The network intrusion detection technology based on data mining technology is studied in this paper. On account of low detection accuracy and efficiency of the network intrusion detection technology established by the common BP natural net⁃work which is easy to fall into least value,the particle swarm algorithm is used to optimize the BP natural network model,the dynamical inertia weight coefficient is adopted to define the parameters of BP natural network,and the parameter of BP neural network are integrated with the characteristics of network intrusion rate,and encoded to a particle in order to realize the synchro⁃nous selection of the characteristics of network intrusion rate and parameter of BP neural network. The detection model estab⁃lished with this method and the common BP natural network are trained and tested by using the intrusion flow data in CUP99 KDD database. The results show that the detection model established with this algorithm has advantages of high detection effi⁃ciency and accuracy.%在此对基于数据挖掘技术的网络入侵检测技术进行研究。考虑到常规BP神经网络建立的网络入侵检测技术存在由于BP神经网络容易陷入最小值导致检测效率和准确率低下等问题，使用粒子群算法对BP神经网络模型进行优化，使用动态惯性权重系数以确定BP神经网络的参数，并将网络入侵流量特征与BP神经网络的参数组合并编码成一个粒子以实现网络入侵流量特征与BP神经网络的参数的同步选取。通过使用KDD CUP99数据库的入侵流量数据对使用该方法以及常规BP神经网络建立的检测模型进行训练和测试，结果表明，研究算法建立的检测模型具有更高的检测效率以及检测准确率。
展开▼
