【摘要】 针对混沌网络流量时间序列预测,提出一种基于自适应变分模态分解AVMD(Adaptive Variational Mode Decomposition)-分散熵DE(Dispersion Entropy)和改进鸟群算法IBSA(Improved Bird Swarm Algorithm)优化核极限学习机KELM(Kernel Extreme Learning Machine)的组合预测模型。利用混沌理论对网络流量样本数据进行分析,采用AVMD-DE方法对网络流量序列分解重构,降低非线性、非平稳时间序列的预测误差及计算规模;采用IBSA-KELM模型分别对重构的子序列进行预测;将预测值进行合成。通过仿真实验分析及与其他预测方法的对比实验,证明AVMD-DE和IBSA-KELM组合预测模型可以显著提高网络流量预测的准确度。更多还原

【Abstract】 Aiming at chaotic network traffic time series prediction,the combined forecasting model was proposed based on adaptive variational mode decomposition( AVMD),dispersion entropy( DE),improved bird swarm algorithm( IBSA) and Optimization of Kernel Extreme Learning Machine( KELM). Firstly,the chaos theory was used to analyze the network traffic sample data,and the AVMD-DE method was used to decompose and reconstruct the network traffic sequence,which reduced the prediction error and calculation scale of nonlinear and non-stationary time series. Then the IBSA-KELM model was used to predict the reconstructed subsequences. Finally,the predicted values were synthesized.Through simulation experiments and comparative experiments with other forecasting methods,it was proved that AVMDDE and IBSA-KELM combined forecasting model can significantly improve the accuracy of network traffic prediction.更多还原