随着信息技术的发展, 对等网络P2P信息流量经常出现偏离正常范围的异常情况, 以决策树算法为基础, 对P2P流量检测和流量异常时的检测技术进行了研究。采用改进的C4.5决策树P2P流量检测模型, 通过P2P流量异常检测模型对大量训练数据集的训练, 实现了对对错误的逐步修正, 通过实验室仿真试验, 经过选择网络流量特征后, 基于改进的C4.5决策树的P2P网络流量分类器能实现较好的分类效果, 分类检测率在94.6％～96.7%, 较高的检测率说明采用改进的C4.5决策树算法能有效对P2P流量进行检测, 为今后研究P2P流量异常检测技术提供了参考。

With the development of information technology, P2P network traffic is often abnormal situation from the normal range, based on the decision tree algorithm, the detection technology of P2P traffic detection and abnormal flow were studied. Using the improved C4.5 decision tree P2P traffic detection model, the model of a large number of training data set of training anomaly detection by P2P flow, the realization of the error gradually modified, through laboratory simulation test, by selecting the characteristics of network traffic, P2P network traffic C4.5 improved decision tree classifier can achieve better classification results based on the classification, detection rate of 94.6%～96.7%, higher detection rate of C4.5 improved decision tree algorithm can effectively detect the flow of P2P, for the future research of P2P traffic anomaly detection technology provides a reference.