中文摘要：

为了改善小波神经网络（WNN）进行流量预测的性能及避免量子粒子群算法（QPSO）搜索后期的早熟收敛缺陷,提出了一种改进的QPSO。该算法定义粒子群聚拢度,改进收缩—扩张系数使其表示为聚拢度的函数并服从随机分布,以使粒子群具有动态自适应性,避免陷入局部最优,并通过搜索使用WNN待优化参数编码位置向量的粒子群的全局最优位置来实现目标参数的优化,使用本算法优化WNN参数,建立了基于改进的QPSO优化WNN的网络流量预测模型。使用真实网络流量通过两组对比实验对其预测精度进行验证,证明了该方法的可用性。实验结果表明,该方法的预测精度优于WNN和QPSO-WNN方法。

英文摘要：

To improve the performance of wavelet neural network（ WNN） model in forecasting network traffic,as well as to avoid the shortcomings of premature convergence of quantum-behaved particle swarm optimization（ QPSO） algorithm,this paper proposed a novel improved IQPSO method. This method defined particle gathering degree and improved contractionexpansion coefficient,which was subject to stochastic distribution,to be expressed as the function of particle gathering degree to make swarm have self-adaption,avoiding falling into local optimum. And by searching for the global best particle,it optimized wavelet neural network parameters which were encoded in the positions of particles. It trained the wavelet neural network with IQPSO to implement the optimization of WNN parameters and established the network traffic forecasting model based on the wavelet neural network optimized by improved quantum-behaved particle swarm optimization（ IQPSO-WNN）. Forecasting results on real network traffic demonstrate that the prediction accuracy of the proposed method is more accurate than that of traditional wavelet neural network and wavelet neural network optimized by quantum-behaved particle swarm optimization（ QPSO-WNN）.