In order to raise the correctness rate of network intrusion detection,we propose a network intrusion detection model (HIWO-SVM)in which the hybrid invasive weed optimisation (HIWO)algorithm is used to optimise the support vector machined (SVM). The model deems the parameter codes of SVM as invasion weeds,and takes the network intrusion detection rate as fitness function of weed seeds,then it finds optimal parameters of SVMthrough simulating the intruding weed seeds progresses of space spread,growth,reproduction and competition,etc.In optimisation process,the crossover operation of genetic algorithm is introduced for enhancing the ability of HIWO to get out of local optimum.At last,we build the network intrusion detection model according to optimal parameter.On Matlab 2012 platform the KDD CUP 99 data set is used for simulation test,results show that the HIWO-SVMcan get satisfactory result of network intrusion detection.%为了提高网络入侵检测的正确率，提出一种混合入侵杂草 HIWO（hybrid invasive weed optimization）算法优化 SVM的网络入侵检测模型（HIWO-SVM）。该模型将 SVM参数编码为入侵杂草，并以网络入侵检测率作为杂草种子适应度函数，然后通过模拟杂草入侵种子的空间扩散、生长、繁殖和竞争等过程找到 SVM的最优参数。在寻优过程中引入遗传算法交叉操作以增强 HIWO 算法跳出局部极值的能力，最后根据最优参数建立网络入侵检测模型。在 Matlab 2012平台采用 KDD CUP 99数据集仿真测试，结果表明 HIWO-SVM可以获得满意的网络入侵检测效果。
展开▼
