量子遗传算法优化BP神经网络的网络流量预测

Network traffic prediction based on BP neural networks optimized by quantum genetic algorithm

为了提高网络流量的预测精度,提出了一种改进的多种群量子遗传算法优化BP神经网络的网络流量预测模型。在确定了神经网络的结构后,采用多种群量子遗传算法对BP神经网络的初始权值和阈值进行优化。该模型利用K均值聚类算法将种群划分成若干子种群,多个子种群分别进化以保持种群的多样性。子种群间通过移民操作进行信息交互,减小了算法陷入局部最优的概率。同时采用一种自适应的量子旋转门调整策略加快算法的收敛速度。仿真结果表明,相较传统方法,该模型在网络流量预测方面具有收敛速度快、预测精度高的优点。

In order to improve the prediction precision of network traffic, we propose a network traffic prediction model based on optimized BP neural networks with an improved multi-population quantum genetic algorithm. After the neural network structure is fixed, the multi-population quantum genetic al- gorithm is used to optimize the initial weights and thresholds of the BP neural network. The model di- vides a population into several sub-populations by using the K-means clustering algorithm, and main- tains the diversity of the population through respective evolution of several sub-populations. Information interaction among sub-populations through immigration operation decreases the possibility of falling into local optimum. An adaptive quantum rotation gate adjustment strategy is adopted to accelerate the con- vergence rate. Simulation results show that compared with conventional models, the proposed model is of faster convergence rate and higher prediction precision in network traffic prediction.