基于IPSO混沌支持向量机的网络流量预测研究

针对传统混沌支持向量机参数寻优算法的不足,提出了一种改进的粒子群(IPSO)算法。该算法通过延长迭代的开始阶段和最后阶段的搜索时间,实现了算法的全局搜索与局部搜索能力之间的平衡,进而优化模型参数,建立了基于IPSO优化的混沌支持向量机预测模型。应用实例结果表明,该模型对网络流量预测是有效可行的,并具有较高的寻优效率、预测精度和较好的稳态性能。

Study on network traffic prediction of C-SVM based on IPSO

Aiming the limitation of the conventional parameter optimized algorithm in C-SVM, this paper proposed an IPSO algorithm. The algorithm extended the search time in the beginning and final phase of the iterative course, in order to achieve the balance between global and local search capabilities. Then it optimized the parameters of C-SVM, and built the prediction model of C-SVM based on IPSO. The results of application example show that the model is efficient in network traffic prediction with higher optimized efficiency, prediction accuracy and better stability.