基于改进黑洞算法优化ESN的网络流量短期预测

网络流量数据序列具有混沌特性.相空间重构后，采用一种改进黑洞算法优化回声状态网络的非线性模型对网络流量进行预测.改进黑洞算法是在现有工作的基础上提出一种新的新解生成机制，可以提高算法的收敛速度和精度;相比于遗传算法、和声搜索算法等其他优化算法，所提出的改进黑洞算法不依赖自身相关参数的准确设定;将其应用于回声状态网络4个重要参数的优化选取，使得预测模型具有较好的预测稳定性.通过Mackey-Glass混沌时间序列和网络流量公共数据集的仿真实验，结果表明所提出的方法具有较好的预测性能.

Network Traffic Short-Term Prediction Based on Echo State Network Optimized by Improved Black Hole Algorithm

The network traffic data series has chaos characteristics. After phase space reconstruction， a nonlinear prediction model based on echo state network (ESN) optimized by improved black hole (BH) was used to predict network traffic. The improved BH algorithm is a new mechanism for new-solution generation based on current works， which can increase the algorithms convergence speed and precision. Compared with other optimization algorithms， such as genetic algorithm (GA)， harmony search (HS) algorithm， etc， the proposed improved BH algorithm is not affected by the accuracy of the setting for some parameters of itself. It is used to optimally select four key parameters of the ESN model， which has better prediction stability. Simulation experiments of Mackey-Glass chaos time series and public network traffic data set show that the proposed method has better prediction ability.