蚁的初始位置信息,通过蚁群之间的信息沟通交流和觅食过程中的相互协作获取BP神经网络各层神经元的初始连接权值和相应阈值等最优参数,建立优化预测模型,对多维加权密集连接网络流量进行预测。仿真结果显示,上述研究方法得到的预测结果与真实值拟合度较高,不仅能够有效降低多维加权密集连接网络流量预测误差,而且提高了预测速度。In this article,a method of chaotic prediction for the traffic of multidimensional weighted dense connection network based on ant colony optimization BP neural network algorithm was proposed.First of all,the wavelet analysis was used to reduce the noise in signals of multidimensional weighted dense network.Based on chaos theory,the network signals after the noise reduction was reconstructed by phase space,and then the change rule of multidimensional weighted dense connection network flow.On this basis,a BP neural network model was established.Then,gradient descent method was used to adjust the parameters of BP neural network.Moreover,the adjusted parameters of BP neural network were used as initial position information of ant.Through the information communication between ant colonies and the mutual cooperation in search of food,the optimal parameters such as the initial connection weights and corresponding thresholds of neurons at each layer of BP neural network were obtained.Finally,the optimal prediction model was built to forecast the multidimensional weighted dense connection network flow.Simulation results show that the prediction result of proposed method has high fitting degree with the real value.This method can not only effectively reduce the traffic prediction error of the multidimensional weighted dense connection network,but also improve the prediction speed.

关 键 词：多维加权密集连接网络 流量 混沌 预测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]