Abstract: In order to solve the problem of IP network traffic modeling and predicting, this thesis puts forward an ameliorated model, which is included ordinary differential equation based on the traditional GP model that has some shortages on the precision of traffic behavior and the extrapolated capability of the evolvement. The dynamic description capability of the model can be improved. The complexity of the model can be reformed, and the IP network traffic transformation orderliness can be reflected accurately.

Key words: IP traffic prediction, Genetic programming (GP), Ordinary differential equation, Fitness

摘要： 为了解决IP业务流量的建模与预测自动化问题，该文针对传统GP模型在流量行为分析对预测精度和外推性能方面存在的不足，提出了在原有模型基础上引入常微分方程进行GP（遗传规划）演化，从而提高了模型的动态描述能力，改善了模型的复杂度，更能贴切反映业务流量的变化规律。

关键词: IP业务流量预测, 遗传规划, 常微分方程, 适应度