摘要
文章针对网络流量时间序列的预测问题,提出一种基于变分模态分解(variational mode decomposition, VMD)-分散熵(dispersion entropy, DE)的多尺度组合预测方法。首先对流量样本数据进行混沌特性分析,使用改进的VMD-DE方法对流量数据分解重构,减少周期性流量序列预测的随机性和计算复杂度;然后采用改进鸡群优化算法(chicken swarm optimization, CSO)优化Elman神经网络与最小二乘支持向量机(least squares support vector machine, LSSVM)的多尺度模型,分别对重组后的高频、中频和低频序列进行预测;最后对各预测值求和。通过实际流量数据的仿真和对比实验,证明基于VMD-DE的混沌网络流量组合预测模型具有良好的适应性和预测效果。
In view of the problem of network traffic time series prediction, a kind of multiscale combined prediction method based on variational mode decomposition(VMD) and dispersion entropy(DE) is proposed. Firstly, the network traffic data is analyzed by using chaos theory, and the network traffic data is decomposed and reconstructed by the improved VMD-DE, which can reduce the randomness and computation complexity of the periodic traffic series prediction. Then, Elman neural network and least squares support vector machine(LSSVM) is optimized by the improved chicken swarm optimization(CSO) algorithm to predict high frequence, middle frequence and low frequence sub-sequences. Finally, the ultimate prediction results are established by sub-sequences prediction. The proposed model is tested using the real network traffic datasets and compared with other models. The simulation results show that the VMD-DE based model has better adaptability and prediction performance.
作者
魏臻
陈颖
程磊
WEI Zhen;CHEN Ying;CHENG Lei(School of Computer and Information,Hefei University of Technology,Hefei 230601,China;Engineering Research Center of Safety Critical Industrial Measurement and Control Technology of Ministry of Education,Hefei 230601,China)
出处
《合肥工业大学学报（自然科学版）》
CAS
北大核心
2019年第12期1625-1629,1671,共6页
Journal of Hefei University of Technology：Natural Science
基金
国家国际科技合作专项资助项目(2014DFB10060)
